Mafia/gag 01cm: 25¢ per day per item RETURNING LIBMRY MATERIALS: Place in book nturn to remove chum fran cit-caution records IMPACTS OF AGGREGATION ERROR ON AN AGRICULTURAL INFORMATION SYSTEM By John Douglas Sutton A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Resource Development 1981 0-: 3 (JD Ll'l/ ABSTRACT IMPACTS OF AGGREGATION ERROR ON AN AGRICULTURAL INFORMATION SYSTEM By John Douglas Sutton Developing countries often rely on agriculture to provide savings, labor, food supplies, and/or foreign exchange for economic development. Yet the sector generally has severe problems that limit its ability to be this engine for growth. To understand sector problems and suggest solutions a variety of agricul- tural information systems have been developed over the past several decades. A frequent criticism of the systems is that they do not incorporate enough detail and are not sufficiently complex. At the same time, others argue that because it is so costly and difficult either to correctly specify model structures or to avoid data measurement errors, models should be small and simple in structure. This dissertation investigates how varying the amount of data detail affects the output of information from one information system -- the Comprehensive Resource Inventory and Evaluation System (CRIBS). The purpose of CRIES is to provide national planners with information useful to analyzing agricultural pro- duction potential. a. f, (‘x 9—0. 1‘" I u-Ivu B . ‘9. I ' V’JU. John Douglas Sutton Three soil inventories utilizing the phases of the soil great group, subgroup, and family, respectively, and three map scales -l:l,000,000; l:250,000; and l:lO0,000, respectively were constructed of the Cibao Valley in the Dominican Republic. The map units and principal soil components of these inventories, crop yield interpretations, and resource use data provided the data for a cost- minimizing linear programming (L.P.) model. Without changing specification of the L.P., the model was run for three policy questions at three different levels of soil aggregation: Detailed, Intermediate Weighted, and Aggregate Weighted. In addition, the model was run at these three levels but with different and "independent" crop yield estimates for each level. These latter three L.P.‘s, the Detailed, Intermediate, and Aggregate differ both in the degree of soil aggrega- tion and yield estimates. Specification error in this simply structured L.P. does not vary with aggregation since only one L.P. is used. Measurement error occurs at each level of aggregation. Aggregation error is measured by comparing the solution values of the two area-weighted L.P.‘s and of the two L.P.‘s with "independent" yield estimates to solutions of the Detailed L.P. The Aggregate soils inventory appears adequate for the user's needs. The map is legible, the number of map units is the same as the number of regions of interest to the user, and the inventory is least costly in time and funds. Data are less reliable than those of the more detailed inventories but sufficiently so given economic data to be associated with the principal components of each map unit. The Intermediate Weighted L.P. may be preferable to the Aggregate Weighted L.P. since its Regional solution values more closely correspond to the Detailed solution, the base model for comparison. Optimal objective function values between the weighted L.P.‘s do not differ significantly. 65' ' John Douglas Sutton Positively biased yield estimates (with respect Detailed yields) were used in the Intermediate and Aggregate L.P.‘s. The Aggregate L.P. may be preferable to the Intermediate L.P. since its Regional solution values correspond best to the Detailed solution and since it is the least costly to construct. However, its yield estimates are subject to greater measurement error. The relative utility of any one information system can only be determined by the user. The issue is one of setting priority between relatively low cost and time requirements but high specification and possibly measurement errors (which occurs with more aggregated, simpler systems) and high cost and time but lower specification and measurement errors (which occurs with more complex, large systems). Although theoretically possible, specification, measurement, and hence aggregation errors in information systems cannot be avoided. It is recommended that system builders be objective, that they faithfully interact with users to determine users' perceptions of the issues and their needs for analysis, and the appropriate type and size of information system to be built. Relatively less research attention should be given to exploring ways to reduce aggregation error and relatively more to identifying the values and needs of user groups. In " its: 5m and n ACKNOWLEDGEMENTS The author wishes to thank Daniel Chappelle for his helpful contributions while serving as guidance committee chairman and as dissertation advisor. In addition, Glenn Johnson, Eugene Whiteside, Milton Steinmueller, and Allan Schmid provided many useful suggestions as members of my guidance and dissertation committees. Sincere appreciation is also extended to John Putman, leader of the CRIES project, for his interest and support throughout the term of this dissertation. Gratitude is also expressed to members of the CRIES project, particularly, Mark Cochran and James Johnson. Weldon Lodwick provided computer programming. Cooperation from the SIEDRA staff in the Dominican Republic, from Trevor Arscott, agronomist, and from Ellis Knox, soil scientist, were also essential to the success of this study. Thanks are also due to Sue Perkins who typed the final copy and many earlier drafts. And finally, I want to acknowledge the encouragement and support that Chris has always provided, and the general acceptance from Laura and Andrea of this use of my time. -11- HI. TABLE OF CONTENTS LIST OF TABLES.. . . . . . ............ . ........ LIST OF FIGURES .......................... CHAPTER I. II. III. IV. INTRODUCTION ......................... Statement of the Problem ................... Study Objectives. . . .................... Research Approach ..................... CRITICISMS OF PREVIOUS AGRICULTURAL STUDIES. ....... Growth and Use of Formal Analytical Systems ......... Two Categories of Criticisms. ................ Poor Identification of Users ............... Models ........................ INFORMATION SYSTEMS. . .................. . A Generalized Information System. . ............. Criteria for Information System Credibility. . . ........ Comprehensive Resource Inventory and Evaluation System (CRIES). . ..... . ............... Conceptualization of Production Potential. ........ Operationalization of These Conceptual Inputs. ...... Interpretation and Analysis. . . . . . . . . . . . . . . . A METHOD TO EVALUATE IMPACTS OF ALTERNATIVE LEVELS OF DATA DETAIL ...... . ....... . . . . . . . . User Group, Problem Identification, and Study Area ........ Three Cibao Valley Soil Inventories . . . . . . ........ Soil Map Comparison Factors . . . .............. Five Linear Programming Models ................ u OOOOOO\O\ l 20 20 21 23 24 31 32 36 37 #1 #9 #9 H}! iv Page Algebraic Structure of the Basic Model .......... 51 Linear Programming Solution Comparisons ......... 5‘} Methods to Estimate L.P. Coefficients ........... 56 V. A COMPARISON OF THE INPUTS AND OUTPUTS OF FIVE INFORMATION SYSTEMS ................... . 67 Comparative Evaluation of Three Soil Inventories. . ....... 67 Preparation Methods ................... 68 Map Analysis .................... . 70 Map Unit Descriptions .................. 81 Soil Inventory Costs .................. 82 Comparative Evaluation of Linear Programming Solutions ..... 85 Detailed, Intermediate Weighted, and Aggregate Weighted Models ................... 85 Detailed, Intermediate, and Aggregate Models ....... 104 VI. CONCLUSIONS .......................... 116 BIBLIOGRAPHY ............................ 130 APPENDIX. ............................. 137 A. Supplementary Data Tables .................. 138 5-6 5—? 5-3 5-9 TABLE #-1 5-1 5-2 5-3 5-14 5-5 5-6 5-10 5-11 5-12 LIST OF TABLES Relationships of Map Units and Their Component PPAs ........ Comparison of Soil Inventory Preparation Methods ......... Characteristics of Cibao Valley Soil Maps .............. Soil Inventory Costs . . . . . . . . . . . ........... Differences in Crop Areas Between Detailed, Intermediate Weighted, and Aggregate Weighted Solutions, Growth and Growth/Shift Runs.. . . . ....... . .......... Area, Production, and Cost Differences Between the Detailed, Intermediate Weighted, and Aggregate Weighted Solutions - Benchmark Run ................... Area Differences Between the Detailed, Intermediate Weighted and Aggregate Weighted Solutions - Benchmark and Growth Runs. . . . ......... . ....... Intermediate Weighted and Aggregate Weighted Solutions as Percentages of Detailed Solutions - Benchmark and Growth Runs ......................... Relation Between Coefficient of Yields Variation and Unsigned Regional Crop Area Differences from Detailed SOlution O O O O O O O O O O O O O O O O O O O 0000000 Rice Production, Transportation Flows, and Production Cost Differences Between the Detailed, Intermediate Weighted, and Aggregate Weighted Solutions -BenchmarkRun.............. ....... .. Rice Transportation Flow Differences Between Detailed, Intermediate Weighted, and Aggregate Weighted SOIUtIOflS - BenChmark Run. 0 o o o o o o o o o oooooooo Differences Between Objective Function Solutions Of All MweISO O O ..... O O O O O O O .......... Correspondences Between Intermediate Weighted, Aggregate Weighted, and Detailed Solutions ......... . . Page 47 71 71 88 93 101 5-13 6~2 6.3 11-1 .i-Z .1-3 .1-5 11-6 1.7 1.3 1-9 1-10 111 Th 6-2 6-3 A-Z A—3 A-4 A-S A-ll vi Page Bias Between Detailed, Intermediate Weighted, and Aggregate Weighted Solutions ............... 103 Rice Production, Transportation Flows, and Production Cost Differences Between Detailed, Intermediate, and Aggregate Solutions - Benchmark Run .............. 106 Rice Transportation Flow Differences Between Detailed, Intermediate, and Aggregate Solutions - Benchmark Run ...... 107 Area, Production, and Cost Differences Between Detailed, Intermediate, and Aggregate Solutions - Benchmark Run ...... 108 Intermediate and Aggregate Solutions as Percentages of Detailed Solutions - Benchmark and Growth Runs ........ lll Correspondences Between Detailed, Intermediate, and Aggregate Solutions ...................... 113 Bias Between Detailed, Intermediate, and Aggregate Solutions ...................... 115 Information Produced by the Detailed, Intermediate Weighted, and Aggregate Weighted Information Systems ...... 119 Revised Soil Inventory Costs. .................. 122 Information Produced by the Detailed, Intermediate, and Aggregate Information Systems. . . . . . . . ......... 123 Measured Areas of Soil Map Units ................ 138 Measured Area Estimates of Intermediate PPAs . ......... 139 Measured Area Estimates of Aggregate PPAs. . . ......... 140 Measured Area Estimates of Detailed PPAs ............ 141 Detailed PPA Measured Areas Normalized to Intermediate PPA Measured Areas ..... . ............... I43 Allocators of Intermediate PPA Areas to Aggregate PPA Normalized Areas . . . ................... 146 Normalized Areas of Aggregate PPAs and Regions ......... 147 Allocators of Intermediate PPAs to Regions ............ 147 Intermediate PPA Cropland Use, Field Estimates . ......... I48 Allocation of Crop Areas to Intermediate PPAs. . . . ....... 149 Crop Annual Costs of Production, by Method of Production ..... 155 L1- 0 . _- \L) .1-23 41-21 .1-22 $23 i-Zh 11.23 in J' (D (7 A-12 A-13 A-14 A-l 5 A-l 6 A-l 7 A—l 8 A- l 9 A-Z O A-Z l A-22 A-Z 3 A—Z 4 A-Z 5 A-26 A-27 A-28 A-29 A-3O vii Distribution of Production Methods, by Intermediate MapUnit . . . . . .................. Sample Description of Cibao Valley PPAs. . ......... Transportation Costs, by Region. . ............. Benchmark Crop Requirements. . . . . ........ I . . Percentage Distribution of Crop Requirements. . . . . . . . Percentage Distribution of Crop Requirements, Assuming Population Shifts. ........ . . . . . . . . . . . Allocators of Detailed PPAs to Intermediate PPAs, Aggregate PPAs, and Regions. . . ............ Allocators of Aggregate PPAs to Regions. . ......... Cartographic Relationships Between Map Units ....... Comparison of Measured and Summed Map Unit Areas . . . . Comparison of First and Second Measurements of Planimetered Areas ..... . . . . . . . ...... Estimated Costs of Three Cibao Valley Soil Inventories . . . . Differences Between Crop Transportation Flows of the Detailed, Intermediate Weighted, and Aggregate Weighted Models - Benchmark Run . . . . ........ Differences Between Area, Production, and Costs of the Detailed, Intermediate Weighted, and Aggregate Weighted Models - Growth Run ......... Differences Between Crop Transportation Flows of the Detailed, Intermediate Weighted, and Aggregate weighted MOdelS "' GrOWth Run 0 o o o o cccccc Differences Between Crop Transportation Flows of the Detailed, Intermediate Weighted, and Aggregate Weighted Models- Growth/Shift Run... . . . ..... Variances and Coefficients of Variation of Detailed Yield Estimates, by Intermediate PPA and Crop ....... Variances and Coefficients of Variation of Detailed Yield Estimates, by Aggregate PPA and Crop ....... Bias Between Detailed and Intermediate Yield Estimates ....... . .............. Page 164 165 166 166 167 167 168 170 170 171 172 I73 176 178 181 I83 185 187 189 1-31 J- .n 13’- 135 A-31 A-32 A-33 A-34 A-35 viii Bias Between Detailed and Aggregate Yield Estimates. . . .......... . ........ Differences Between Crop Transportation Flows of the Detailed, Intermediate, and Aggregate MOdeIS - BenChmal’k Run 0 o o o ooooooooooo Differences Between Area, Production, and Cost of the Detailed, Intermediate, and Aggregate Models - GrOWth Run. 0 o o o o o o o o o o o o o o ooooo Differences Between Crop Transportation Flows of the Detailed, Intermediate, and Aggregate Models - Growth Run. . ..... . ............. Differences Between Crop Transportation Flows of the Detailed, Intermediate, and Aggregate Models " GI‘OWth/Shift Run. 0 o o o o o o o o o o oooooo ix LIST OF FIGURES Page FIGURE 3-1 A Generalized Information System .................... 21 3-2 An Iso-Product "Curve" ......................... 34 4-1 Detailed Soil Map of the Cibao Valley, Scale 1:100,000 ......... In packet 4-2 Cibao Valley Map Unit - PPA Sample Form ................. 43 4-3 Intermediate Soil Map of the Cibao Valley, Scale 1:250,000 ....... In packet 4-4 Aggregate Soil Map of the Cibao Valley, Scale 1:1,000,000. ......... 46 4-5 Soil Map Comparison Factors ....................... 50 5-1 Southeastern Cibao Valley, Scale 1:100,000 ................ 78 5-2 Western Cibao Valley, Scale 1:100,000 ................... 79 CHAPTER I INTRODUCTION Statement of the Problem In many developing countries, one-half or more of the population lives in rural areas and gains its livelihood from agriculture. The agricultural sector is frequently burdened with a major responsibility to promote national economic development. Ways of promoting such development include increasing food supplies for domestic consumption, releasing labor for non-farm employment, enlarging the size of the domestic market for manufactured goods, increasing the supply of domestic savings and exchange (Myint, 1975). At the same time, these countries commonly face human problems such as low per capita food consumption, poor nutrition, high unemployment and low incomes; and natural resource problems such as subutilization of potentially highly productive soils, overutilization of relatively low productive soils, declining soil fertility, erosion, eutrophication of water supplies, high rates of slash and burn agriculture and deforestation, and others, all in the face of severely limited financial and skilled human resources needed to resolve them. Governments of developing countries recognize that they must understand the nature and extent of sector problems if they are going to be able to obtain the expected contributions to development. Ignorance about agricultural production potential can be overcome only by systematically gathering data relevant to clearly specified problems, analyzing those data, and providing information useful to those responsible for formulating agricultural policies. 1 2 National and international agencies have financed innumerable efforts to produce data and useful information on sector problems. These efforts, some of which may be labeled agricultural information systems, have had the potential to help resolve such problems by improving the decision making process of national planning bodies. Unfortunately, literature of the natural resources disciplines suggests that the proliferation of data and analyses has not produced this desired result. Typical criticisms of marginally useful information systems--of well- intentioned but irrelevant studies -- include: users are not clearly specified; problems are not clearly described; - relationships modeled are too simple to be useful in such a heterogeneous sector; or, models are too complex for users to understand chains of causality; - models are too normative; - the simplifying assumptions of theory and the resultant implications are never explained to the users; - models have too much or too little detail. Errors of specification, measurement, and aggregation seriously limit usefulness. Study Objectives This dissertation is an inquiry into the information production function of one agricultural information system--generally represented by the Comprehensive Resource Inventory and Evaluation System (CRIES)l project in the Dominican Republic (D.R.). CRIES provides subject matter research to assist participating countries develop an agricultural resource inventory and analytical system that lCRIES is funded by the Office of International Cooperation and Develop- ment in the U.S. Department of Agriculture through a Participating Service Agreement between the U.S. Agency for International Development and the USDA. Participation of Michigan State University is covered under a Research Agreement between the Economics and Statistics Service of the USDA and MSU. 3 will help national planners explore the extent, quality, and use options of natural resources. The CRIES D.R. system is based on the CRIES' conceptualization of agricultural production potential: maximum crop production given limits set by the distribution and quality of natural and human resources, production tech- niques, and societal institutions. Within this broad subject matter area, data are made available for resource planners and users with concerns about solving specific resource problems. The general objective of this dissertation is to investigate how varying the amount of spatial data detail affects system output: information useful to national planners and policymakers about how the land resources of the Cibao Valley might be most efficiently used to increase crop production. The general procedure is to a) construct three soil inventories of the Cibao Valley and specify one linear programming (L.P.) model aggregated to three different levels, and b) compare the information produced by each soil inventory-L.P. information system. This dissertation is only a partial investigation into the information produc- tion function. For example, the question of whether any system's analyses become useful to policymakers is largely a function of the latter's relationship to the subject matter area and the analyses undertaken. Hence, a more complete investigation of usefulness would involve user interviews as one way of establish- ing both monetary and non-monetary values for inputs and outputs. Intended users of the results of this study are information system builders, such as CRIES personnel. These persons must interact with potential system users to establish the relevant data detail for soils, land use, production methods, etc., that are necessary for useful subject matter research. Research App roach The specific procedure to be followed includes these major steps: a Review the literature for criticisms of national/regional agricultural planning efforts. Establish the need for an information system approach to the subject matter area of agricultural production potential. Define data, infor- mation, and information system and criteria for developing systems that are credible to intended users. Build five information systems for the Cibao Valley. a) Describe a problem within the subject matter area to which the information systems will be applied: making the least costly use of Cibao Valley land resources to meet crop production require- ments. b) Prepare and describe the three soil inventories of the Valley. c) Build five interregional L.P. models. Three of the L.P.‘s will differ only in the degree of soil aggregation. Otherwise, they have identical data inputs. Two will differ both in degree of soil aggregation and in crop yield estimates. Compare inputs and outputs of the five information systems: a) Develop performance indicators of changes in the data inputs (the soil maps and the L.P.'s) and information outputs (from the soil maps and the L.P.'s). b) Examine the information content of each. soil map and the impact of land aggregation on the solution values of each L.P. Discuss the differences in policy prescriptions of each system for resource use in the Cibao Valley. Formulate conclusions regarding the importance of errors of specifica- tion, measurement, and aggregation to developing information systems for national policymakers concerned with agricultural production potential. TE ‘1‘.) (1 5 The organization of this dissertation is as follows: Chapter II is a literature review of criticisms of earlier agricultural information systems including a section on aggregation error. In Chapter III, a generalized information system is presented and the differences between data and information are discussed. In Chapter IV, a method is developed to evaluate information content from soil maps and linear programming models, the two analytical tools under study. The information output of the Detailed, Intermediate, and Aggregate soil maps and five L.P. models are presented and evaluated in Chapter V using the performance indicators of the previous chapter. In Chapter VI, conclusions relative to the effects of specification, measurement, and aggregation errors on the production of useful information are made and recommendations for further research are suggested. Won “In“ CHAPTER II CRITICISMS OF PREVIOUS AGRICULTURAL STUDIES Growth and Use of Formal Analytical Systems During the past two decades, support for policy research and analysis has grown impressively. In a recent five-year period, 1969-1974, expenditures on nondefense program evaluations alone rose to $146 million, a 500 percent increase (Rein and White, 1977). In spite of the growth in formal planning-policy systems, documentation of direct use is rarely found. An exception is found in the National Water Commission which sponsored a linear programming (L.P.) analysis of optional water-resource policies to the year 2000. Based upon model projections, the Commission concluded that "no additional irrigated agriculture was needed to meet domestic and export needs, and that Federal subsidies for development or additional irrigation in the United States should be discontinued (Holcomb Research Institute, 1976). During the past decade, most major national agricultural and related resource commissions have used large-scale L.P. models to assess the impacts of different policy alternatives at both national and regional levels; to evaluate the productivity potentials of the agricultural sector and the nation's land and water resources; to indicate generally 'the what and where' as variables at the national level and in international trade have impacts which thread back to individual producing and resource areas; and to reflect interdependencies among regions, technologies, resources, commodities and policies (Heady, 1975, p.4). L.P., the mathematical tool under study in this dissertation, has the ability to recognize interregional relationships, which are important because producers and resources are spatially distributed. Interregional L.P. serves well as a tool to 6 7 generate information at both national and regional levels. National-level infor- mation is needed so that supplies can be equated with demand and interdepen- dence can be established among regions. Regional detail is needed so that the flexibilities of, or restraints on, production and resource use can be measured, and impacts can be expressed at the regional level. Accompanying the growth in support for analytical systems has been a chronic sense of frustration about their usefulness both on the part of those who have conducted the work and those who have sponsored it. For example, despite the large number of soil surveys--one kind of information-generating effort--undertaken in this period, it is felt by many soil scientists that "these have not influenced agricultural development...or...increased productivity as ought to be expected" (Stobbs, 1970, p. 41). In spite of the feelings that policy research and evaluations have been largely ineffective, there has been very little systematic investigation into the extent to which research enters the policy process. One of the few studies of this kind found that "rarely is policy formulation determined by a concrete point-by- point reliance on empirically grounded data" (Caplan, 1975, p. 47). Attempting to determine why some studies were used while others were not used for public policy, Caplan concluded that lack of contact between researchers and policy- makers and lack of trust were the major factors for variations in utilization. Mathematical models, which are often a major part of informational efforts, have also been repeatedly criticized. In 1966, a participant at a conference devoted to agricultural sector analysis commented that, "there is nothing like the array of formal planning models presented at this conference to give one particularly interested in the agricultural sector even more neuroses than he might already have" (Heady, 1966, p. 381). Eleven years later, the U.S. General Accounting Office reported that while there had been a proliferation of quantita- tive models in response to the growth in concern about the workings of an 8 increasingly complex and interdependent agricultural sector, the number of models had raised doubts about which models would be useful. Two Categories of Criticisms For discussion purposes, criticisms of previous agricultural studies may be grouped roughly into two broad categories. The first deals with questions of who is the user, what are his needs, and how congruent are the analytical tools to these needs. The second deals with questions of the structure and detail of mathematical models, particularly L.P. Obviously, the two categories are closely related. Poor Identification of Users Eisgruber points out that the major limitation to the usefulness of many information efforts is poor definition of user groups. In most agricultural information systems, "the modeler rarely interacts with the policymaker but instead with a technical staff" (Eisgruber, 1967, p. 1548). It should not be assumed that the users of information-generating systems are the policymakers themselves since technically competent policy analysts may act as important intermediaries between the policymaker and the system builder. Without con- siderable interaction between the analyst and the system builder and also between the policymaker and his analytical staff, differences in values and objectives cannot be sufficiently recognized and considered.1 Identifying the users as "policymakers" does not resolve the problem of definition since this term can have a variety of meanings. Policymakers vary in professional background: they may have technical skills, or they may have policy lThis criticism is not restricted to the public sector. A recent article in Fortune Magazine describes part of the problem and some means to deal with it. The article contends that data processing managers invent solutions without identifying the problems. The last thing considered would be impacts on the user. Currently, the successful data processing shops have good relations with top management, not because top management understands computers, but because the data processing manager understands the business. Data processing managers are moving higher up in the management structure (Bylinsky, 1978). 9 skills. They are sensitive to different issues and to different inputs. In addition, because policy issues are typically highly complex and uncertain in most agencies, there may be a large number of influential decision-makers having conflicting views. Needs and objectives of users are often not clearly identified by system builders. In the context of developing nations, Vernon discusses the extent to which large-scale model-building supports the needs of national planners. Among the international agencies, three basic criteria seem to have evolved for a country to be said to have a national plan: first, the plan must be comprehensive, i.e., it must explicitly state a set of output and income targets and relate these targets to other performance variables both within and among sectors of the economy. Second, it must be consistent, i.e., the composition of goods produced must be consistent with the composition of goods demanded. Third, it must make optimal use of a country's scarce resources (Vernon, 1966). Vernon questions the extent to which information-generating efforts that are comprehensive, consistent, and optimal meet the needs of policymakers. Aside from these perhaps secondary characteristics of national plans is the overriding need for a plan to justify a claim for foreign aid. Top echelon decision makers are acutely aware of the scarcity of national resources for development. They may be quite willing to sacrifice consistency, for example, in order to meet this primary purpose. The national plan also can serve as a document to generate local hopes, describe the policymaker's interest in local problems, and strengthen the incumbent's hold on his office. Incompatibility between national plans and national planners often rests on different assumptions regarding the nature of information itself. The planning process in western developed economies depends for its validity upon certain critical capacities. These include the capacity to gather information and to create common planning assumptions among sectoral planning agents. 10 Information should not flow in a single direction but be part of a dynamic process. This is not the case in Latin American nations where communication channels are more nearly monologue than dialogue, more vertical than horizontal. This situation stems from an acute shortage of administrative and technical skills in the public sector; from fears of proposals that recommend significant change; from a society that has long been based on personal relations that run from a "patron" to those protected; and from a deep-seated suspicion between the private and public sectors. Another problem of developing models for policymakers is the professional requirement to satisfy not only the needs of the policymaker but also the demands of the professional community. Policymakers may change with elections, but the modeler's peer group remains. This dual audience can create problems. For example, a goal of merely collecting statistically reliable data useful at some uncertain later date for policy analyses may be preferred by some disciplinary specialists. This long-term approach may not satisfy the most urgent needs of the policymaker who must also consider changing political forces in his decision making (Cummings, 1977). MEI—85 The second general category of criticisms deals with confusion over the appropriate type of model to be constructed, appropriate use of values, and appropriate levels of data detail. Where desirable, discussion will be related to linear programming models. Too often, individuals have approached problem identification and solution from only the narrow perspective of their own disciplines. For example, having identified a problem of chronic low incomes in a rural area, an agronomist may recommend introduction of a new high-yielding crop. Unless other aspects of the problem such as local and distant markets for the crop and peasants' attitudes 11 toward the uncertainty involved in changing their habits are also studied, the proposal may be ill-conceived. At the heart of this issue is confusion over the type of research that ought to be undertaken for a specific purpose. Three broad types may be cited: disciplinary, subject matter, and problem-solving (Rossmiller, 1978). The purpose of disciplinary research is to develop or refine disciplinary theoretical knowledge and/or methodology. Although disciplinary specialists may regard this research as problem-solving, it may not be so due to inadequate treatment of facets of the problem which lie outside of the discipline. Subject matter research develops information about an area of concern that contains many specific interrelated problems. Being multi-disciplinary, it brings together relevant bodies of know- ledge. Additional data, analytical tools, and information must be provided to solve specific problems. The agricultural information system considered in this dissertation reflects subject matter research. It can make a contribution to knowledge about agricultural production potential but cannot solve specific problems. Credibility gaps between economists and policymaking users are created when the latter wanted problem-solving efforts but received subject matter efforts. The third type of research is problem-solving. Information that is derived from this research is also multidisciplinary, but it is problem-specific as well. Output from a problem-solving analysis is a prescription for action. A further criticism is that models are normative, i.e., evaluative and value- laden, and therefore of limited operational value in analyzing development problems and prescribing solutions (Singh, 1976). As noted above, policymakers may not have the same values as model-builders. Thus, a credibility gap could arise if the former did not fully accept the modeler's single goal of maximizing efficiency. Use of the normative is not undesireable & s3 but not recognizing what constitutes use of the normative may be confusing. For example, resource 12 economists who assume that technology, institutions, and people are fixed are not avoiding the normative. Their acceptance of the existing distribution of oppor- tunity sets and of power between persons as a valid starting point for trade; their definitions of which outputs are products and which are waste; their use of market valued inputs and outputs; and their assumption that the rational actor is either a profit or utility maximizer; are all normative decisions. In information-generating efforts that use maximizing techniques such as L.P., the frequent failure to recognize the four normative preconditions for maximization described below damages credibility (Johnson, 1971). Furthermore, these preconditions must be established before credible and acceptable decisions can be made using L.P. alone. The inability to satisfy them is a major reason L.P. can provide only limited information that may be useful to making decisions. The first precondition is the existence of a common denominator which reduces the "goods" being sought and the "bads" being avoided to a single dimension. Monetary values may or may not be usable for reducing the health objective of one policy, environmental objectives of another, and income objec- tives of another policy to comparable levels. Second, such a common denominator should be interpersonally valid since some policies may confer "goods" on some people and "bads" on others. Third, the maximum to be found must exist. When programs and policies involve technological, institutional, and social changes, there is no automatic ordering of programs and policies that will meet the mathematical conditions necessary for locating an optimum. Until such an order is established or found to be unnecessary, it is impossible to maximize the difference between good and bad, even if a common denominator is available. Fourth, a decision-rule is needed to prescribe the right action to take. In a static economy when the first preconditions are met, the rule is simply to maximize the difference between good and bad. In real world situations, however, when positive and normative knowledge is imperfect, when actors may not always be man 500‘. the A the r ”C 6er W111 be 13 maximizers, when optimizing order of programs and policies is unknown, and when an interpersonally valid common denominator may not have been found, the definition of an acceptable rule by anyone other than the user of the information system may be impossible. Aggregation Error.l Much concern is focused upon the significance of different amounts of data detail in an information system. This concern will be discussed in terms of aggregation error and its components specification error and measurement error. The aggregation problem surfaces whenever the economist, who wishes to study aggregate supply or demand relationships, must choose between macro or microanalytical techniques. Macro techniques would provide for direct statistical estimation of the aggregate functions without requiring analysis of the individual decisionmaking units. Micro techniques, by more manageably focusing on the individual unit, would much better represent the social, economic, and cultural forces that condition behavior. On the other hand, the huge number of micro models needed to represent the behavior of all firms in the market precludes full and complete application of microanalytic methods. One approach for proceeding beyond micro to macro analysis involves identifying an hypothetical farm whose supply function is "representative" for farms in a given region. Aggregation of farm supply functions based upon this Marshallian concept of typical farms involves a) sorting farms in a region into homogeneous categories and deriving supply functions for "representative" farms in each category; b) aggregating these functions to provide a regional estimate of supply response; and c) adjusting results on the basis of experience to make predictions (Barker, 1965, pp. 705-6). 1The term "aggregation bias" is frequently used. "Bias" denotes a systema- tic error rather than a random error. Since errors in aggregation may be biased or unbiased, depending on the method of aggregation, the term "aggregation error" will be used instead of "aggregation bias". 14 A conventional presentation of the aggregation problem is as follows: Assume farm A has 100 units of capital and 50 units of labor; B has 100 units and 150 units, respectively. It requires 10 units of labor and capital to produce one unit of product Y. Programmed separately, A and B will produce a total of 15 units of Y. However, if their resources are averaged, the representative farm would have 100 units of capital and 100 units of labor. Programming this farm and multiplying by two (to represent the two farms in the aggregated farm) allows production of 20 units of Y. The supply function for this representative farm will thus be shifted to the right. When the representative farm aggregate is represented by a single linear programming (L.P.) model, the question to be asked is, to what extent must the component farms be alike in order for the single L.P. to represent their supply functions without distortion. Day has shown that aggregation error can be avoided only if farms are classified into groups which are defined according to rigid theoretical requirements for homogeneity (Day, 1963): a) technological homogeneity: each farm has identical production possibi- lities, resources and constraints, levels of technology, and managerial ability; b) pecunious proportionality: individual farmers hold net revenue expec- tations that are proportional to average expectations; and c) institutional proportionality: the constraint vector of the model for each farm must be proportional to the aggregate constraint vector (Buckwell, 1972, pp. 122-123). These may be algebraically stated as: 1) B1 = B2 = Bn = B where Bi is the input-output matrix for each farm 1, n. 2) Zi = aiZ where Z1 is the net return expectation of the ith farm. 15 3) C1 = biC where th Ci is the resource constraint vector of the i farm. "ai" and "bi" are scalars z 0. A consequence of these requirements is that each farm should be pecuniously and institutionally proportional to the arithmetic mean farm for its group. Others concluded that it is "not possible to find empirical cases to fit Day's conditions" (Paris and Rausser, 1973, p. 660). Attempts to develop less demanding conditions include those which suggest that farms be classified on the basis of the most limiting resource for each product (Sheehy and McAlexander, 1965). Aggregation error would be less than under the conventional grouping used as an example above. Miller would include in one group all those farms that have a common range of resource ratios; the result of such a grouping is individual farm solution vectors all containing the same activities although not necessarily at the same levels (Miller, 1973). Unfortunately, as Buckwell pointed out, the application of both Sheehy and Miller's conditions requires full previous knowledge of the technology matrices of the individual farms. Hence, not even these methods are easy to apply. In contrast to defining the "representative farm" and then aggregating to regional totals is the "aggregate resource" situation in which the L.P. considers a region as an entity with aggregate relations estimated for it. The L.P. uses the geographic map unit directly as the unit of optimization or relevant resource restraint. In concept, the restraining resources of individual farms are summed to a regional (map unit) total. Specification of the areas is simply a special problem in aggregation over inputs. Some of the assumptions underlying this aggregation should be noted. First, a single value is used for technical coefficients in the resource area even though these may be very different within the components of that area. Second, a single farm model is assumed although farmers' 16 characteristics across the resource area may vary a great deal. Third, except for land, the supplies of other production inputs are assumed to be perfectly elastic. The "aggregate resource" approach is under study in this dissertation. The authors cited on the previous two pages dealt with errors in L.P. "supply functions" resulting from different methods of aggregating the representative farm. This dissertation explicitly considers the question of the effect of aggregating soils of different qualities and to a lesser extent the resources attached to the soils. gecification Error. Some economists criticize information systems for being too aggregative & s3. For example, Singh believes that for agricultural models to be useful in so heterogeneous a sector, they must incorporate a great deal of detail. "Until farm-level behavior is first understood and modeled, normative national models for the spatial allocation of agricultural resources provide few operational clues to the development of the sector in LDCs" (ibid., p. 771). Much more detail would be needed to adequately represent the institutional structure and its constraints. This view is echoed by Falcon who states that unless mathematical programming models are founded on detailed farm-level models of specific agricultural regions, they are "likely to give misleading implications for policy in developing countries" and should be used with "extreme caution" (Heady, 1966, p. 382). The validity of this "disaggregated-first" approach rests on the ability a) to specify perfectly conditions actually facing the farm firm for a given length of run; b) to specify perfectly interrelationships among farms; and c) to meet Day's conditions for aggregation without error. It was concluded above that Day's conditions (or those even less demanding) could not be met. It is also argued that it is impossible to avoid specification error since it is impossible to construct complete structural or behavioral systems for representative farms. Aggregation of misspecified relationships increases the amount of misspecification and hence l7 aggregation error. Further, as Fromm notes, it is highly doubtful that economic theory is so encompassing that it is possible to faithfully relate microtheory to macrophenomena. Even when microtheory provides behavioral propositions, normally they refer to static or dynamic equilibrium states and not to the disequilibrium situations of the real world. Functional specifications are always underlain by the strict, simplifying assumptions of the theory (Fromm, 1973). Finally, the "disaggregated-first" approach seems to be based on the questionable assumption that national policy decisions are made on the same basis as are firm-level decisions. This is not the case inasmuch as the national decision-maker may be more concerned with the values of several interest groups as well as with his own and also with the long run. Measurement Error. Measurement errors both for micro and for macro variables include errors in technical coefficients, the resource restrictions, input cost, etc. Data may be drawn from a number of sources. Thus, some data may really be modal, or average, or drawn from a very small and unrepresentative sample. Yet all inputs represent the total number of farms or resource areas. If sampling techniques are used, errors arise if the distribution of the model's parameters over all farms was not known but estimated. Without an historical record, it may be difficult to establish the relevant range for any variable. If each of several variables is felt to have wide ranges, the utility of model solutions may be seriously questioned. If the distribution of a variable's values is expected to have a large variance and/or if the distribution is expected to be badly skewed, the use of some representative value has less meaning. Citing the nearly full utilization of some forms of capital, human, and natural resources in developing countries, Vernon feels such skewed distributions may be characteristic of certain sectors. Accordingly, and in contrast to Singh and Falcon, he concludes that the proper amount of data detail may be quite low, for "the best of models cannot be expected to yield a great deal more than the simpler varieties" (Vernon, p. 65). 18 Since measurement error will normally occur before aggregation, its magni- tude influences aggregation error. It is not obvious that aggregation of measure- ment errors will produce greater error, however. If measurement errors at the disaggregated level were not biased, their aggregation might result in a net reduction of error as overestimates cancelled out underestimates. Two additional issues surface in the discussion of small, simple vs. large, complex models: cost and time required to build, run, and interpret them. Because the costs of obtaining sufficient data and estimating microrelationships are very high, some favor more aggregated or simpler models. Because the time required to build complex models is greater than for simpler models, the latter may be favored also. There are at least two aspects. First, for political reasons, timeliness of analytical results may be highly important to the user. Second, and related to cost, large and complex models may require greater amounts of scarce supplies of skilled technicians and analysts than smaller models. In summary, over the past several decades the number of quantitative models constructed to address agricultural issues has increased dramatically. Lack of documentation on their actual use is consistent with the variety and severity of criticisms levied upon them. A repeated issue is that of the proper amount of data detail to be incorporated into model structure. This is the issue of large, complex vs. small, simple models. Inability to perfectly specify functional relationships and to avoid measurement errors and hence aggregation errors either in complex or aggregated models is apparent. The question of the correct degree of aggregation may be one of setting priorities between control of specification, measurement, and aggregation error on the one hand and the cost and time required to build alternative systems on the other hand. In this dissertation, one L.P. model structure (see p. 51-2) is specified. It is quite simple and aggregated. Without changing the L.P. specification, the model is run at three increasingly aggregated levels of detail. Hence, trade-offs of different amounts of measure 19 ment errors, a given high level of specification error, and costs of the three different levels of detail will become apparent in the following chapters. CHAPTER III INFORMATION SY STEM S In order to begin to resolve criticisms of information-generating efforts such as those discussed in the previous chapter, it is important to understand how useful information may be produced. This chapter first presents a generalized information system and then the specific case of the CRIES information system developed for the Cibao Valley. A Generalized Information System Figure 3-1 depicts the major components of a generalized information system. Given the infinite variety of the real-world, the researcher must simplify reality in order to be able to deal with it. If data produced are to be internally consistent and correspond to reality in important ways, classification must itself be based on concepts that are meaningfully related to one another and to the real world. Thus, to produce accurate data, conceptualization of some aspect of reality must first occur. These steps of conceptualizing an area of concern, determining which data to gather, and deciding which methods to use, gather, store, and retrieve them, comprise the data subsystem of the information system in Figure 3-1. An information system includes not only data production but also analysis and interpretation in the context of some problem or set of problems. Decision- makers rarely use raw data. Rather, the data gathered may be interpreted or 20 21 revised, or new data may be gathered and interpreted.1 Analytical tools that provide the informational inputs needed to solve the sets of decision rules established by the decision-maker are critical to the successful operation of the system. F... Decision-makers 1—9 Interpretation/ Analysis ——.> Data Measurement/ Storage ,____, Classification/Operationalization ——+ Theoretical Concepts ,___, Real World Problems Figure 3-1.-- A Generalized Information System Source: Adapted from Bonnen, 1975, p. 758. Criteria For Information System Credibility The fundamental purpose of our information system is to increase the user's knowledge about the set of problems in his area of concern. In order that this be possible, there should be interaction between the researcher and system users. Such communication may have several consequences. First, it allows a better understanding on the part of users of the conceptual, operational, and analytical structure of the system. Second, the system builder can better determine whether the system represents reality as the users perceive it. Third, the builder 1Data are not equivalent to information although the terms are often used interchangeably. Data become information by being associated with and inter- preted with respect to an issue which is of concern to a decision-maker. It is in practice difficult to distinguish between data and information within the context of an information system since unprocessed, unformatted, raw data are rarely used by either the final user or his staff. Instead, at a minimum, data are presented in some particular format that carries with it some degree of interpretation. 22 can become better aware of the roles played by his values and by those of his client(s). Production of useful information requires that both positive and value-laden normative knowledge be utilized. Positive knowledge refers to concepts that describe conditions, situations, and things without assigning degrees of "good" or "bad." Positive knowledge is nonevaluative. Normative, on the other hand, means evaluative. Hence, normative knowledge includes standards for judging condi- tions, situations, and things in terms of (not as) good and bad. Inasmuch as the use of the normative is inevitable, system builders must be objective if their work is to be accepted by the user. Normative knowledge is considered objectively true if it meets four tests. They are a) logical consistency, b) correspondence with reality, c) clarity, and d) workability (Johnson, 1971). The important question in determining logical consistency (or validation) is whether the underlying assumptions and theories and the interactions among the parts of the system are compatible with the general set of problems to be analyzed. If, as sometimes happens, the system builder omits this test, or any of the other three tests, and tries to function as the policy analyst without the latter's much greater breadth of knowledge, the usefulness of the system is apt to be seriously jeopardized. Although logical consistency implies some sort of a comparison between model and reality, there may be little quantitative informa- tion about the real system that can be used as a basis for comparison. In such a case, the subjective judgments of the system builder and user are heavily relied upon. Correspondence with reali_ty, also called model verification, is closely related to logical consistency. It involves testing concepts with respect to their ability to reflect real world phenomena accurately. Since the researcher's perspectives and experiences are necessarily limited while reality is infinite, it is 23 at this point that one is faced with deciding when singular statements (summaries of observations) can lead to the universal ones (generalizations) of economic theory. While final proof is never possible, the test of verification can be met if empirical theories are intersubjectively testable (Shapiro, 1973). The third test is glam. The system's conceptualization of reality and the underlying theory and assumptions must be explainable and understandable to the user. Finally, the decision-maker tests the system for workability. This is a test of the system's prescriptive ability and is undertaken by the user. The system must provide information usable to the decision-maker in the process of solving the variety of problems found in the area of concern. Comprehensive Resource Inventory and Evaluation System (CRIES) The real-world set of problems (refer to Figure 3-1) about which the CRIES information system attempts to provide information relate to a country's agricul- tural production potential. The objectives of CRIES are to 1. develop a natural resources classification and analytical framework adaptable to many countries and capable of accumulating and trans- ferring reliable information among countries. 2. help national planning agencies strengthen their capacity to determine the extent, quality, and use of their resources, and to evaluate partial impacts of alternative uses. I Intended users of the information are public officials with agricultural policymaking responsibilities at the national level and their immediate technical staff. Development of this resource information system is not intended to provide solutions to specific problems falling within the subject matter area of agricul- tural production potential, nor is it intended to provide comprehensive recommen- dations about needed sources of action. It is designed to provide a specific, 24 limited amount of information that together with other sources of information will permit setting policies to attack problems in effective ways. Conceptualization of Production Potential. Conceptualization of production potential is based on host country docu- ments that contain statements of agricultural problems and public policies and interpretation of these problems and policies by U.S. and host country specialists in agricultural production economics, soil science, botany, and agronomy. A development plan for the northwest area of the Dominican Republic discusses agricultural production potential of the Cibao Valley: ....a disequilibrium exists that will be accentuated if adequate mea- sures are not enacted immediately. This disequilibrium is determined by....rapid population growth; migration, principally to Santo Domingo and Santiago; low real incomes of the rural people; scarce educational and health opportunities; the lack of remunerative employment and of adequate communications; inadequate food supplies; and poor nutri- tion. From the viewpoint of production, only limited supply is observ- able, as well as underemployed labor, land held idle for lack of water, credit, and other factors. Among others, the following actions are needed: identify possibilities for increasing yields and production of currently cultivated crops; identify new crop possibilities; identify new activities such as livestock, forestry, mining, and tourism; adopt natural resources conservation practices (Organizacion de Estados Americanos, 1977, p. xxx). Another Cibao area study found that soil resources considered suitable for food production are quite limited in the Dominican Republic because of the naturally steeply sloping and broken relief of much of the island. Due to increasing demands of a growing population, and to national needs such as more employment, higher incomes, and more foreign exchange, "it is necessary to use natural resources to their natural productivity" (FAO, 1974, p. I). In subject matter research such as this dissertation, each specialist is expected to contribute with scientific objectivity to the conceptualization of the area of concern. 25 Contributions from Production Economics. The production economist is concerned with the efficient allocation of resources to produce desired outputs. His theory may be static in that there are a given number of exact and unchanging relationships among a number of economic variables. The static system is timeless; it does not deal with the path to equilibrium. Assumptions of static production economics fall into three broad sets. The first set consists of assumptions that fix: a. production functions by assuming a constant state of technology and input-output relationships; b. utility functions by assuming unchanging tastes, habits, and customs, the ownership pattern for resources (and hence the distribution of private real incomes and power relationships; there is no assumption of perfect competition), and population levels; c. the institutional mix of the economy by assuming unchanging public- private sector relationships and the existence of markets where goods and services are produced and exchanged without coercion. The second set of assumptions eliminates random elements by assuming perfect knowledge and rational behavior. The third set concerns the behavioral forces that drive the system. Consumers are assumed to act to maximize their utility, and producers, their profits. Not only does such behavior work to solve individual producer/consumer problems, but it is assumed to promote resolution of national/regional problems. A consequence of these assumptions is that purely static analyses may tend to create credibility gaps between the analyst and decision-maker. While the assumptions make possible a theory that is manageable conceptually, the decision- maker may want to go beyond the market to non-market changes in technology, institutions, and people and into situations where knowledge is imperfect. Because of his concern for societal welfare, he may go beyond market pricing to 26 value the net changes in utility. Most real-world problems and solutions involve going beyond free exchange to include the use of power and coercion to redistribute the ownership of rights and property. Static theory lacks the ability to resolve dynamic problems in which changes occur that confer benefits on some and damages upon others. All adjustments are "better" because of the perfect knowledge assumption: ex a_1_n_t_e_ = ex M. Nobody will enter into any transaction unless that person is made better off. In spite of these limitations, the theory is highly useful in that, by simplifying the complexities of reality, it becomes possible to gain at least a partial understanding of how resources are allocated. It is critical to credibility, however, that the decision-maker understand the limitations of the analysis. The decision-maker with his concern for increasing the welfare of society, needs to be aware of both an analysis that accepts the s_ta_tu_s M distribution of opportunity sets and power and other types of analyses that provide information essential to his other decision rules. Two concepts of production economics have been drawn upon in particular, to help conceptualize the subject matter area, i.e., the agricultural production potential of the Cibao Valley. These two concepts are comparative advantage and marginal productivity analysis. With respect to the former, it has been demon- strated that while the possession of a substantial quantity of natural resources of a relatively high quality is conducive to a region's economic development, it is not obvious that any particular type or amount is absolutely essential to such growth. What is required is access to natural resources via trade with other regions as well as the availability of substantial quantities of capital and labor (Herfindahl, 1969). This principle of comparative advantage underlies the production economist's conceptualization of agricultural production potential. It was first stated by the classical economist David Ricardo in 1817. Refining Adam Smith's conception of absolute advantage, Ricardo showed, in the context of international trade, that 27 trade might take place to the advantage of both trading partners even if one had an absolute advantage in both of the commodities exchanged. Trade, and hence resource allocation, would thus occur because of comparative advantage, a situation depending only on productivity ratios of its resources (Lekachman, 1959). In this study, the trading partners are regions within the Dominican Republic. Marginal productivity analysis is based on the concept that, given a set of fixed inputs, as the amount of a variable resource used in the production of some good is increased in successive equal increments, output will at first increase at an increasing rate, then at a decreasing rate, reach a maximum, and then decrease absolutely. This is the familiar law of variable proportions or law of diminishing returns. Given the assumptions of a static economy, producers will produce up to the point where marginal value product equals marginal factor cost. At that point profits are maximized. This optimal or high profit point thus depends upon the relationship of the prices of the product and the resource inputs as well as upon the physical production function. Contributions from Soil Science. The soil scientist views soils as indepen- dent natural bodies. Each is composed of a unique combination of four soil forming groups and time: climates, living organisms, topography, parent materi- als, and age of land surface (Johnson, 1978). According to this concept, the kind, thickness, and arrangement of soil layers in the profile are the result of interaction of temperature and moisture and their seasonal distribution; and of vegetation, animals, humans, and micro-organisms on the parent materials in particular topographic situations over time. Since soil is one of the fundamental components of terrestrial ecosystems, it must be studied to determine how it interacts within these ecosystems in the production of food and fiber products. Soil resource inventories provide data for the study of this part of the ecosystem. At best, these inventories are incomplete records of soil conditions within an area. They are documents that record the geographic distribution of 28 unique kinds of natural soil bodies on maps and identify the properties that characterize them (Cline, 1978). The three major elements of a soil inventory may be noted: characteriza- tion involves the field and laboratory study of soils. It is based on grouping individual pedons (observed elements) into taxons to define the ranges in their characteristics or properties, and to understand the evolution of landscape relationships of the soils present. The second element, classification, is done in such a way that one soil varies from another in one or more properties to such a degree that the combinations of all the properties result in different responses to management for growing crops and facilitate the evaluation of production potential (USDA, Soil Taxonomy, 1975). When consistently and reliably done, soil classification makes possible soil surveys that are consistent between different places and that include accurate interpretations of soils for specific uses. The Taxonomy, with its hierarchical classification, provides the flexibility needed. There are six categories in the classification. In order of decreasing rank and increasing number of differentiae and classes, the categories are order (10 in the U.S.), suborder (47), great group (185), subgroup (970), family (4500), and series (10,500). The most heterogeneous category is the order; the most homogeneous is the series. Further, phases may be used as subdivisions of any category; and even phases of series may vary in their productive potential. This dissertation deals with the great group, subgroup, and family taxa since they have frequently been utilized in national/regional resource analysis. At the great group level, the soil scientist is able to consider the whole soil, the assemblage of horizons, and the most significant properties of the soil. Subgroups are subdivisions of great groups. In addition to the processes that have dominated soil development, the soil scientist also considers subordinate proper- ties that are marks of important processes. At the still more detailed family 29 level, the intent is to group soils within a subgroup that have similar physical and chemical properties which affect their responses to management. The agronomic responses of comparable phases of all soils in a family are nearly enough the same to meet most needs for practical interpretations of such responses. The purpose of the family (and also series) category is primarily pragmatic in contrast to the conceptual natures of the higher categories. Phases add to the pragmatic value of all categories. The third element, mapping, is done in order to be able to make more precise statements about the mapped subdivisions of the area than about the area as a whole (Beckett, 1968). Descriptions of the soils including statements about their morphology and their important qualities and the proportions and patterns of the soils in each map unit are described in the text accompanying the map. Soil inventories are made for many different purposes. Some purposes, such as farm planning, require refined distinctions among kinds of soil that occupy small homogeneous soil areas (usually phases of soil series). Others, such as national planning, require a broad perspective of the soils of large and commonly heterogeneous but distinctive areas (usually phases of the highest categories). The attributes of the maps and the intensity of application of the methods by which they are made vary greatly within this range. Theoretically, basic input elements of a soil inventory can be adjusted to provide the most useful product for the principal intended purposes. Different intensities of field study, different degrees of mapping detail, different phases or levels of abstraction for defining units, and different map unit designs produce soil surveys of widely ranging applicability (USDA, Soil Survey Manual, 1975, with parenthetical amplifications). The desirability of different levels of detail in a soils inventory is a function of the amount of information that is needed and can be utilized for the specified purpose as well as on the availability of base maps. If topographic maps, for example, are unavailable, other sources such as aerial photos, radar imagery, 30 satellite imagery, mosaics of such photos and/or imagery, planimetric maps, orthophotos, tabular information, or other information, would have to be used. To a much lesser extent, the availability of other land resource information (weather records, geologic maps, vegetation maps, weather station data, land use maps, etc.) influences the degree and/or feasibility of soil resource inventories of differing levels of detail. Collaboration with Other Disciplines. Climatology and botany are important to conceptualizing the stated area of concern. Soil productivity is a synthesis of conditions of soil fertility, water control, plant species and varieties, pest control, and physical environment. Plant growth depends on the environmental conditions in the immediate vicinity of the plant. Plants, as integrators of a variety of environmental variables, are potentially better indicators of climate than many instru ments are because they are continuously receiving climatic inputs. Due to a scarcity of meteorological time series data, field verification of vegetative assemblages is frequently necessary in order to understand the climate. Environ- mental conditions (temperature and precipitation) that the botanist can help identify are a required datum to specify certain soils taxa. Consideration of both soils and climatic data make possible the identification of areas that demonstrate similar production potential. Resource interpretation for more than broad suitability ratings for general crop groups requires the added input of agronomy. Soil productivity is that quality of a soil that summarizes its potential for producing plants under defined sets of management practices. Comparison of soil and climatic requirements of plant cultivars found in a region, with the characteristics of the agro-ecological areas, is the basis for a suitability assessment. The agronomist describes and defines regional cropping practices and systems, both current and possible, and estimates expected crop yields for each agro-ecological area under a specific management method. 31 Operationalization of These Conceptual Inputs Operationalization of this conceptualization of production potential requires establishment of data categories and gathering of those data. The CRIES information system does these things by means of the following components: a. A resource classification that defines land units in a way consistent with established taxonomy of soils. b. Land units that are sufficiently homogeneous with respect to plant productivity and adaptability to be reliably depicted by agronomic parameters for national agricultural resource analysis. This land unit, the Production Potential Area (PPA), is conceptually mappable, but this has been considered unnecessary given the area of concern and level of planning. The definition of a PPA follows: A PPA is an aggregate area of individual soil bodies and associated climates within a Resource Planning Unit (defined below) which is sufficiently homogeneous with respect to plant adaptability, potential management requirements, and producti- vity to be reliably depicted by unique estimates of those parameters for national analysis and planning (USDA and MSU, 1980, p. 3). c. Land units that are spatially identified to permit the association of representative social and economic data to the agronomic areas. These landscape units are composed of discernable patterns of PPAs. They are homogeneous only in their repeating heterogeneity of soils and climate.1 1In the CRIES system, these are called Resource Planning Units. They are defined as a geographically delineated unit of land that is relatively uniform with respect to land forms, kinds, and patterns of soils, climate, and vegetation (ibid., p. 3). 32 d. Representative sets of agricultural production practices and costs and the spatial identification of these representative sets to map units and then to PPAs. Interpretation and Analysis. The next component of an information system is interpretation and analysis.1 The analytical tools of interest to this dissertation are soil maps and cost-minimizing linear programming (L.P.). Map information content involves relationships found in the number, size, shape, frequency, and pattern of the various soil delineations of each map unit. Map units are rarely considered pure--composed of only one kind of soil body--but are variable to differing degrees. Another map characteristic is scale. As map scale decreases, ground area represented by one square centimeter of map area increases. In order that map delineations be legible, increasingly aggregative and increasingly heterogeneous soil area are required at increasingly smaller map scales. Although map scale is frequently expected to indicate the possible uses of a map, the level of detail incorporated in it, and the kinds of interpretations that could be made, these are often erroneous expectations because of the differences in inventory methods upon which a given map is based. Eswaran notes that the proper decision on map scale requires one to determine the following: first, the purpose of the survey; second, the degree of precision desired for the location of map unit boundaries, and the maximum percentage of error acceptable; third, the size of the smallest map unit which will not exceed this maximum error; and, lastly, the map scale at which the user's minimum area of interest is the same as this smallest acceptable map unit (Eswaran, 1978). lData measurement is discussed in detail in Chapter IV. 33 Soil inventories with excessive detail may provide more information than can be used effectively and efficiently for a particular purpose. Excess information is manifest in too many map units, too many map unit components, or maps too large or in too many sheets to be used conveniently. The size of the study area or country then, is roughly correlated with the number of resource units at any given level of intensity (Knox, May 1979). Linear programming (L.P.), the second! tool of interest, has been in use by agricultural economists since the early 19505 "to specify the 'optimum' organiza- tion of resources and enterprises on farms, , profit maximizing mixes of commodities produced , cost minimizing methods of processing products , spatial equilibrium patterns in the flow of agricultural products, 'optimum' interregional patterns of resource use , and to solve related problems" (Heady, 1971, p. l). L.P. is a budgeting procedure that breaks the production process into a series of linear relationships or activities. The ratios between any two inputs and between any input and the output are fixed and hence independent of the level of production. By comparing the single input-output ratio of one method of doing an operation to the ratio of another method and then comparing the two methods with the factor price ratio, the most advantageous method can be determined. The linearity assumption for any method of production does not imply that only constant returns to scale may be handled by an L.P. "Diminishing returns" can be incorporated by specifying additional methods of performing an operation. In Figure 3-2, a series of independent input-output points, each a defined method of production, are connected to suggest an iso-product "curve" for a product Y as well as a production function. However, in order to produce Y in a specific way, inputs X1, X2, and X3 must be combined in a definite proportion, i.e., the inputs are perfect complements. Inasmuch as dy/dxi cannot be defined for perfect complements, the function is not defined. 34 X2IX3 Figure 3.2 -- An Iso-Product "Curve" Any L.P. problem has three quantitative components, they are: a. C. An objective function that is to be either maximized or minimized. Often profits are maximized or costs are minimized. The existence of alternative processes by which the objective may be reached. A process is a method of converting resources into a product. Two processes are the same if they use the same resources or restrictions in the sine proportions. A process may be represented by one point of the production function concept used in marginal analysis. Where a production function is linear, it is defined as different levels of the same process. A nonlinear function is a collection of different processes. A restricted or limited availability of resources. Basically, some combination of processes, each of which has resource requirements, is found that maximizes or minimizes the objective function subject to two restraints: 35 (l) the total amount of resources used must be equal to or less than the supply; and (2) the level of any process must be equal to or greater than zero. In addition to the linearity assumption and those stated earlier for static economics, L.P. has several additional assumptions: a. Additivity. Processes must be additive in the sense that when two or more are used, their total product must equal the sum of the individual products. No interaction between certain inputs can occur. If there is interaction, the entire combination must be treated as a single process. The additivity assumption also implies that the sum of the resources used in each process equals the total resource use of all processes. Divisibility. All non-negative levels of input use and output are production possibilities. Resources and products are assumed infinite- ly divisible; the lumpiness of inputs and outputs is ignored. Inputs are perfect complements; outputs are perfect substitutes with respect to contribution to the maximand. Finiteness. It is assumed that the number of alternatives and resource restrictions is finite. CHAPTER IV A METHOD TO EVALUATE IMPACTS OF ALTERNATIVE LEVELS OF DATA DETAIL A major criticism of information systems has been that the data detail employed is not appropriate for either the identified problem, the analytical tools used, or the needs of the user. The purpose of this chapter is to present a method to understand more clearly the process of producing information. The method will involve first, varying the degree of detail in one data input, soils, and studying the change in information produced (output) by the analytical tools employed--soil maps and a least cost linear programming (L.P.) model; and second, varying both the degree of detail and crop yield estimates and making a similar study of change in information produced. The other steps in the generalized information system presented in Figure 3-l--the real world subject matter area and conceptualization of it, data classification, storage, and retrieval, analytical tools, and the user--are not varied. First, the user group, study area, and the problem to be analyzed will be described. Next, three soil inventories, each of a different level of aggregation, will be presented. Data from the inventories will provide inputs to a) three L.P. models that differ from one another not in specification but in the number of activities and constraints stemming from the number of principal soil components (PPA's), and to b) two L.P. models that also differ in crop yield estimates. Performance indicators will be developed that will suggest, by showing differences in the data inputs and information outputs, the nature of the information production function. It must be emphasized that this analysis is only 36 37 a partial one. Interaction with users regarding clarity, validation, verification, and workability of alternative levels of detail and hence their usefulness is a step not in the scope of this thesis. This step would be required for a complete analysis. User Group, Problem Identification, and Study Area Policy analysts and planners concerned with the capability of the Cibao Valley and major regions within it to produce food and fiber for both its own and for non-Valley requirements are the w of the information systems being constructed. In this regard, the Cibao is treated as if it was a country. The professional interests of this group of planners would be diverse, as they in fact are in the Subsecretary of Natural Resources. This diversity might include remote sensing, agrometeorology, forestry, soils science, agronomy, range science, and agricultural economics. Within agricultural production potential, the general area of concern, the problem to be analyzed is how to use the land resources of the Cibao Valley in the least costly way to meet current and future crop requirements, both internal and external. In order that the analysis not lose credibility among hypothetical users due to lack of clarity about the analyst's value assumptions, the following limitations are identified: 1. Only a partial, noncomprehensive, analysis is to be made. The focus is primarily on one aspect of production potential--the physical resource possibilities. The region's ability to realize these possibilities or even the desire to try will depend on societal institutions, e.g., land tenure, educational levels of farmers, pricing policies, and customs, that are exogeneous to this analysis. The impacts of resource use changes on other sectors of the economy, such as transportation, and even on aspects of the agricultural sector itself, such as input prices, are not considered. 38 2. Linear programming solutions are not assumed to be sufficient per se for making "real world" decisions since the four normative pre- conditions for maximization (of minimization) discussed earlier cannot be substantiated. 3. The analysis must be interpreted with recognition given that informa- tion on the reliability of the parameters included in the analysis is largely absent. 4. For these reasons, the information provided by this particular informa- tion system cannot be used alone for decision-making. However, it can serve as a needed input to the users' decision-making process. Methods used to develop each of the data inputs are professionally acceptable. The data themselves are considered internally consistent, and their behavior corresponds with reality as previously experienced by the professionals involved. The Study Area is the Cibao Valley. There are two major reasons for choosing this area: first, it is a very important agricultural area in a developing country; second, a sufficient amount of data on soils and general land cover is available from previous work. In its 1968-1985 Development Plan, the National Planning Office delineated four national planning regions. The most important, with an estimated 39 percent of the nation's land area and 51 percent of the 1971 population, was the greater Cibao Valley. The highly productive flatlands of this Valley were studied by the FAO between 1969 and 1974. Somewhat more than 70 percent of this 6,299 square kilometer flatlands area was estimated to be in farms. Of the total land in farms, 44 percent was estimated to be either in cultivation or idle; 27 percent was in cultivated pasture, 18 percent in natural pasture, and 11 percent in other uses. The agricultural sector of the Study Area contributed 24 percent of the national agricultural product in 1971. It was also the principal source of area employment. 39 With respect to national exports, the Cibao accounted for 80 percent of those from tobacco, 42 percent from coffee, and 27 percent from cocoa. In terms of production, the Cibao produced 90 percent of the nation's tobacco, 33 percent of its coffee, 25 percent of its cocoa, 29 percent of its rice, 53 percent of its plantain, 46 percent of its yucca, and 32 percent of its sweet potatoes. As part of the FAO study program, a semi-detailed soil inventory of 222,000 hectares was done for the flatlands between Monte Cristi on the west and Pimentel-Cotui in the east (FAO, Informe Tecnico 6, 1974). It is this area that is being studied in this dissertation. Two major river systems drain this inventoried area: the Yaque del Norte flows westward from just east of Santiago to the Caribbean Sea at Monte Cristi; the Licey, Cenavi, and Camu Rivers in the eastern end of the Study Area are major tributaries of the Yuna which is just east of the Study Area. The Study Area is quite homogeneous in many respects. The soils are almost all nearly level, very deep, and formed in water-deposited sediments. Most are medium, moderately fine, or fine textured. Much of the area is cultivated and irrigated. On the other hand, there is a great range in precipitation. The climate is humid in the east and arid or semi-arid in the west. Soil moisture regime is probably the most important variable characteristic. 1 staff Three major regions were identified by the CRIES-assisted SIEDRA generally taking into account population, climate, type of agriculture and economic activity, and the resource base. The Western Region extends from Monte Cristi to Esperanza. Except for a small area east of Esperanza, the region is dry, with average annual rainfall of 400-800 mm., and hot, with an average annual temperature of 27°C. The region has slopes of 0-396. No cropping is possible without irrigation, but with irrigation, the soils are quite productive. lSIEDRA is the Spanish acronym for Inventory and Evaluation System of Agricultural Resources. SIEDRA is located within the D.R. Subsecretariate of Natural Resources. 40 Flooding is generally not a problem; salt and sodium accumulation is found where irrigation and drainage practices are unsuitable. The major crop is rice, with some plantain, sugarcane, and natural pastures. To the south and east of Esperanza, where precipitation is higher (probably 1,200-l,400 mm.), more plan- tain and some tobacco are found. One major east-west highway and numerous hard-surfaced feeder roads are located in the region. Only three notable centers of population exist: Monte Cristi with 15,800 persons (1971), Esperanza with 22,700 persons, and Valverde (Mao) with 46,600 persons. The Central Region extends from just east of Santiago to Esperanza. Physiography changes to include some rolling plains with slopes of 3-15 percent and scattered others of up to 30 percent. Along the Yaque del Norte River are the flood plains noted above. Average annual rainfall is 800-l,400 mm. in the rolling plains areas and 1,200-l,800 mm. along the Yaque. Supplemental irrigation is generally needed although not always available. Major crops are tobacco in the rolling hills and rice and plantain on the lower terraces. The transportation network is similar to that in the West except that it is quite extensive around Santiago, a city of 268,400 persons in 1971. The East Region extends from Licey, east of Santiago, to Pimentel-Cotui, just outside the eastern rim of the Study Area. Physiography returns to the 0-3 percent slopes of the West. Terraces of stream, lacustrine, or marine origin and alluvial foot slopes occur in this area. This is the wettest of the three regions with average annual rainfall ranging from 900-1,500 mm. in most of the region to l,600-2,000 mm. on the easternmost side. Irrigation is required only for rice. Land use includes much interplanting of food crops (corn, beans, yucca, plantain, sweet potatoes, and tobacco) in the west and central parts to extensive and highly productive rice and pastures in the southeast; cocoa is found in the hillier areas. Except for those used in some of the rice fields, production methods here, as in other regions, are quite simple. A number of hard-surfaced roads transect the 41 region. Population centers are found all along the edges of the region: Moca - 96,200; Salcedo - 40,500; San Francisco de Macoris - 133,800; La Vega - 165,500; and Fantino - 16,000. . Three Cibao Valley Soil Inventories Three soil inventories were constructed: Detailed, Intermediate, and Aggregate. Ideally, the three inventories would have been developed completely independent of one another, perhaps by different soils scientists using the same technical guidelines. Inasmuch as this was not possible, a soils scientistl prepared the three inventories for the purpose of permitting analysis of the Valley's agricultural potential. Each was developed according to the following criteria: Detailed Inventory: 1) Map units are properly described as phases of soil family taxa; 2) Map unit delineations are legible at a 1:1 00,000 map scale. Intermediate Inventory: 1) Map units are properly described as phases of soil subgroup taxa; 2) Map unit delineations are legible at a l:250,000 map scale. Aggregate Inventory: 1) Map units are properly described as phases of soil great group taxa; 2) Map unit delineations are legible at a l:l,000,000 map scale. The FAO map was published at a scale of l:50,000. Its map units are mostly consociations or associations of phases of "series." The FAO "series" (Knox, 1978) are not taxonomic subdivisions of soil families so do not fully qualify as series. They are more broadly and loosely defined than required by current U.S standards. They are classified to the subgroup level of the soil taxonomy; however, some of the classifications are contrary to other information about the soils, and some of the series seem to be broader than the subgroup to which they are assigned. Some lEllis Knox, Salut Incorporated, Columbia, Maryland. 42 FAO map units, particularly those in flood plain areas, are very vaguely defined, loose associations or undifferentiated groups of suborders and great groups. The number of "series" and map units is excessive compared to the amount of information given for each unit. In the Study Area, 55 distinct "series" and 9 associations of "series" are identified. As a result, the FAO map shows a number of boundaries of which the significance is not at all clear. The FAO report was augmented by information from the 1967 OAS study of the natural resources of the D.R., by the previously prepared CRIES national-level soils inventory, and by brief field observations. The OAS soil map was published at a scale of l:250,000. Map units are loosely defined associations of very broadly conceived series. They are "broadly conceived" in that The OAS report does not attempt classification of the soils, and the descriptive material provides a skimpy basis for classification. I The third data source, the national CRIES soil map, was previously drawn at a scale of 1:250,000. Map units are associations of phases of subgroups. The Detailed soil map was drafted on the FAO map at 1:50,000 scale, but its delineations were drawn for presentation at 1:100,000 scale. See Figure 4-1 in the packet. The 31 map units are associations of phases of families. Ninety-five delineations and 45 unmapped principal components, PPAs, of the map units were also defined. The soil classification of the FAO report was used in most cases, but Salorthids were changed to Camborthids, and the aridic moisture regime was expanded at the expense of ustic moisture regime. The classification of soils as Mollisols or Alfisols was accepted in large part. (In the national CRIES map, Argiustolls and Haplustolls were reclassified as Ustropepts, Argiaquolls as Pellusterts or Fluvaquents, and Tropaqualfs as Tropaquepts.) Figure 4-2 is a sample Map Unit-PPA form used for all three soil inventories. The Intermediate soil map, Figure 4-3, closely follows the national-level CRIES map. It was also drawn on the FAO soil map base at l:50,000. The 43 Map Unit 01 Characteristic Unit PPA 011 PPA 012 General Corresponding Map Symbol Ang 2 Mt from FAO Soil Map F0 2 - Position in Landscape Terrace Terrace Ave. Annual Rainfall mm. 1600-2000 1600-2000 Ave. Annual Temperature 0C. 25-27 25-27 52$ Principal Soil Component Abruptic Tropoquolf Abruptic Tropoquolf Slope 96 0-3 0-3 Depth to Bedrock cm. 150+ 150+ Surface Texture Medium Medium Stoniness Not Stony Not Stony Permeability Very Slow Very Slow Drainage Imperfect Imperfect Flooding None None Reaction Slight/Moderate Acid. Strong Acidity Base Saturation 96 50+ 50- Salinity Nonsaline Nonsaline Figure 4.2-- Cibao Valley Map Unit-PPA Sample Form 44 Figure 4.3 - Intermediate Soil Map of the Cibao Valley, Scale l:250,000 (in packet) 45 criterion for legibility is specified at l:250,000 scale. Seven map units, described as associations of phases of subgroups, ll delineations, and I4 PPAs are defined. The dominant textural phases of the subgroups are specified so that interpreta- tions can be made as though the components were families. Contrasting soils of minor extent and more extensive intergradational or intermediate soils that are identified on the family-level map are ignored at the subgroup level. The Aggregate map, Figure 4-4, was also drawn on the FAO soil map at l:50,000. Only three map units, each with one delineation, are shown. They separate the udic, ustic, and aridic moisture regimes. They are associations of phases of great groups, but because the total range in soil characteristics in the Study Area is very limited, the nine PPAs are described and interpreted more or less at the family level. Following common soil survey practice, at all three levels the Soil Taxonomy is used to organize information, guide the design of map units, and name PPA components of map units. Almost never does a component cover the whole range of the class used to name it. Within the restricted universe of the Cibao Valley, relatively narrow ranges are presented by Taxonomy classes even at the great group level. By emphasis on the dominant and representative segments of the soils, use and management interpretations can be considered more or less at the family level with all three maps. Taxonomically, each Aggregate map unit could be a perfect composite of Intermediate map units which in turn could be perfect composites of Detailed map units. However, because of the intermingling of soil bodies, difference in map scales at which the three maps were drawn and need for legibility, the carto- graphic correspondence is not so exact. Table 4-1 presents the taxonomic relationships as modified for cartographic needs between the three levels of map units. In several instances, a Detailed PPA, 051 for example, belongs to more 46 Figure 4-4 -— Aggregate Soil Map of the Cibao Valley 47 Table 4-1.- Relationships of Map Units and Their Component PPAs Aggregate Map Unit 1 Aggregate Intermediate Detailed PPAs by Dominant, Similar, and PPAs PPAs Intermediate Map Unit Included Soil Subgroups A G 11 A1 011, 012, 013, 042 Abruptic Tropaqualfs A1 021 Aeric Tropaqualfs A1 031 Aquic Haplustalfs 12 A2 041, 051 Udic Chromusterts A2 081 Aquic Chromuderts 13 A3 101 Fluvaquentic Eutropepts G1 311 Aquic Ustifluvents G3 313 Typic Ustifluvents G2 312 Aeric Tropic Fluvaquents Aggregate Map Unit 11 Aggregate Intermediate Detailed PPAs by Dominant, Similar, and PPAs PPAs Intermediate Map Unit Included Soil Subgroups B C 111 B1 061,071 Udic Pellusterts B1 051,091 Udic Chromusterts B1 081 Aquic Chromuderts 112 C1, F2 191,201 Fluventic Haplustolls C1 171 Cumulic Haplustolls Cl 181 Typic Haplustolls Bl, C1 151 161 Typic Argiustolls Cl 131 Cumulic Haplaquolls 82 092 Typic Haplaquolls Bl 121 lypic Argiaquolls 113 F1, G3 301 313 Typic Ustifluvents Bl 072 Aquic Ustifluvents G1 311 Aquic Ustifluvents G2 312 Aeric Tropic Fluvaquents :T :D to. .'-.'.\. [TVA .1; .—< t——‘ 1 I Table 4-1.-- (continued) 48 Aggregate Map Unit 111 Aggregate Intermediate Detailed PPAs by Dominant, Similar, and PPAs PPAs Intermediate Map Unit Included Soil Subgroups D E 1111 E1 271, 291 Ustic Torrifluvents E1 262 Ustic Torrifluvents (Saline) E1 272 Aquic Ustifluvents El 281 Aeric Tropic Fluvaquents E1 282 Aeric Tropic Fluvaquents _ (Saline) 1112 E2 141 Typic Haplaquolls 1113 E3 241, 292 Fluventic Camborthids E3 251, 261 Fluventic Camborthids (Saline) E3 242 Fluventic Camborthids (Alkaline) D1 221 Pachic Haplustolls E3 211 Aridic Haplustolls E3 232 Fluventic Haplustolls E3 231 Fluvaquentic Haplustolls E3 111 Aquic Ustropepts 49 than one Intermediate and/or Aggregate map unit. Appendix Table A-20 presents the relationships between PPAs at the three levels of aggregation. Soil Map Comparison Factors Soil map evaluation suggests an assessment of the accuracy with which different soil areas are indicated and of how useful the map is to users. While most users do not have the opportunity to "ground truth" soil maps, an investiga- tion of the methods to prepare the inventories can indicate which of several maps may more closely correspond to actual resource distribution. Secondly, estimates of relative costs, a factor important to the workability of any proposed survey, can also be made by studying the methods to be used. Finally, the map itself can be inspected for its information content, that is, the relationships found in the number, size, shape, frequency, and pattern of soil areas on a map; soil variability; and map legend characteristics. Various comparison factors have been suggested by the soil science pro- fession to compare soil maps of different intensities. The factors may be placed into two groups: those based directly on each soil map and those based on assumptions as to how the inventories would have appeared had they been constructed independently. See Figure 4-5. A description/definition of each factor and a comparison for each soil map will be made in the next chapter. Five Linear Programming Models Five cost-minimizing L.P. models were built to analyze how to use the soil resources of the Cibao Valley in the least costly way to increase crop production. The models have identical structures. Three models (Detailed, Intermediate Weighted, Aggregate Weighted) differ only in the disaggregation of the Valley's land base as provided by their respective soil inventories. Two others (Intermediate and Aggregate) differ from the Detailed both in disaggregation and in crop yield estimates. The models are II. III. IV. 50 Methods of Preparation A. B. C. Base data collection Base maps adequazcy Observations/ Km Map Characteristics A. B. C. Scale 2 Observations/ cm Map delineations 1. Average size (ha.) 2. Least size (ha.) 3. Number 4. Area covered by lines (ha.) Map units 1. Taxonomic level 2. Phases 3. Soil variation 4. Purity 5. Agronomic interpretations Map Analysis A. B. C. D. Map legibility 1. Map texture intensity 2. Index of maximum reduction 3. Minimum scale of reduction Reliability of boundaries Measured vs. summed areas of map delineations Map unit descriptions Estimated Inventory Cost Figure 4-5.-- Soil Map Comparison Factors Based on Soil Map XXXX XXXXX Inventory Assumptions 51 1. Detailed (crop yields vary by PPA); 45 PPAs. 2. Intermediate Weighted (crop yields are area weighted averages of the Detailed PPAs comprising each Intermediate PPA); 14 PPAs. 3. Aggregate Weighted (yields are weighted averages of the component Detailed PPAs); 9 PPAs. 4. Intermediate (crop yields vary by PPA); l4 PPAs. 5. Aggregate (crop yields vary by PPA); 9 PPAs. Current use of the Intermediate-level map units and PPAs, by crop and production technique, was established and mechanically allocated to the Detailed and Aggregate PPAs through the use of percentage ratios discussed below. Thus, only one set of production methods and of production costs was used for all three sets of land areas. Two sets of crop yield estimates were made. In the first set, Aggregate Weighted and Intermediate Weighted PPA yields were simply weighted averages of the yields estimated for the component Detailed PPA's. In the second set, differences in land quality of the Detailed, Intermediate, and Aggregate PPAs were recognized by separately making yield estimates for the Detailed PPAs, then for the Intermediate PPAs, and then for the Aggregate PPAs. Hence, costs per unit of production vary. Neither the level nor the mix of variable factor proportions vary from one PPA to another. Algebraic Structure of the Basic Model. The structure of the model is as follows: Objective Function: MIN 2 = h’ijk (Cijk ' xhijk) " hrh'k (C'hh'k ' thh'k) Land Constraints: Z 7 1k xhijk Xhi for each h k=l,'",l3 for each 1 52 for each i jk xhijk $1°2xhijkcn for each h k=1- 3, 6- 9, 11 [k for each 1 hi ) Xhijkcn [(#13631017foreachh . x.. .<.X' .. foreachi jk h1jk h1jk k=I-3,6-9,11 CD for each h x. :X. foreachi k211k2 hlcn for each h x .. = X .. for eachi h1jk hijk k=4,5,10 cn for each h Transportation Constraint: 2 t for each R =1,'",10(11 is h hh'k h'k interplanted package of crops) for each h' Production is Transported: 1; (xhijk ’ Yuk) ‘ E thh'k z 0 12:23:: = 1’ ’11 Where: i xhijk Cl hh'k thh'k production potential area (PPA). In the Detailed model, i = 1 45; in the Intermediate models, i = l 14; in the Aggregate models, i = l 9. crop production method, j = l, 2, 3. crop, k = 1, ..., 13. 1-10 are individual crops in which areas of 4, 5, 10 are fixed; 11 is interplanted group of crops 2, 3, 6-9; 12 is pasture; 13 is idle. production region, h = 1, 2, 3. requirements region, h' = l, 2, 3, 4. cost of producing crop k by production method j on PPA i; (pesos/hectare). area of crop k under production method j on PPA i in region h; (hectares . cost of transporting crop k from producing region h to requirements region h'; (pesos/unit of production). units of crop k transported from producing region h to requirements region h'; (units of production). 53 Xh = area available for crop production and pasture in producing region h; (hectares). X'hi = one-half of area of PPA i in Region h not used for any type of production in Current Normal. y.. = estimated yield of PPA. for crop k under production method ijk . . . 1; (units of production/hectare). Rh'k : requirement for crop k in requirements region h'; (units of production). Z = total cost of crop production and transportation; (pesos). cn = Current Normal conditions In obtaining the optimal solution subject to the constraints noted, land in pasture, and several small area crops (banana, cocoa, and sugarcane) were fixed at their present Current Normal levels.1 In order to respect implicitly the socioeconomic factors that have determined current PPA use, L.P. solution values of PPA areas in each crop-production technique combination were restricted to 70-120 percent of their Current Normal levels in the Benchmark run and to 70-130 percent in the Growth and Growth/Shift runs (more fully defined below). In those instances where some crop-production technique is not found on a PPA, 50 percent of the total area of the PPA is available for use by the L.P. For example, suppose the area of PPA 01 is 4000 ha. and only 500 ha. is in use (corn-low technique). The relevant constraints on use of PPA 01 would be: a) corn/low area could range from 350 to 600 ha., 70-120 percent of 500 ha. lThe L.P.‘s were run on an IBM 370-138 using the Magen L.P. system. The Haverly optimizer takes as input an L.P. matrix and converts it to the form required to produce an optimal solution. Because the optimizer does not have the ability to bound columns, the 70-120% bounds had to be done with rows. As a result, the sizes of the matrices were increased considerably. The optimizer is most efficient when the number of rows is less than 1800. Removing cocoa, sugar cane, and banana from the matrix was necessary to reduce matrix size. The size of the three matrices was: Rows Vectors Elements Detailed 2044 3076 7531 Intermediate 795 1256 3066 Aggregate 600 955 2308 54 b) beans/low (a crop-technique combination not currently found on PPA 01) could enter the L.P. solution on up to 2000 ha. on PPA 01, 50 percent of 4000 ha. Linear Programming Solution Comparisons Without changing specification of the L.P., the model was run for three policy questions at three levels of soil aggregation: Detailed, Intermediate Weighted, and Aggregate Weighted. In addition, the L.P. was run at these three levels but with different and "independent" crop yield estimates (see section below on "Crop Yield Estimates") for each level. These latter three L.P.‘s, the Detailed, Intermediate, and Aggregate differ both in the degree of soil aggrega- tion and yield estimates. The three policy questions are: given the Current Normal pattern of cropland use, 1. What would be the most "efficient" (least costly) means of meeting regional crop requirements? This the the Benchmark run. 2. What would be the most "efficient" means of meeting regional requirements that are 20 percent greater than those of the Current Normal? This is the Growth run. 3. What would be the most efficient means of meeting regional crop requirements that, in addition to experiencing an overall 20 percent growth, also reflect shifts in population from rural regions to the Santo Domingo region by increasing the requirements of this latter region for non-export crops relative to the other regions. This is the Growth/Shift run. The following comparisons will be made between the Detailed L.P. and each one of the four aggregated L.P.‘s: 1. Absolute solution values for regional and Valley crop area, production, production cost, transportation flows, and value of the objective function. 55 A summary comparison statistic, Theil's Inequality Coefficient, U, will be used to compare Regional and Valley changes, from Current Normal values, of cultivated area and production as calculated by each of the models. The formula for U is: ~’ >3 (021’011)2 [0112 where 011 are the differences between the Detailed L.P. solution and U : the Current Normal value of some parameter i; 02i are the differences between the Intermediate Weighted L.P., for example, and the Current Normal. When there is perfect correspondence between all of the changes predicted by the Detailed and another model, U is zero. Non- negative, it has no upper limit. This paper does not attempt to determine the significance of any particular U value. This can only be done by the user of the information produced by the L.P.‘s. Results will also be evaluated by computing bias (average signed error) in the more aggregated models with respect to the Detailed model by determining if their solution values are either larger or smaller than the Detailed values on the average. If the weighted sum of the overestimates equals the weighted sum of the underestimates, the bias is zero. The formula for computing bias is 0 0-0 o 21 I1 ——0. ) ]x 100 Bias = [2013' 1 I where 021 is, for example, the Aggregate Weighted solution value and OH is the Detailed value for the specific variables noted above, and W1 =_2£ This formula can be simplified to £011 2: ) .(0 .-0. Bias = 12%; 1‘ x100 11 56 No estimates were made of economic data gathering and assembly costs since only one land use inventory was undertaken. Relevant records would require developing data independently for each model. In addition to the above and in order to suggest further the loss of information from land aggregation, the variance of Detailed PPA yields around the Intermediate Weighted (Aggregated Weighted) yield of each respective Intermediate (Aggregate) PPA and the coefficient of variation will be calculated by crop. The formula for variance is Z (Vi-1’)2 Variance : = O n-l where y is the weighted average yield of the crop on the Intermediate (Aggregate) PPA and yi is the yield on the Detailed PPA. The formula for coefficient of variation is: % where 9 is the mean yield and o is standard deviation. Percentage bias among the Detailed, Intermediate, and Aggregate sets of yield estimates will be calculated to suggest in a general way the significance of additional measurement error incurred due to making "independent" crop yield estimates for PPA's at each level of aggregation. '3 O'i'oki Bias =[ _ (w ° —1——)] x100 k-l k 0ki where Oji is the weighted yield of crop i on Intermediate PPA j; Oki is the yield of crop i on the kth Detailed PPA composing that Intermediate PPA; Wk = A_k_ where A. l Ak is the area of the Detailed PPA and Aj is the area of the Intermediate PPA. Methods to Estimate L.P. Coefficients Map Unit Areas. Area measurements of each map delineation on each of the 14 sheets of the 1:50,000 FAO base map were made by planimetering the 57 boundaries as drawn on the base map. The summation of delineations composed of the same soil type resulted in area measurements of each map unit. Table A-1 contains these data. The total measured area of the Cibao Valley varies from one map to another due to operator errors in using the planimeter. The Detailed 2; the Intermediate measurement was 2221.3 km2; and the Aggregate measurement was 2286.3 kmz. measurement was 2179.5 km PPA Areas. Since PPAs are unmapped, indirect means had to be used to estimate areas. Two different methods were used. The Intermediate soil inventory records an approximate percentage composition of each Intermediate map unit by its PPA components and for "undescribed inclusions." This percen- tage was allocated to the PPAs in the map unit on the basis of their area. See Table A-2. The Aggregate soils inventory also records the PPA composition of Aggre- gate map units. As was done with Intermediate PPAs, Aggregate PPA area was adjusted to include the undescribed inclusions. See Table A-3. This procedure could not be used for Detailed PPAs since the Detailed inventory did not include this quite specific area information about the unmapped PPAs. Instead, the original l:50,000 FAO base maps on which Detailed map units were drawn were studied to determine the FAO soil "series" within each. These "series" were then correlated with the PPA definitions of the Detailed inventory in which predominate "series" had been denoted. The FAO estimated area of each "series" in each map unit was recorded. Realizing that map units often have dissimilar inclusions because of cartographic construction, a m estimate of the percentage of each map unit's area occupied by each "series" was made. With no FAO "series" information in map units 29-31, an arbitrary allocation was made. An example of the procedure used for Detailed PPAs follows: 1) Detailed map unit 9 has 2 PPA - 091 and 092. 2) The major FAO soil "series" by map symbol in each PPA are: 58 091 - Ar 092 - PC 3) FAO estimated Ar area at 500 ha. and PC area at 900 ha. 4) Using 1,400 ha. as equivalent to the size of map unit 9, 091 is considered to be 3596 and 092 to be 6596 of this Detailed map unit. Percentages such as these were applied to the planimetered Detailed map units to estimate measured PPA area. See Table A-4. In order to test the effects of land aggregation, it is important that the area of an Aggregate PPA be equal to the sum of the Intermediate PPAs that comprise it and that the area of an Intermediate PPA equal the sum of its component Detailed PPAs. Since PPAs are not mapped, the method of determining PPA measured area is only an approximation. Separately measuring the Aggregate, Intermediate, and Detailed map units and then allocating to PPAs did not permit these equalities. Detailed and Aggregate PPA areas are "normalized" to the Intermediate PPA areas to provide these equalities. The Intermediate map unit measurements and PPA percentage composition are used as control totals. Since they do not correspond to measured area of their respective Intermediate PPA, Measured Detailed map units and calculated PPA areas are used only to establish relative percentage occupancy of the Intermediate PPA area (the control total). The same method was used to allocate Intermediate PPA area to each Aggregate PPA. The example below demonstrates the normalization method and calculation of Detailed PPA areas. The percentages, 19.896, 14.796, and 35.5% are used to allocate A2 area to Detailed PPAs 041, 051, and 081. See Tables A-5 and A-6. Intermediate PPA Detailed PPA in A2 Int. Area Normalized Allocators Name Measured Name Measured Det. Area Det. PPA of A2 Area Area A2 6210 ha. 041 2430 ha. 0.5056 1228.6 ha. 19.896 051 5493 ha. 2777.5 ha. 44.796 081 4359 ha. 2203.9 ha. 35.596 12282 ha. 6210.0 ha. 100.096 59 Regions. Three production Regions--East, Central, and West--were identi- fied by the SIEDRA staff. Briefly the sparsely populated, dry West requires irrigation for crop production; irrigated rice is the major crop. The moister Central region has a more rolling terrain. Tobacco is the major crop in the uplands; rice and plantain are on the flood plains. Santiago, the second largest city in the country, is located here. The East is the wettest; irrigation is practical only to obtain a second rice crop. Interplanting of annual food crops is widespread. The East is composed of 90 percent of Aggregate map unit I and 69 percent of II; the Central is 12 percent of II; and the West is all of map unit 111 and small portions of I and 11. See Tables A-7 and A-8. Cropland Use. Due to time and budget limitations, the land use and agrono- mic data that were developed to describe the Cibao Valley's resource base are not considered sufficiently accurate to be actually used by planners who wish to analyze the area's agricultural production potential. The data were developed in order to illustrate the nature of impacts on information generated by a particular CRIES system, and in this respect, they demonstrate internal consistency and correspondence to reality. The data could be used to illustrate (in an ordinal way) how Cibao Valley production potential may be affected by alternative uses of its land resources. The methods used to determine land use were influenced by the total absence of secondary data for the Study Area. This absence occurs largely because existing secondary sources present data by administrative boundaries, such as provinces, rather than by the physiographic boundaries used to identify map units and PPAs. Given the lack of reliable secondary data sources, the non-coincidence of Study Area boundaries with administrative divisions, and the methodological purposes of this study, the author decided to rely on the informed judgment of persons familiar with the Cibao Valley and trained in soils, agronomy, and 60 agricultural economics to develop Current Normal estimates of land use by each crop-production method combination for the Intermediate Map Units and PPAs only. The first step to determine PPA use was done in collaboration with agronomist Francisco Rodriguez of the SIEDRA program in Santo Domingo. Drawing upon Rodriguez' previous field work and supervisory experience in cost of production surveys in the central and eastern parts of the Valley, upon national production cost budgets of the Agriculture Bank, regional cost of production budgets of the field offices (Northwest, North, and Northeast) of the Secretary of Agriculture, and upon the CRIES national soil map, a first approximation of the production regions just discussed, of the major crops and production methods found in each region, and production costs was made. Total area cultivated was the criterion for selection of major crops and production methods.1 This first approximation provided a framework for a three-day "windshield survey" directed by Ingeniero Jose Santiago, a SIEDRA soils scientist who had been the principal Dominican counterpart in the FAO soil survey of the Cibao. An agronomist with extensive experience in tropical cropping systems (including previous experience in the Cibao) was also a major participant.2 Utilizing the FAO l:50,000 topographic map showing land uses3 as of 1965, and major land use interpretations of December 1973 and July 1978 LANDSAT imagery,“ current lGiven other purposes of the study, other criteria such as the proportion of farms using a particular production practice or using a certain set (or specified range) of inputs, the proportion of production, the amount of labor inputs or of other specified inputs, foreign exchange earnings, etc., could have been used. Selection of what is major may vary yearly as markets, price, and climatic conditions change. "Normal" conditions with respect to such conditions were assumed. 2Dr. Trevor Arscott, Department of Agronomy, Ohio State University, Columbus, Ohio. 3 O O O O Categones 1ncluded plantam, cocoa, coffee, rice areas, sugarcane areas, and pasture. #Categories included sugarcane, forest, and intensive mixed agriculture. Scattered cloud cover limited interpretations east of Santiago. Interpretations were done by the Remote Sensing Project, MSU. 61 cropland use was visually identified in each Intermediate map unit. The FAO soil inventory included only the generally flat and rolling lands of the Cibao which were in agricultural use. Extensive areas in urban or forest use or regarded as unsuitable for agriculture were designated as "not mapped." These non-agricul- tural uses are thus not in the Intermediate land use. This identification served as the basis for revising Rodriguez' estimates of crop area and production practices. For each Intermediate map unit, percent estimates of land use by crop were made. These estimates were in turn modified and estimates were made for the map unit PPAs in discussion with agronomist Ramon Jimenez, the sub-director of the FAO—sponsored Centro de Desarrollo Agropecuario (CENDA) in Santiago, and his staff. The crop categories considered in the field reconnaissance were rice, tobacco, sugarcane, plantain, bananas, yucca, corn, beans, cocoa, sweet potatoes, a general "annuals" category, pasture and idle. See Table A-9. CENDA staff felt 30 percent of this general "annuals" area was interplanted. Interplanting is found only on those Intermediate map units largely in the East, that is, A, B, and a minor part of C. The remaining 70 percent of the "annuals" is planted to single crops. The interplanted hectare is defined to include com, beans, yucca, sweet potatoes, tobacco, and plantain. Except for rice, which is double-cropped in all three regions, only very limited double-cropping is practiced. Table A-10 shows the allocation of crop use, idle, and pasture to the Intermediate PPAs. Idle is assumed to occupy 20 percent of each PPA. For map units C, D, E, and F and the unmapped PPAs within them, the crop distribution percentages estimated in the field reconnaissance were multiplied by the total active cropland in the particular PPA to obtain crop area estimates by PPA. In order to consider interplanting, the allocation of cropland use on Inter- mediate map units A, B, and C differs from the other regions in treatment of the general "annuals" category. An example illustrates the assumptions and method: 62 Assumption: - 30% of the "annuals" area is interplanted and 7096 is planted alone. The relative distribution of crops in the 70% area is: yucca - 3096, corn - 2596, plantain, sweet potatoes, and beans -l596 each. Assumption: - of the interplanted area, 3096 is in corn and beans during the first planting and yucca and corn during the second planting; - 3096 is in plantain and beans; no second planting; - 10.96 is in tobacco and yucca; no second planting; - 3096 is in plantain, sweet potatoes, and beans; no second planting. Assumtion: - whenever plantain is interplanted, it occupies one-third of the cultivated area. Calculations for the yearly use of 100 hectares of interplanted "annuals": - 30 hectares in corn/beans and yucca/corn (equivalent to 15.0 ha. in corn, 7.5 ha. in yucca, 7.5 ha. in beans--two plantings); - 30 hectares in plantain/beans (equivalent to 10.0 ha. in plantain and 20.0 ha. in beans--one planting); - 10 hectares in tobacco/yucca (equivalent to 5.0 ha. in tobacco and 5.0 ha. in yucca- -one planting); - 30 hectares in plantain/sweet potatoes/beans (equivalent to 10.0 ha. in plantain, 10.0 ha. in sweet potatoes, 10.0 ha. in beans--one planting). In summary, the yearly use of 100 interplanted hectares is _3_7_._5_ hectares in beans, 20.0 hectares in plantain, 15.0 hectares in corn, 12.5 hectares in yucca, 10.0 hectares in sweet potatoes, and 5.0 hectares in tobacco. These crop totals 63 converted into percentages are used to determine Intermediate PPA use in the East. See Table A-10. Crop Production Methods and Costs. One to three "representative" produc- tion methods with an associated yield and variable on-farm production costs (pesos/hectare) were defined for each crop by the consulting agronomist, and the SIEDRA agronomist and soils scientist mentioned earlier. The specific methods, yields, and costs are in Table A-ll. Production costs would be expected to vary from PPA to PPA as fertilizer application rates, the amounts of labor for fertilizer application, etc., would also vary. For example, use of the same method for corn may yield 30 qq. on one PPA and 40 qq. on one of greater inherent productivity. Thus, harvesting costs on the second PPA should be one-fourth greater. Absence of reliable data on such relationships led the analyst to vary costs only by yields. Thus, a "representative" method with a cost of $60.00/ha. is applied to every PPA. Cost per quintal varies with yield: on the first PPA, cost is $2.00/qq.; on the second, $1.50/qq. The current distribution of these methods by Intermediate map unit and PPA was calculated using percentage allocators developed by the scientists cited above. Crop Distribution to Detailed and Agggate PPAs. The only land use inventory done was for the Intermediate map unit and PPA level. Allocation of crops and production methods to Detailed PPAs was done on the basis of the percentage compositions of Intermediate PPAs by Detailed PPA. Allocation to an Aggregate PPA was done on the basis of its composition by Intermediate PPAs. See Table A-12. Crop Yield Estimates. In the absence of any crop yield data by soil, estimates of yield potential for each crop were made by ordinally ranking the productivity of each Detailed, each Intermediate, and each Aggregate PPA, on an index of 0 to 100 under one of the "representative" production methods. Each index was converted to an estimated yield by multiplying the index by the yield 64 associated with the "representative" method. See Table A-ll for the base yields. The resulting yield estimates by PPA are consistent relative to one another. They are not valid in an absolute sense for lack of more information on PPA physical and chemical characteristics. These yield estimates are inputs to the Detailed, Intermediate, and Aggregate models. As discussed earlier, weighted averages of the Detailed yields were inputs to the Intermediate Weighted and Aggregate Weighted models. Conceptually, the amount of descriptive information was least for the soil great group (the Aggregate PPA) and greatest for the soil family (the Detailed PPA). As stated earlier, however, due to the restricted universe of the Cibao, the difference in information between soil subgroups was not as great as would occur worldwide or even in a much larger country. Limited information on soil properties such as salinity, reaction, base saturation (properties that directly affect various crop responses) was also used to estimate the yield indices; however, one soil property and dominant crop requirement often tended to predominate in estimating an index. In the case of paddy rice, poor drainage is desirable for good water control and management, but such a condition is detrimental for good banana production. An example of a crop's ability to attain the "normal yield" value on a particular PPA can be demonstrated by cacao. This tree possesses a deep tap root system which requires a relatively deep permeable soil. A soil with an impervious texture and structure such as an Aquic Chromudert would, therefore, be undesirable for good cacao production (Arscott, 1979). Factors such as slope and erosion potential were also considered. A given crop may do quite well for one cropping season on a PPA with a high erosion risk under the respective cropping practice. However, yields in following years may fall significantly due to the severe erosion produced by the first crop. Table A-13 presents the descriptions of Aggregate PPA Il, Intermedi- ate PPA A1, and Detailed PPA 011. 65 Transportation Costs. Tariffs established by the Dominican government to transport one quintal of produce from designated sites to Santo Domingo were used to estimate transport costs for the L.P. All map units and their PPAs that are in a production region have the common transport cost of the region. Trade may occur between any of the three production regions (East, Central or West) and any of the requirement regions (East, Central, West or Santo Domingo). These costs are presented in Table A-l4. Production Requirements. Relative crop production requirements in the four requirements regions were estimated in different ways. Rice marketing is based on the quantity and distribution of rice storage facilities. Tobacco is assumed to be required only in the Central Region since its production is totally controlled by the nation's major export houses all of which are located in this Region. Sugarcane is assumed to be evenly required between the West, the location of the Valley's only sugarcane mill, and the non-Cibao area south of the East Region. This non—Cibao area is part of the Santo Domingo Region. Corn is largely utilized in the poultry industry which is concentrated in the Santo Domingo Region. Cocoa is primarily an export crop. An estimated 90 percent is marketed through Santo Domingo for export. Current requirements for plantain, bananas, beans, yucca, and sweet potatoes are assumed to be based on the relative populations of the four regions. For these crops, as for those above, production from non-Cibao areas of the country is ignored. No differentiation is assumed between population parameters on diets between the regions. No production is derived from pasture. Tables A-15 and A-16 present percentage distribution and of crop requirements, and percentage distribution of crop requirements assuming a shift in population toward Santo Domingo. In the absence of independently derived production levels by crop and region, the product (crop-production method area per PPA) x (crop-production method yield on each PPA) is used to estimate current production of each crop. 66 These sums for each crop, one based on the detailed yields, one based on intermediate yields, and one based on aggregate yields, were calculated. The least restrictive sums for each crop were chosen as right-hand side constraint on production. Multiplication of these output levels by the distribution percentages results in current production requirements by crop and requirements region. See Table A-l7. Aggregation Rules. Cropping patterns and costs generated by the three L.P. models will be compared at the Region level. The rules to allocate solution values to the Regions are found in Tables A-18 and A-19. CHAPTER V A COMPARISON OF THE INPUTS AND OUTPUTS OF FIVE INFORMATION SYSTEMS The purpose of this chapter is to discuss each of the three inventories in terms of the soil map comparison factors presented in Figure 4-5, and to discuss the differences in solution values of the L.P. models. Inasmuch as it was not possible to undertake three completely independent soil inventories, the discussion of soil inventory methods and costs was based on assumptions of SIEDRA soil scientists for conditions in the Cibao Valley. Comparative Evaluation of Three Soil Inventories The assumed conditions are as follows: The size of the Study Area would be 250,000 hectares; the purpose of the study would be to assist Study Area planners to analyze the area's agricultural potential; the smallest map delineation would be consonant with that purpose; map scales and soil mapping units would be: --1:100,000 and associations of phases of soil families, 1:250,000 and associations of phases of soil sub-groups, and 1:1,000,000 and associations of phases of soil great groups; topographic and climatic maps would exist; aerial photos at l:20,000 and smaller scales would exist; sufficient labora- tory equipment, staff, materials, and budget would exist to finish any one of the inventories in one year. Insofar as the 67 68 latter conditions seemed unrealistic, a conservative bias . . . 1 was given to the costs of the more 1ntens1ve surveys. Preparation Methods The preparation of each inventory would generally require that 10-15% of the allotted work time be spent compiling existing tabular data, imagery, and maps and doing preliminary photographic interpretation to establish map units. One-half to three-quarters of the time would be spent in the field making observations, boundary checks, analysis of soil samples, and map unit revisions. Finally, 10-2596 would be devoted to classification, laboratory analysis, and report writing. For each, map units would be initially established through interpretation of existing photography. A sample area (not necessarily contiguous) representing 10% of the 250,000 ha. Cibao would be established. For the Detailed inventory, aerial photos of l:20,000 scale and a topographic map 1:50,000 scale would be used. Also for the Detailed inventory, 10 observa- tions/km2 in the sample area would be made. A detailed field analysis would be made of three of these observations. Tentative typical profiles would be described. Outside of the sample area, five observations/km2 would be made. A detailed laboratory analysis in a central laboratory would be made of two of the five observations/kmz. Map unit boundaries would be based on the photo interpretations and field and laboratory analysis. Publication scale would be 1:100,000. Map units would be phases of families and associations of phases of families (Jose Ogando, 1979). For the Intermediate inventory, aerial photos of l:60,000 scale and a topographic map of 1:250,000 scale would be utilized to establish map boundaries. 2 Only five observations/km would be made in the sample area. A detailed field analysis would be made of one of these observations/kmz. In the rest of the lA revision of these cost assumptions is suggested in Chapter VI. 69 non-sample Cibao area, 2-3 observations/km2 would be made, with a detailed laboratory analysis of l observation/km2 made in the central laboratory. Extra- polation would be less precise than in the Detailed inventory. Publication scale would be 1:250,000. Map units would be phases of subgroup level. For the Aggregate inventory, aerial photos of 1:80,000-100,000 scale and a topographic map of 1:250,000 scale would be utilized to establish map unit boundaries. Three observations/km2 in the sample area would be made and a detailed field analysis of 0.5/observations/km2 in the sample area. There would be no central laboratory analysis. Outside of the sample area, one observa- tion/km2 would be made. Publication scale would be 1:1,000,000. Soil map units would be phases of great group level. This description of methodology suggests greatest reliability (and cost) would be found at the Detailed level and least, at the Aggregate level. A relative scale proposed by Cline (Cline, 1978) is also used to rank the three inventories on the reliability of preparation methods. Means to collect base data would range from 1 (most reliable) to 5 (least reliable). His scale is defined as follows: 1 - detailed field work with remote sensing support; 2 - detailed field work without remote sensing support; 3 - reconnaissance field work with remote sensing support; 4 - reconnaissance field work without remote sensing support or only remote sensing with field checking; 5 - exploratory field work without remote sensing or only remote sensing; 6 - schematic compilation only. "Detailed", "reconnaissance", and "exploratory" are defined as follows in the Soil Survey Manual (USDA, 1951): Detailed - boundaries are sketched from observations of their entire occurrence on the ground. 70 Reconnaissance - only part of the boundaries are seen. First, a schematic map is made from available evidence on relief, geology, climate, and vegetation. Choose sample areas for detailed work and extrapolate results. Exploratory - boundaries are usually obtained from compilation of existing sources; identify soil associations through limited field checking. Based on this scale, the means to collect data in the Detailed inventory are the most reliable (3), while those of the Aggregate are least reliable (5). While the same type of base map would probably be used in each inventory, the information content of the large scale 1:50,000 topographic map would probably be greater than that of the 1:250,000 map. Therefore, the map used in the Detailed inventory would be more informative than that used in the other inventories. Map Analysis The characteristics of the three soil maps are presented in Table 5-2. They provide the basis for map analysis.l Map Legibility. Map legibility is largely subjective. Factors that influence it include the number of delineations per unit of map area, colors or patterns used to represent them, ground features shown on the map, and quality of map presentation. Only the first factor, the number of delineations per unit of map area, is varied between the three maps, and it is discussed here. 1Before turning to map analysis, mention is made of the regtionship between scale, field observation cm. of map and field observagions/km of land. At a scale of l:100,000, 1 cm. on the map represents 1 km. ; at 1:250,000, l 2 cm.2 represents 6.25 km.2; and at l:l,000,000, 1 cm. represents 100 km- - Thus, although the number of observations/cm. increases as scale decreases, the number of field observations per unit of land area and, therefore, information content decreases. 71 Table 5-1. Comparison of Soil Inventory Preparation Methods Preparation _ Soil Inventory Methods Detailed Intermediate Aggregate Adequacy of base map Type topographic topographic topographic Scale 1:50,000 1:250,000 1:250,000 Photos Scale l:20,000 l:60,000 1:80-100,000 Field observations/km2 5.5 2.7 1.2 Observations y/lab. analy51s/km 2 1 Collection of base data (rank) 3 4 Table 5-2.-- Characteristics of Cibao Valley Soil Maps Map Characteristics Detailed Intermediate Aggregate Scale l:100,000 l:250,000 1:1 ,000,000 Field observations/km.2 5.5 2.7 1.2 Field observations/cm.2 5.5 17.2 120.0 Map delineations Number 95 11 3 Minimum size (cmz) 2.04 2.04 2.04 Minimum size (ha.) 204 1275 20,400 Least size (no.) 3 l 0 Average size (cm.2) 21.9 32.3 7.6 Average size (ha.) 2,190 20,190 76,200 Smallest size (cm.2) 0.3 1.2 3.0 Area covered by 1 cm. length of line (ha.)1 8 50 800 Map units2 Number 31 7 3 Taxonomic level family subgroup great group Phases yes yes yes Agronomic interpretation crop yields crop yields crop yields lBoundary lines are 0.8 mm wide. 2A map unit may be composed of l or more delineations. 72 Legibility decreases greatly if a large number of map symbols must be placed outside of the delineations and keyed to them by arrows. In this regard, a "minimum-size" delineation is one in which a simple map unit symbol (2-3 digits) can be printed legibly. Alternatively, it is the smallest area that can be easily discerned by the user. Its size is, of course, affected by its shape. Detailed map unit 31, for example, is long and narrow and so would have to be larger, for legibility than if it were round. While the concept is a parameter for the determination of a suitable published map scale, it is a highly subjective concept. Sizes of 0.4 cm2 to 1.5 cm2 (1‘. in.2 = 1.61 cmz) have been suggested where 2-3 digit map symbols are used (Eswaran, 1978). Inasmuch as the Intermediate map units would be designated by symbols with 8-15 digits,l a slightly larger "minimum size" delineation, one of 2.04 cm2 (3/4 in. X 3/4 in.), is used here. The actual land area represented is 204 ha. at a scale of 1:100,000, 1275 ha. at l:250,000, and 20,400 ha. at l:1,000,000. On many soil maps, there may be a few delineations smaller than the theoretical minimum called "least-size" delineations. Obviously, a relatively large number of these areas, so small that map symbols would have to be keyed outside of them, would decrease legibility. These "least-size" delineations are particularly significant for large-scale (1:20,000 and larger) maps since the delineations may imply highly contrasting small areas that are not just incidental observations. Given the small scale of the three maps, "least-size" may not be of great significance except possibly for some areas in the Detailed map. On the Detailed I:100,000 map there are only three, and on the Intermediate Map just one "least-size" delineation. lMap unit symbols have not been prepared. However, the symbols used in the CRIES soil mapping for the Dominican Republic were 8-15 digits. This mapping is closely equivalent to the Intermediate soil inventory. 73 The "average-size" delineation for each map is presented in Table 5-2. Their ground representation ranges from 2,200 ha. at 1:100,000 scale, to 20,200 ha. at l:250,000 scale, to 76,200 at l:1,000,000 scale. The relative complexity or simplicity of a map can be indicated by relating the number of delineations per unit of map area to the maximum number allowable while maintaining legibility. If the minimum-size delineation is 2.04 cm.2, then the reciprocal, 0.49, is the maximum number of delineations/cm.2. If n = the number of delineations/cm.2 of map area then n also 1 ‘average-size delineation and map texture intensity (MTI) = 0—2-9 __ l + 1 ’ average-size minimum size . . . 2 m1n1mum-51ze (cm. ) . 2 average-Size (cm. ) The larger the MTI, the more complex the map. The MTI for the maps are: Detailed - 0.093, Intermediate - 0.063, and Aggregate - 0.268. Relative to one another, the Aggregate map is the most complex, the Intermediate map the simplest; all are below the maximum number of delineations/cmz. A map becomes illegible if a large proportion of the delineations are "least- size". Map intensity can also be understood by determining the degree to which map scale can be reduced before the average-size delineations become "least- size", i.e., smaller than the theoretical minimum-size delineations. The m_ap index of maximum reduction is thus defined as the square root of the ratio between the average-size and minimum-size delineations of the actual map. The minimum scale of reduction, then, is the smallest scale to which a map can be reduced before the average-size delineations become smaller than the minimum- size delineation. 74 Minimum scale of reduction 2 actual scale + (actual scale) (index of maximum reduction) The index of maximum reduction and the minimum scale of reduction for the three maps are: Detailed 3.28 and 1:428,000 Intermediate 3.98 and 1:1,245,000 Aggregate 1.93 and 1:1,930,000 The minimum scale of reduction certainly does not indicate a useful scale for, if average-size delineations were reduced to the minimum-size, a large proportion of the delineations would be "least-size". A much smaller scale than that of the published map suggests that the surveyor might have worked with a smaller scale from the standpoint of legibility. A high (e.g., 10-20) index of maximum reduction may indicate relatively homogeneous soil bodies, improper scale definition, or an inadequate survey. A low (e.g., 1-2) index suggests intensive mappings (Eswaran, 1978). Only the Aggregate map approaches the intensive mapping range of the index of maximum reduction. The values for the three Cibao maps suggest that smaller survey scales (and hence lower costs) may have been desirable from a legibility perspective for the Detailed and Intermediate maps. At the same time, the previously noted homogeneity of the soil resource would also seem to support smaller scales. In addition to the above considerations about map complexity, the total number of delineations is important: given a purpose for a study, the fewer the number the easier and less costly the tasks of gathering soils data and comple- mentary socioeconomic data, correcting errors in existing data, analysis and interpretation, and presentation of analytical results to decision-makers. The number of delineations for the three maps are: Detailed - 95; Intermediate -1 1; Aggregate - 3. 75 The adequacy of map scale must also be related to the user's area of planning interest. The largest map scale, then, is defined by the size of the smallest area of interest the user would want to identify on the map. Given the national orientation of intended users, the three Cibao regions could serve the users initially as minimum areas of interest. Broad plans could indicate general use patterns that would make most efficient use of Valley natural resources. Later, the same planners could be expected to focus on subregions within a particular region or regions for more intensive investigation and data collection. From a natural resources aspect only, these subregions would be relatively more homogeneous and could correspond to individual or small groups of map units. From a natural resources standpoint only, individual map delineations would define the minimum area of interest for the Valley's regional planners. In this regard, then, the Aggregate inventory with three map delineations and nine unmapped PPAs may provide the users with sufficient geographic information. The detail of the Detailed (95 delineations and 45 unmapped PPAs) and Intermedi- ate inventory (11 and 14) may not be any more useful than the Aggregate inventory.1 The smallest map delineation in each inventory is: Detailed - 30 ha., Intermediate - 750 ha., and Aggregate - 30,300 ha. Given the minimum-size delineations (defined for legibility at 2.04 cm.2), only the Aggregate map appears to be adequate since only its minimum-size delineation (204 ha., 1,275 ha., and 20,400 ha., respectively) is smaller than its smallest map unit. Reliability of Delineation Boundaries. The intensity of the soil inventory and map scale determines the reliability of delineation boundaries. Theoretically, lOnly limited planning would be possible if only resources were considered. Planners would need to address social and economic factors, the relationships between these factors and with natural resources, and hence the relevant minimum areas of interest. Given the scarcity of reliable socio-economic data the additional usefulness of an inventory as disaggregated as the Detailed is particularly questionable. 76 the larger the scale, the larger is the number of observations required for boundary placement and the more accurate is the placement. Given the discussion on inventory methods, the Detailed inventory ought to have the most reliable boundary placement, the Aggregate the least reliable. The accuracy of boundaries may also be a function of line width: the smaller the size of individual delineations, the greater the effect of width. Should a boundary line be misplaced, the percentage error in the area of the delineation is greater the smaller the delineations. Ten measurements were made of the width of the delineation lines on the 1:100,000 map and averaged. The land area covered by a 1.0 cm. length of this thickness (0.8 mm.) of line would be 8 ha. at l:100,000, 50 ha. at 1:250,000, and 800 ha. at 1:1,000,000. Assuming circular average size map unit areas, the areas of their boundary lines would be 6, 5, and 10 percent, of the circular areas in the Detailed, Intermediate, and Aggregate inventories. Thus, effect of line width on Intermediate delineations is relatively smaller, 5 percent, since average size (32.3 cm.2) is greater than in the other inventories. With respect to the Detailed inventory (6 percent error), however, this Intermediate advantage may be cancelled out by the less reliable means used to collect Intermediate data and hence locate boundary lines. Given the less reliable means used in the Aggregate inventory to collect base data as well as smaller average delineation size (7.6 cm.2), errors in placement, 10 percent, are more sizable than in the other inventories. The area of a boundary line of a minimum size area on the Detailed, Intermediate or Aggregate inventory maps represents 20 percent of its area. Comparison of Measured and Summed Detailed Areas of Map Delineations. The area of each map delineation, at all three levels of detail, was planimetered as drawn on the 1:50,000 scale FAO base maps. Theoretically, area measurements would sum exactly across the three levels of soil detail. That is, Aggregate map 77 delineations (associations of phases of great groups) would be perfect sums of the Intermediate map delineations (associations of phases of subgroups) which in turn would be perfect sums of the Detailed map delineations (associations of phases of families). For several reasons this boundary coincidence does not occur. Cartographic error may occur when delineation boundaries from maps of three different scales (l:1,000,000; 1:250,000; and l:100,000) are drawn on the same 1:50,000 base map. The lines from a small l:1,000,000 scale map do not exhibit as much precision as the lines from a l:100,000 map. Conversely, if a l:100,000 map is reduced in scale to 1:1,000,000, its delineation boundaries lose their twists and turns and become smoother. Intermediate delineation A in the southeastern part (Fantino) of the Cibao illustrates such cartographic errors. Intermediate map unit A is defined to include only Detailed map units 1, 2, 3, 4, 5, 8 and 10 (see Table A-18); yet it has been drawn to include minor parts of 31 (and one major part of 31). See Figure 5- 1. Intermediate G defined to include only Detailed 31 has been drawn to include parts of 6 (of Intermediate B), 10 (of Intermediate A), and 2 (also of A). Aggregate map unit I defined to include Intermediates A and G, see Table A-6, has been drawn here to include parts of G (also of Aggregate II) and of B (only of Aggregate 11). Aggregate 11 defined to include Intermediate B, C, F and G here also includes part of A (of Aggregate 1). Another aspect of cartographic error relates to the minimum size needed for legibility on the published map. Map delineation may retain inclusions of non- similar soils simply because to split the delineation unit into two or more parts and maintain purity would mean illegible map units when published. For example, south of Castanuelas in the western Cibao, Intermediate delineation D has been drawn to include Detailed 27 even though 27 properly belongs to Intermediate E. See Figure 5-2. 7636 15m Is'oo' DOMINICAN REPUBLIC FANTINO SOILS UNMAPPED SOILS MAP 7dw’ rwa m 7d3d Figure 5—1 ~— Southeastern Cibao Valley map boundary 1 associations of families : associations of subgroups ossocmtions of great groups unmapped area approxlmate scale 2 3 4 5 Km l I 1 fl l-’ 100,000 81 I9 10 I9' 50‘ 71°45 REF’JBLIC DOMINICAN PEPILLO SALCEDO SOILS MAP 9 I I 5875 "II: 7 30 L 7K (f 23 UNMAPPED SOILS 27 28 I 25 E ATLANTIC OCEAN 27 111 M0 /\ Castanuelas "AQUE DELI\. - NokrE K \ 27 22 Pepillo Salcedo \ x 22 20 27 I \ 19° 46 . \ 71°45 __ map boundary 7| 30 —— ossocnations of families —— —- associations of subgroups Figure 5—2 —— Western Cibao Valley “““ “55°C‘°"°"5 °' 9'9“ 9'01"” [227723 unmapped area approximate scale 2 3 1 i:ioo,ooo 19'40' 80 Other possible sources of error arise from neglecting to include small areas of Detailed delineations which are on the borders of the fourteen 1:50,000 base map sheets in an Intermediate or Aggregate delineation, improperly locating soil boundaries, mislabeling, using line markers of varying widths, and inexact use of the planimeter. The planimeter, for example, approximates curves with linear segments. Area measurement of those map delineations with a higher proportion of curved boundaries (Detailed would generally have more than Intermediate which would have more than Aggregate delineations) would therefore be subject to greater error. Tables A-20 and A-Zl present the cartographic relationship between the Detailed, Intermediate and Aggregate units and the measured areas of each map unit. The area measurements at the three levels are 2179.53 kmz, 2221.32 km2, and 2286.34 kmz, respectively. The Intermediate measured area is 1.9 percent greater and the Aggregate measured area is 4.7 percent greater than the sums of the measured Detailed map units. Measurement error appears when comparisons are made between the plani- metered Intermediate (Aggregate) map delineations and the sums of those Detailed delineations of which the Intermediate (Aggregate) delineations are composed. At the Intermediate level, the difference between measured and summed areas is least for the larger map units (5.3% for B, 0.3% for E, and 4.5% for C) and largest for the smaller map units (17.6% for D, 22.5% for F, and 72.9% for G). A, a relatively large unit, has a relatively high difference of 15.4%. This pattern seems explainable largely by noting that small areas of dissimilar soils often would be included within a map unit for legibility purposes. G, for example, is typically a long, narrow shape immediately paralleling rivers. Measurement of map delineations is another minor source of error. An indication of the relative size of this error was made by remeasuring every 10"h 81 Detailed delineation (16 of 153), based on the original order of measuring on the individual map sheets, every 5th Intermediate delineation (7 of 36), and every 4th Aggregate delineation (7 of 22). Table A-22 compares the initial and follow-up measurements of the chosen map delineations. The average absolute difference between the first and second measurements was 0.8% among Detailed delinea- tions, 0.9% among Intermediate delineations, and 1.0% among Aggregate delinea- tions. Even the errors of individual areas never exceeded 4.0%, less than the percentage the boundary line represents of the average size map unit even in the Intermediate inventory. No defensible conclusion regarding the relative error encountered in measuring the three sets of map units is drawn. Detailed lines with their twists seem more subject to mechanical planimeter error yet those twists which have soils informational significance are smoothed out in the more generalized, smaller scale maps. Map Unit Descriptions The soil taxonomic classification of the map units in the three inventories are family, subgroup, and great group. Because of the restricted extent of the Study Area, map units almost never cover the whole range of the class used to name it. The total range in soil characteristics for Cibao Valley great groups and subgroups is quite limited. Thus, great group descriptions are "more or less" (Knox, July 1979) done at a family level of significance. Soil phases, for example, are the same as those at the other levels. Similarly, the dominant textured characteristics of the subgroups were able to be specified and so descriptions could be made "more or less" at the family level. The family-level map is most closely based on the FAO "series-level" map. With some assumptions about clay mineralogy, the "series" were placed into families based on the information given in the report on particle size distribution. There is considerably more doubt about 82 placement at the higher subgroup and great group levels than at the family level. Although the FAO information leaves questions about the degree of flood hazard, the nature of soils on the flood plains, and the drainage restrictions, the family map could present "reasonable analysis for interpretation of management require- ments and use potentials" (Knox, July 1979). At each of the three levels, both consociations and associations of soils are utilized. By definition a consociation is purer than an association. It is a mapping unit in which only one identified soil component (plus allowable inclusions) occurs. An association is a geographic mixture of areas of two or more distinctive kinds of soil, or of a soil and a kind of miscellaneous area. Principal components of soil associations are generally not delineated at scales smaller than 1:20,000. In the Detailed inventory, l9 map units are consociations and 12 are associations. The Intermediate and Aggregate inventories have 3 and 0 consociations and 4 and 3 associations, respectively. This suggests a highest level of purity, given the taxonomic level, for the Detailed inventory and least for the Aggregate inventory. Soil Inventory Costs The costs of gathering and developing soils information depend on such a variety of factors that to date, no neat and tidy formula for their estimation has been proposed. These factors include climate; transportation difficulties; health problems of field personnel; levels of wages; imposition of completion deadlines; availability of secondary materials such as topographic maps, planimetric maps, geologic maps, meterological data and maps, recent aerial photography, photomosaics, satellite images, soil surveys at larger or smaller scale; type of equipment for field and laboratory work; and availability of trained personnel; number of cloud-free days; severity of rainy seasons; and complexity of the terrain. Most developing countries require that all work be done through their national soils agency; although the impact of this requirement on costs will vary 83 by country, it may cause at least a doubling of costs over that demanded by foreign consultants (Bleisch, 1979). A brief review of cost estimates suggests their great variability. The U.S.D.A. estimates the relative costs of detailed soils surveys for U.S. counties of 700 square miles of survey area as 1.0 for low intensity, 1.4 for medium, and 2.0 for high intensity land use (Herfindahl, 1969). In a low intensity area, 75 percent would be in extensive agriculture, such as range. In a medium intensity area, 50 percent would be in cropland with the remainder in range and forest. A high intensity area would have a mixture such as 33 percent cropland, 15 percent range and forest, 25 percent urban, and 25 percent other. Field work accounts for 80 percent or more of total costs in each type. In 1964, the OAS estimated costs of various soil surveys in the Guayas River Basin in Ecuador (OAS, 1964). Direct costs (those associated with a particular type of survey activity) and overhead costs (such as facilities, topographic map work, mosaics, and drilling) were of approximately equal importance. The direct costs of different survey activities were as follows: Survey Activity Cost/ Acre Relative Cost Reconnaissance soils capability 0.015 1.0 Semi-detailed soils capability 0.113 7.5 Detailed soils capability 0.358 23.9 Finally, Eswaran presents the following relative costs for three classes of surveys as defined by the Soil Conservation Service (Eswaran, 1978). Relative Class of Survey Objectives Map Scale Cost Order V national planning 1:1,000,000 0.2/ha. Order IV regional planning 1:500,000 l/ha. Order 111 cultural systems l:100,000 2/ha. It is obvious that these four sets of estimates are not comparable. The lack of comparability may stem from differences in definitions as well as from the factors noted above. 84 The procedure to obtain cost estimates for the three Cibao inventories was to specify conditions relevant to soils inventory in the D.R. A SIEDRA soils scientist was asked to estimate the costs as if the three inventories were independently constructed. The assumptions about conditions, materials, facili- ties, and labor in the D.R. and in the Cibao Valley were stated above. Table 5-3 summarizes the costs of each inventory. Appendix A-23 presents the cost detail. Table 5-3.-- Soil Inventory Costs Item Detailed Intermediate Aggregate ($1000) Personnel 271 . 2 90 . 0 21. 2 Materials 4.4 I. 3 0. 5 Vehicles 70 . 0 21 . 0 4.2 Lab. Analyses 150.0 33.8 20.3 Publications 74 . 3 29 . 2 9_._2_ Total Cost 569.9 175.3 55.4 Cost/Ha. ($/ha.) 2.28 0.70 0.22 The relative costs for the three inventories are 1.0 for the Aggregate, 3.2 for the Intermediate, and 10.3 for the Detailed inventory. The purpose of this section has not been to draw precise conclusions regarding the information content and costs of the three soil inventories. Rather, it was to present a method which planners could use to make such an evaluation in their specific situation. In general though, it appears that much more soils information is generated in a "detailed" than in an "intermediate" or "aggregate" inventory. Not surprisingly, it is more costly. The issue is to present intended users with a comparative statement of the information produced by each. It is their responsibility to choose. Attention was paid to only one presentation tool- soil maps. Several measures of information content were discussed. 85 Comparative Evaluation of Linear Programming Solutions The DetailecL Intermediate Weighted, and Aggregate Weighted Models The three models - Detailed, Intermediate Weighted, and Aggregate Weighted - were run for three different policy questions. The questions are: 1. Given the Current Normal pattern of cropland use and Regional crop requirements, what would be the least costly means of meeting these requirements? This is called the "Benchmark" run. 2. Given the Current Normal pattern of cropland use, what would be the least costly means of meeting Regional crop requirements that are 20 percent greater than those of the Current Normal? This is called the "Growth" run. 3. Given the Current Normal pattern of cropland use and Regional crop requirements at 120 percent of Current Normal, what would be the least costly means of meeting requirements if, due to shifts in population away from rural Regions to Santo Domingo, the require- ments of this Region for non-expost crop are increased relative to the other Regions? This is called the "Growth/Shift" run. Except for corn and for idle area in the Detailed model, solution values of the three models, do not vary between the Growth and Growth/Shift runs. See Table 5-4. However, transportation flows between Regions are different in 28 of 90 (3 models x 3 Regions x 10 crops) possibilities. See Tables A-26 and A-27. This indicates the insignificance of transportation costs relative to production costs in the Objective Function. Total production cost of the Detailed Growth/Shift solution is only 0.6 percent greater than the Growth solution in the East, 0.1 percent less in the West, and less than 0.01 percent less for the entire Valley. The Objective Function value is 0.3 percent greater in the Growth/Shift run. 86 6.5328 oumwocww< ..o .voEwE? oumwocww< 63:30? oumBoEcouc— 05 E .5000 305.02% oZm N6 Nd m4 To «6.. .. 1 m6- 22 m . T .. 1 w . T 1 1 1 1 63:29.3:— 1 1 1 1 1 1 .. 1 8000 ~.o 521 061 w.m 1 1 1 1 mooamuom $026 w . o n .n o .L 1 1 1 .. 1 88> Yo 1 m4: :6. 1 1 1 .. mcmom «.0 .. m4: wé- :6 Tan 1 o.T EoU .. 1 .. 1 1 1 .. 1 250 ..mem 1 1 .. 1 1 1 1 1 9853 m . o 1 a . N 1 1 1 1 1 5355 o . o 1 _ . o 1 .. 1 .. 1 00030... 1 1 1 1 1 1 1 1 avg mo=m> “mm? .9550 “mom xozm> to? 3.550 “mom no.0 omeoEcoH-E Wolzimaloni— Amcoflgom 5380 we m8 . mmczm SEQ-$380 vcm 530.5 55:28 68.33? oumwocwwt. paw five-Eu? pumBoEcoE— .8238 5003qu moc< aocU E moo-5.635 I .aun 2am... 87 Due to the trivial differences between Growth and Growth/Shift Solutions, comparisons between Detailed, Intermediate Weighted, and Aggregate Weighted solutions in the Benchmark and Growth runs are now made. Although not always true, the difference in solution values between the Intermediate Weighted and Aggregate Weighted models are less than the difference from the Detailed solution. The differences between the cropping pattern of the Detailed model and those of the two aggregated models are presented in Tables 5-5, 5-6, and 5-7. Each table is progressively more in summary form. For both aggregated models, the difference is much less at the Valley than at the Regional level. See Table 5- 6. Also, for both models, the total (unsigned) difference at the Valley and Regional levels is less for the Growth than for the Benchmark run. Regional differences from the Detailed solutions are much larger for the Aggregate Weighted than for the Intermediate Weighted model. For example, in the Benchmark and Growth runs, respectively, the sum of unsigned differences for 11 crop and idle areas is 516 percent and 643 percent, respectively, for the Intermediate Weighted model and 1506 percent and 2033 percent, respectively, for the Aggregate Weighted model. For the two runs, the Intermediate Weighted model shows a smaller amount of Regional variation around the Detailed solution in 4 of 6 instances. See Table 5-6. The areas of three crops that were relatively small in the Current Normal were not allowed in the linear programming algorithm. The crops and their areas are banana 4005 ha., sugarcane - 8237 ha., and cocoa - 9313 ha. Difference in their areas from the Detailed solution occurs outside of the models. Aggregating Detailed map units did lead to some noticeable differences at the Regional level for each of the three crops. :-~1h \‘HL-YI:~F\1\F-fifif~ lv-hhvmhiufifilvr. 33:34.13}! ..UHF.§LLIX< :5: Josue—3.203 1Hs-4__.urv...1.hsup,._ Jere—...,..uavf. «up: Zarvxsbmvc 233:5-.1shuac «WOU WC: .CC..~1,M:~1v0LL ..flde.‘ ll 1h1|h 1U~n~nw1~1 88 90:50:80 n4 “.0 n; “ OR 23 Em " N.» N.” N.» 53> 0.0 0.0 0.0 u 000 000 000 u 0.0 0.0 0.0 $03 2.0 2.0 2.0 u 02 02 02 u 20 20 20 35:00 * * * u * * * u * * * ummm " ” 0:00 ..mwam 0.~ 0.~ 0d u Sm; Sm; 8m; " 0.0 0.0 o0. >23 0.~ 0.N 0.~ u 0 0.0 own 0 02.. 0 u 0.0 0.0 0.0 $03 * * * u * * * u * * t 3.3ch 0 0 0 u 0 0 0 u 0 0 0 $0M " n 20:00 0.2 0.02 0.2 u 000 000 000 u ”AN 0.n~ 0._N >0=m> 0.0 0.0 NK ” 02 03 02 u 0.02 0.02 2.2 503 2— 0._ 2; u 00 00 00 u TN 0.~ 0.2 35:00 0.0 0.0 0.0 u 03 02 02 u 0.0 0.0 0K 501". u n 53:20 20 2.0 0.0 u 0R0 0R0 0R0 " 0.2 22 0.2 >0=m> 0.2 0.2 0.2 u 00 cm 00 u w.~ 0.0 0.0 $03 0.» 0K 20 u 000 000 000 u 2.2 22 22 11.3000 0.0 0.0 0.0 u 00 00 0N u 0.0 0.0 0.0 “mam " " ooomnoh 2mm 0.3 2.3 u 00R.N 00R.~ 00R.~ u 23 10m 0.0m >0=0> 0.2 0.2 0.2 ” 00w: 03$ 00m: " 0.2 2.2 12 «m0? 0.2 0.2 0._ u 0: 002 002 u 0.2 0.2 22 35:00 Ni 0.2 0.2 u can; on} 2m; " 1: Z: «.2 88m. u u 00E @800 5:20.: ” 03:5 000: ” $0.300; 0005 SEE»? 08.30? n .8206? 82003 ” BEE; 82303 0~mw0cww< 002006.005 00:08.00 " 0umm0cww< 000605.005 00:08.00 " 0umw0cww< 05:00:00“:— 002300 co2w0~0=0> 0N ~00? . 0 .3200 00 wmmm 0:: n : 0:3 0 0 083 05 _2200 n a «80 0.5345 00.280 .2 033 91 “00:53:80 n .00 0 . :2 0:3 200 0.00 20— ”.»: :41: 82298: 0.002 0.2 A0.030000 00030 0 . 00w 0 . 00 000.; 0.00: 0.00 200.0 0.3— 0.00 :80 0 .00 200 53:05 200 0. 2 800:8. 0.02 2.0 00E 352 m.“ R 2200 0005.035 00:30:00 0.0 0.00 0.~2 0.0 0.01 0.01 0.01 0.0 00o0oU 0.0 201 0.0 0.02 0.0 0.0 0.0 0.0 00:00 03.0 201 0.01 0.001 0.0 201 0.01 0.001 0.0 00:0:00. 0.21 20 0.02 0.001 0.22 0.021 0.02 0.02 0.0— 200 2.000 2201 0.0 0.00 2.00 0.N~ 0.00 03:29.3:— 201 2.001 0.02 0.2 0.01 0.2 201 22 003390 00030 0.0 2.000 0.02 20 20 0.00 0.01 0.2 000.; 2.2 0.002 0.001 0.2 221 0.0 20:1 2.021 0:000 0.2 0.001 0.001 20 0.01 2.02 0.21 0.01 :80 0.0 2.21 0.2 0.0: 0.02 0.00 0.0m 0.21 E02020 2.0 202 0.2 202 2: 0.2 20 0.0 800:8. 0.2 0.01 2.0: 20 _._ 0.0 2.0 0.N 0000 >0:0> 50? 03:00 30.0 U:0> 50? 03:00 «0.01"— 09.0 03:30? 030M003? 03:30? 30:03:03:— cam x..0E :oc0m 05028 8:00: Ho 80 230 5320 0:0 x..0E:o:0m 5:22:00 03:30? 30w0hww< 0:0 63:30? 32005.3:— .000300 0:» :00>$0m 000:0:305 00.3 11 .010 030... .m..m..._ 0:0 00 002330 .5000 000:00300 00:0 2:505:58 .04 0:0 35 ..3:0 30: 020 000.0 0005.0 92 m .0” N3 0:3 2.02 200 0202 m . 000 N . 00 03:00:35 N . 00_ 0 . no: 003300 00030 0 .NON 0.00 000:> 0.00: 200 0:00m 0.3. 200 :00 0 . :0 2 . Nu 53:05 0 . 20 0 . K 0000000. 02m 2.0: 0000 0.38 0.90 2200 000:0..305 00:30:: 0.0 0.00m 002 20 0.0 0.0 0.01 0.0 000000 o.o ~51 m.» ”.3- ca 0.0 o.o 9: 0.050 .008 201 0.0 0.001 0.0 201 0.0 0.001 0.0 00000 2.001 2.01 200 0.001 0.~01 0.001 0.0 0.001 0.0— 0.N 0.000 0.001 0.0 0.0 0.00 2.01 0.0 03:00:35 0.0 0.02 0.01 0.02 0.0 0.001 0.01 0.02 0030qu 00030 0.0 0.00m 0.02 0.: 0.0 _.00 0.01 0.0 000.; 0.2 0.002 2001 0.00 0.2 0.00 2.21 0.0 0:000. 0.2 2.00 0.001 0.: 0.01 0.001 22 0.~ :00 0.0 0.0 0.0 0.02 0.0m 0.00 0.0 202 53:07. 0.0: 0.21 0.2 0.0N 0.2 0.21 20 0.00 0000000. 0.0 201 0.2 20 20 20 0.01 0.0 008 00:0> «m0? 03:00 000m 00=0> «00? 03:00 0.00.1... 0000 03:30? 30000030. :sm 530:0 03:30? 302005035 0.280 .1 0-“ .033 93 .808 80.00 :05 000— 03000 0:0 .033: n 320: 0_0>0_ :0300003 .03300: on :05 3:050 000:0 000:3 00288 0030.. .00:0_:0>:00 00n— 3.305003 0300505 0:3 :03 020> :03200 000.300 0:3 0:320 :030:0> 300:050 05 030:0 :03200 _0005 623 030:0 w55t00cb0 03 S— me. o: u 0o: 00 .3— N: 03000 .003 _30._. 2703 09-3 3 Tnm 372 n 3700 wmnunm a: 7: ST? 03000 30:0 u 0.3? 30w0hww< 03 N: mm 0o: “ n3 2: 3— mo: 3000 .003 .303. 3703 «Sun: 03-3 373 " 370» 37; wawn $5.6m 03000 30:0 n 0.3? 305005035 oo 73 80-3 : 5 70m on Tun u co TR com- 3 momu00 am Thu 3302003 n_0._0 " 0.3? 30w3ww< co 7mm semi: 52.0“ 0 07¢ u 0o Twm oomnww 3 TS 370 :0303003 30:0 n 0.3? 305005035 mm 3 n2 mm " Nw we: a 2 G 000.0 0.0— 373 09.3 3 Tom 02-: " onTNm nonnwn m: Tn0 N07: 000.0 30:0 u 0.3? 30w0..ww< n0 mm 00: no 0 mu 3 mm R 000:0 0.0— 0.2-3 3 7m: 0o Tom 8 Tan 0 0n To» an TR nmfiuon N0 7mm 000:0 30:0 u 0.»? 305005035 $0 " $3 N0=0> 300? _03:00 300m " >0=0> 300? _03c00 300m ..30503mb0002 cam 530:0 053 0:06 :0:0@ 050% 530.0 0:0 x..0E:0:0m 05328 002300 30 00000300000 00 0:0350m 03:30? 30w00ww< 0:0 03550? 305005.35 I .50 030... 94 General relationships for area differences of individual crops are somewhat difficult to discern. In the Intermediate Weighted model, the unsigned sums of differences for specific crops show that rice, tobacco, and sweet potatoes are the closest to Detailed values in the Benchmark run. In the Growth run, this can be said only about rice. In contrast, the further degree of aggregation in the Aggregate Weighted model seems to have more clearly established area differ- ence relationships. In both runs rice, tobacco, and plantain are in one group - that with the least difference. Area idle in the East and West, and at the Valley level is less in both aggregated models than in the Detailed solution. This may be due to greater flexibility and efficiency of land use in the Detailed model. Difference from the Detailed solution seems to be positively correlated with the coefficient of Detailed yield variation, see Table 5-8, particularly in the Aggregate Weighted model. The lowest coefficients of variation are for plantain, rice, and tobacco, the crops with the least difference from the Detailed solution. Unexplainable is crop area interplanted. Whereas this activity has the lowest coefficient, it shows the greatest amounts of difference. Absolute solution values for Regional production of individual crops also show considerable variation from the Detailed values. See Table 5-7. For both aggregated models, the Valley solutions are nearly the same as those of the Detailed model. The relatively smaller variation in production (versus that of area) is consistent with the L.P. structure. The L.P. attempts to minimize costs; reducing production down to the right-hand side crop requirements constraint, a greater than or equal to constraint, helps to do this. Conversely, the relatively greater variation in area than in production stems from each model's ability to choose to use from 70 to 120 percent, in the Benchmark run and 70 to 130 percent in the Growth and Growth/Shift run, of the 95 .380 0:03 30 30:3 :05 000: 003 00:0 0:0: :03 :0:30::0> 32000< 0 .:00:: 0:: >0 000:>:0 :0:30:>00 0:00:30 u :0:30::0> 30 3:0:033000 0 .300? 0:3 5 5:353 0.333 n: :0 0:00 0: 305300 0:0; 0 303035330 .3:— 003 :3:0 30: 0:0 030:0 000:3 0050 00000 :0 .0:00 330 .0:0:00 00205 30: 00000 003303 03:03:35 30030 303003 03:03:35 03:03:35 303003 0:0 .: 00:> 00 x 0 : 3 0 _ 00?: 50 :0 3 . 0:00 : 0030::35 A. 0.3: > 03o 0 m 00:000 53:03 0:000 0:000 :36 0::00 ::00 0000003 ::00 00?: 00w . 3 0003303 0 3003m 0003303 0003303 30030 00:» 30030 ::00 Rn . o 000000 .3. 0003303 0000003 ::00 00000003 30030 : :n . 0 00:0 53:03 0:000 53:03 0000003 m : 0 . o 53:53 00:: 00:: 3000: 00:: 00:: 3000: 2nd 03:03:35 03:w:0? 033:0? 00:0:300 03:w:0? 03540? 00:0:300 0:0:30::0> 30w0:ww< 30:00E:35 30w0:ww< 30:85:35 0:0; 00::300 030:0 :3: 5.30:0 :3: {0:50:00 30 3:0:0:3000 0:0:320w 00:< 30:0 000300 50:": 0030:0300 00:w:0:3 :0320m 000.300 503 033300 00:< 30:0 :0:0:w0m 00:w:0:D 0:0 5:30::0> 00:0; 30 3:0:033000 :003300 5:30:00 l .wum 0:003. 96 Current Normal area in each crop-production method combination, by PPA. In addition, in both runs the models can use up to 50 percent of the area of any PPA not in use in the Current Normal. The larger number of Intermediate PPAs, 14, than Aggregate PPAs, 9, implies more flexibility in the Intermediate Weighted model (most in the Detailed which has 45 PPAs) and hence relatively greater area variation with respect to the Detailed solution. Regional values of both models for production cost of individual crops show considerable range, similar to that for area, about the Detailed values. Total production costs for all crops are greater than the Detailed solution in all areas except in the West where the Aggregate Weighted solution is 5.14 percent less. Regional production costs are never more than 12.9 percent greater (3.8 million pesos in the East Region - Aggregate Weighted solution) and Valley production costs never more than 5.14 percent, 3.8 million pesos, greater than the Detailed solution. Differences from the Regional Detailed solutions are usually but not always greater in the Growth run than in the Benchmark. The largest differences are in the Aggregate Weighted solution for the West, 18.6 percent greater, and for the Valley, 8.9 percent greater. Neither of the two aggregated models consistently show a greater (or lesser) difference from the Detailed solution production costs. Regional crop requirements can be met either from production within the Region or by importing from another Region. In both models, crop transportation flows from one Region to another were nearly always (29 of 30 times) different from those of the Detailed model. Total flows, by crops, of course, were usually the same. See Table 5-9 and 5-10 for an example with rice. Transportation costs are insignificant relative to production costs. For example, when rice produced under the low production method is transported from East to Santo Domingo, 97 .030005 :003300 033:0 :0:w0: 03 :0:w0: 50:3 030: 003 3003 03000 :3 x:0:033< 03 00:0:030m .3000 0:00 30 3:003: 03 03:08:03 0:0 530: 0 503:3 0003003 350050 0003050 0.0005 0.30.33. 0.02.30 02.0 02.0 05.0 0.02.0 0.02.0 0.02.0 :33 o o o 000: 000: 000: o o o 00558 03:0m 3.02.0: 0.02.2 0.03.2 EN EN EN 0.32.: 0.303.: 0.003.: 383 0.30.: 0.3.} 3.8.0.: 20 20 20 1:: 0.8: 100 3:350 702.: 0.08.2 0.3.0.2 9:0 20 0% 0.80.: 0.000.: 3.20; 30mm 388: 80: :00 08: :00 80: 830063 82063 002303 82063 002063 824.003 030m0:ww< 030:005:035 00::0300 030w0:ww< 030:005:035 0020300 030w0:wm< 030:005:035 0020300 0:0:mmm 03000 5300023 0030—03 :0:303:030:0: 3. 330393 0305050003 .0303. :33 5:056:00 .0:0:3=_0m 0030.333 030w0:ww< 0:0 003—3:03 030:005:035 0000300 003 :003300 000:0:0335 3000 5300093 0:0 .035". :0:303:030:0:3. .:0:30:00:3 00:13 l .06 0:00 3. 0.80.0 0.02.: :.::: 0:00.: 0.80.0 0.000.: 0.0:: 0.000.: 0.80.0 0.000.: :00 0.20.: :33 0% 0.000.: :.000 u 0.:n0 0.000.: n.:m0 u n.000 0.000.: .. u 0.000%: 005500 03:00 0.000 0.000 .. . 0.000 0.000 . . 0.000 0.000 . .. 300B 0.0:0 0.000 :.::: . 0.0:0 0.000 0.8: .. 0.0:0 0.0:0 . n 33:00 0.000 . .. 0.0% 0.0% u .. 0.0% 0.00:0 n.000 :00 0.000 300m :00 80:: :00 80:: :00 80:: :030 3. 3003 33:00 300m :030 3. 3003 33:00 300m: :030 3. 300? 33:00 3000: 0:509: 0:503: .0333 05:00,»: .0333 0:503: 03300 0505050300”: :08: 0030003 800200,: :08: 820:0; 80:85:85 :08: 0000300 :3: 30050050.. 0:530:00 0300:0033 03000:00< 0:0 0300:03 3000535: 00:33 :003300 08:30:05 30:": :0:303:030:0:3. 00:0: .1. .0:un 0:003. 99 transportation accounts for 5.6 percent of total costs; and from East to Central only 0.7 percent. Objective function values for the three models and runs are shown in Table 5-11. In each instance, both aggregated models (the Aggregate somewhat less than the Intermediate) have a higher optimal solution than the Detailed model. The differences range from $2.9 to $5.4 million which is 3.9 to 6.4 percent greater than the Detailed solution. Another method of comparison is done with the Theil Inequality Coefficient, U, discussed above. The coefficient compares L.P. changes predicted from the Current Normal level of some parameter. The U coefficients in Table 5-12 compare projected changes of each of the two models to the projected changes of the Detailed model for Regional and Valley cultivated area and for production. Since a Current Normal cost was not calculated, U is not available for costs. When there is perfect correspondence between the Detailed change and that of another L.P., U has a value of zero. Non-negative, it has no upper limit.1 Coefficents are the same in the Growth and Growth/Shift runs for production and either the same or only slightly greater in the Growth/Shift run for area. A markedly higher degree of correspondence with the change projected by the Detailed L.P. is noted for production than for cultivated area. Although even in production, in the East and Central Regions, the correspondence is not very close (with U ranges from 0.34 to 1.00). Correspondence in production with the Detailed solution is closer for the Valley than for the Regions. Coefficients for area do not indicate close correspondence to the Detailed solution in any cases, Region or Valley, nor do Regional area coefficients seem to be consistently larger or smaller than that of the Valley. Each of these points are consistent with the 1Suppose a change in cultivated area of all crops from the Current Normal as solved in the Detailed and some model A is to be computed. If A predicts a change from 7 in the Current Normal to 14 and the Detailed predicts the same, 7 to I4,thenU=0;ifAis7to4,thenU = 1.4;if Ais7to 21, thenU = l;if A157 to 17, then U = 0.43; and so on. :.w- :5- 3.8- 8:38 - .ww< a}- :38. :5- 8:38 - 35 2.“ :3 8a 8:38 - .83 .ww< 2.: «ma 2.“ 8:38 - 6.5 .3: :2?qu no 8 mm 888035 33 8:. on“- 8m- 82mm; 35 .. 6.5 .ww< ea . 7 OR. _- SN. 7 3385.35 - .ww< egg- 3:. 93;. 8:38 - .mw< 8a.? 23...? 8%? 8:38 - 3.: m 08:. :8; S» .N 8:38 - .83 .ww< 1 93K 93K coma 8:38 - 6.5 .35 mcomugom comuuczn— Cm mvucvuvwwmo 885 08.3 ORR Emé 383:? 95.3 833 03.2 3385.35 818 03.3 8:6: 82:83 382%,: 03.8 238 ORR 8233: 3385.35 93 . 2 E _ .2 2m . R 8:38 mcofliom 5385 035030 Aooo_mV EDN— CDM EDN— ~08: w:£m\£w30h0 LHBOLU meESUcvm 228.2 =< Ho 35328 cofiucsm 8:830 5233 305835 l ._ Tn 03m.— 22.50 05 E8 owcmco 05 m. .m .5528 6025.60 of 3 88> 88.52 .5528 6.5» .ww< ..o 6.5» .uE 2t 9. 88> _mEcoZ .< 883: n o Kn : 222380 22:. 05 ..3 2253 8.: 22.50 of E0: : couoEmcma 2:9. 5 owcmzu 2t 3 _ 101 N AH OIfiNO v\ :36 B— 6 $66 omm6 «86 SN; men A awn 6 6.33 oumwocww< cam $235380 95.: B: .o 93.: :86 86.: SN; New; 3%: 6.5 383%.: and 539.0 :66 I: 6 in 6 3:6 Sn 6 amm 4 Rm ._ $66 6.5, oumwocwwxx cam {3:588 6.6.3 8662:52— 8. 69.8800 acorn—om 266 31: 95.6 80.. So; :3 9: $6.: 6.5 383»? :36 626 3:6 :36 86.: Kim :26 in6 6...? 9.3.6253:— cam “HEQEBSU 266 n26 Q36 86; moo; Rn; Q2.— mmw6 6.5» oumwocwwx‘ wmo6 on: 6 £66 366 «8.: $36 :1 ._ mnm6 6...? 83606.6»:— EE 5320 .266 mm_6 _NR6 636 6R6 :34 Rm; 366 6.33 oumwocww< $66 .86 ~36 Rm6 .34 :36 3:6 3:4 .PpB 08:86.63:— csm xcmEzocom 63:22:80 3 “6358580 3 >o=m> “mm? .3550 3mm .32.; 9.33 .3550 5mm cots—om 8:300 op 088800 mcoflgom cofiosvem mcofiaom 853.00 new 6358? oumwocww< .UoEwEB 8.2.86.8:— cooapom moucovcoamotou n.. .NTm 2an 102 L.P. structure discussed above. Considering Regional area coefficients only, correspondence with the Detailed L.P. change from the Current Normal is greatest in the East except for the Benchmark run where the Central Region in the Intermediate Weighted model has the best correspondence.l Area correspondence between changes projected by the Intermediate Weighted and Aggregate Weighted L.P.‘s from the Current Normal are generally closer, as indicated by lower U values, than comparisons with the Detailed model. The least correspondence occurs at the Regional levels and appears to be quite low. Bias, or average signed error, of L.P. solutions is a means of showing if the estimates of the more aggregated models are either larger or smaller than those of the Detailed on the average. If the weighted sum of the overestimates equals the weighted sum of the underestimates, the bias is zero. Percentage bias is presented in Table 5-13. More often than not both aggregated models showed a positive bias to Detailed solution values for cropland area. A negative bias of 1.9 to 6.5 percent appeared, however, in the Central Region. With respect to production, the aggregated models were as often positively as negatively biased. Except for the Central Region where a strong negative bias was always exhibited, 1This performance is explainable if one considers the relative (to the total number of Detailed PPAs) number of PPAs that entered into solution in the Intermediate Weighted and Aggregate Weighted models: the larger the relative number that entered, the greater the degree of flexibility and the closer its flexibility to the Detailed L.P. PPAs in Benchmark Solution Relative to L. P. Number Detailed L. P. East Central West East Central West Detailed 16 18 24 - - - Int. Wt'd. 5 8 6 0.31 0.44 0.25 Agg. Wt'd. 5 u 6 0.31 0.22 0.25 103 OM w6: m6: wk “)0 O \D'fi \OI\ \Dh hi" >o:m> m6—: wé m67 ”6 1N7 33 UN d‘N MN u-i “N O.\D N\D NO Wk 33 “no? 35000 380 025.605 66 06 06 \00 CM :6: :6: duel CO CO 30.; n4: «A: 66—: “no? 38 OM m6: m.n: ~67 w6N: n67 w.o~: n67 :.:N: v.2: MO Wu N“ 00 #N --IO\ IAN 6.2 5N _mbcoU ummm 00300095 :6: :6: m4: 00:!" 005' \00 \D@ $0 06! DON zo=m> :.n: :.:: 5: :.m: :.:: 5.: n67 68 0—1 “N \DO 35 to? 3.5000 “mom moc< “08.03200 A65 ozumwoz moEE. A65 o>Emom 3:5. 6.; as up; .ww< Sm :Efifiaeu .95 .20 29$ .ww< and 539.0 6.5 .2. up; .ww< cam xcmEcucom A65 ozpmwoz 85:. A65 03:80 moEE. .50 up; .ww< .50 up; as .50 5.02396 .30 2p; .ww< .30 3.; as :3— 539.0 .30 up; .ww< .30 up; as and {08603 mCOmHD—OW mcofiaom 08.5303 vopmwocww< pcm .voEwEB 30:35.35 602030 5253 v.05 :: .SA 033. 10¢: the degree of bias is less than for area. At the Valley level, for example, production bias ranges from -O.l percent to 0.3 percent while area bias ranges from 6.# percent to 8.6 percent. With respect to production cost, the aggregated models were positively biased in 21 of 2# instances with a magnitude more similar to that of the area than to the production bias. Table 5-13 also presents the average signed error of the Aggregate Weighted model with respect to the Intermediate Weighted model. Except for the East, the Aggregate L.P. is usually, 21: of 27 times, negatively biased. As with comparisons of other L.P.‘s, the bias is least at the Valley level. The variance and coefficient of variation for the Intermediate Weighted yield estimates and for the Aggregate Weighted estimates each of which are composed of Detailed yield estimates suggests part of the information loss that occurs with aggregation. Tables A-28 and A-29 present the range of variance and of the coefficient of variation for the 14 Intermediate and 9 Aggregate PPAs. The narrowest ranges for the coefficient of variation are found for corn (0.10- 0.30), beans (0.12-0.33) and sweet potatoes (0.12-0.36); tobacco has the broadest range (0.16-1.74). Detailed, Intermediate, and Aggrgate Models The two models - Intermediate and Aggregate - were also run for the three policy questions and compared to the Detailed solutions. Input to these models varies both in number of PPAs and crop yield estimates. Table A-30 shows that the bias of the "independent"1 Intermediate yield estimates with respect to the estimates made for the Detailed PPAs composing each Intermediate PPA is positive 79 times out of a 100. For only two crops, rice 1The estimates are not truly independent from one another. Using the soil scientists' description of the PPAs, the agronomist ordinally ranked the production potential of each Detailed PPA on a 0-100 scale for each crop-low production method combination. He then ranked the Intermediate PPAs and then the Aggregate PPAs using the descriptions of these PPAs. Multiplication of the rank by the yield estimate for the combination produced the yield estimates used in each of the L. P.'s. 105 and corn, are the number of overestimates equal to the number of underestimates. Bias between yield estimates made for Aggregate PPAs and those for Detailed PPAs is positive 61 times out of 70. See Table A-31. The range in bias varies considerably among crops and is greatest for tobacco and rice and least for sweet potatoes. With the two exceptions of differences in crop transportation flows, the Detailed and Aggregate model runs have nearly identical Valley solution values in both the Growth and Growth/Shift run. See Tables A-34 and A-35. Except for transportation flows, Aggregate L.P. values for the Regions are identical and the Detailed L.P. shows only a 0.4 percent difference in corn and 0.2 percent in idle area between the two runs. In the Intermediate model, area and production cost differ in small amounts between runs at the Region level. Valley area for six crops are larger in the Growth/Shift run than in the Growth run but always by less than 0.6 percent. See Table 5-4. Transportation flows from supply to require- ment Regions often changed in this model: 35 out of a possible 90 (3 L.P.‘s x 3 Regions x 10 crops) opportunities. Rice flows between these three models are compared in Tables 5-1# and 5-15. However, impact on objective function values is very small. These values of the Intermediate L.P. differed only 0.91 percent and those of the Aggregate L.P. only 0.006 percent between the two runs. Consistent with the positive bias (with respect to the Detailed yields) of the Intermediate and Aggregate yields, solution crop area for all crops but rice are either less than or equal to the Detailed solutions in both runs. See Table 5-16. Recall that rice yields had little or no positive bias. As with the weighted yield models, differences between solution values seem to be greater at the Region than at the Valley level. Idle cropland is 17-2# percent greater in the Benchmark Intermediate and Aggregate than in the Detailed solution. See Table 5-17. Although the absolute amount of idle land is expectedly lower in the Growth than 106 .m_0_uoE 000.300 .026 SE0. 3 co_w0. 50.: 96: 05 «05 9.6% = 5.600%? on 005.00% .58 o..0~ 0.0 :03. 9. 000.0005: 000 co.w0.. 0 55.? 0002020 “caoEm 0.002050 3&5. 3.25 0.3.5 SEN oE.~ 85:. o.8k.~ 0.830 0.850 .28. - - - So: So: So: - - - ow£5o0 300m «$2.: .52.: 98...: EN EN EN N0. 5%. him} .83 o. 8%. n.m$.~ :69... o... a... 2.. man. “SN ..3 .2200 90%.... 5:53 .382 0% Sn 0% SE; 0.80.. :.:m; .80 .88.. 8.... 80 80.. 60 8o: 0pmw0..ww< 000.000.000— 00:m..0Q 0umw0..ww< 000605.000. 002900 Emmeg 000.000.000— 00:00.00 mco_w0.m 3.80 0030009.“. 00.32.... 00335029.... 02002095 300600.063. .38. cam x..mE;uc0m : mcofiaom 0umw0cmw< 0:0 000.006.000— .00:300 05 c00300m 0000.025 umoU cofiuaoocm 000 £32... 0230:8028... 6030200.... 00.”. :: .:.:n 030... 107 m.a$.~ ~63 n.wn. ~.wR.. o.oon.~ n.~au 5an 0.80.. o.oon.~ némmi ..wm 0.2m; .30... 0.803 . u 0.037. 0.08.. . u 0.08“. 0.08.. .. u 0.08.. 0mc.Eo0 020m 0 .EN o .03 u . o.on~ 0.03 .. u o .onm 0.03 . n 3003 0.00» Wm? néw. ~.wn. 0.0.x n.-n anN u 0.0.» o.o.w . . .3200 0.0% .. u 0.93 0.0% u . o.o..n 0.0% méom ..wm ...nmm 300m. .00 8o: .00 So: .00 8o: .30... $03 .3200 $00 .30... $03 .3200 $00 .30... 3.03 .3200 3000 0:0.m0m 20.304303 20.30 ...aanm 20....00 ...aaam 0202.33.00... 030w03ww< 30.002302. 02.300 :3. {0:20:00 u 20.2.8 30w3ww< .20 .30..00E..02. 602300 :003300 80:00.00 30.". 5.33.5023... 00.0 .i A...“ 0.00... 108 0000.200. @800 00.2.0. 0...... 8.... 00.88.. 8.... 30w00ww< 30.00.0002. 000300 0.0m0..ww< 0.0.00E..02. 000300 0.0m0..ww< 30.00.0002. 000300 00.w0~.\00..0 n... n... n... . E» o.» E» u N.» N.» N.» 3:3 ..... ..... ..... . 8» 8» on» . o.» o.» o.» .3... ..o ..o ..o u o. o. o. . N... N... N... .8030 * * t n t * * u * x. 0.. ummm . n 0000 ..0w..m ».N ».N ».N n 8..» can ON»... . c... o... o... 3:3 0.N 0.N ».N . ON... can.» ON»... . o... o... o... ...3. t .x * u * t * u * t 0.. 00.5000 9 o o . o o o . o o o .80 u " mcmcmm ».o. o.N. ».N. . 8.. 8.. 8.. . ».». ».N ...N 3:3 m.» 0.» N.N . 08 o». o». . n..: .... ..N. .83 N.. N.. ... . on on on u ..N ».N 0.. .2030 a.» N... a... . om. SN 2. u 0.» N.N ».N .30 n u pinata—n. N.» N.» m.» . 2.. 2.. 2.. . 0.2 0... ».N. 3:3 n.. ».. 3.. u 8 o» o. . ».N ...m 0.» .8... n.. ..0 N.» u E» on» 8» . ...N. 0... N.N. .2030 ».o ».o m... . o» o» 0» . n... 0... ».o .80 n u OUUMDOP ».NN ».NN ..NN . 8N.N 8N.N 8N.N . m...N »...N »...N 3:3 m... 0... ...m. . :2 EN EN: . .... n... ..n. .3... a; N.N .... . 8. RN 8. . 0.. N.N N.. .8030 0.... ».m. ».N. . o»N.. ON... 8»: . N.N. m... n..: .80 u u 003— 380 0030000.... 00.200000 000< 03. {0.000000 . 20.2.8 0.0w00ww< 000 62.00.0002. .00...300 0003.00 0.0003035 .000 000 60.303030 .03.... I .m...n 0.00... Table 5-16. (cont'd.) 109 hN*r\ hN—ch -tN-t:1~ ..._.*m O\N*-—a th—o 0 0 0 000 O 0.00 00 0 .0 0 0000 '00 -‘ -—aoo~ NOON —o -‘ HO N NOON \ON*°O #MMO -‘M*¢ fiN-‘é O‘N*—4 [\M*O 0 0 0 000 0 00 0 .000 00 0 00 0 —O -a ~OON NO N -:oo—a -¢O N NO m NIH-0M {\M*O (\n*N MMNOO O\N*-' °O¢*m 0 00 00 0 .0 0 0000 0 0 0 .0 0 ~OON —‘O N NO m -coo—n ~03 N NO M 0000 COCO 0000 0000 OO O Nm—«so urn—mo (\\DV\OO oocuma- n—*\0 SE"; 05““ 00 3.00. 5H 00 O 0000 0000 0000 OO 0 8:0 OW*\D O‘fi'ffiifi O\ONO mnnm \D—‘*\D had) 3‘ --' N 'AN 00 O\—t — (3.03 d d —-I use; U"C 030 “c- L- 0538 oooo oo o 000:: oooo oo 0 80-0 #0010 -n:r*tn a-N—ooo No—m can-”k: 05 :5 M—t if N N #M f\ CON—‘fi >~ ..4 0.: III! 00 .0 .0 00 00 00 00 00 00 0. 00 0. .0 00 00 .0 00 .0 00 00 00 .0 .0 .0 00 00 00 .0 0. 00 00 00 0. .0 00 00 .0 00 \DO‘MN O‘C‘NO OOOMN O‘ONN IANNM CONN OOOOO‘ “NOON O\—IO—I d'OO‘fi OOOOO‘ é—IOM I-d --4 —Il #N-iOO NNOO‘ OON—tN O‘NNOO d'OO—dm O\O—4f\ m—oo “Nd-0N O‘fiO—d GOO“ OOOOO‘ :f—tolfi C—l —4 --1 III! M—«OO—c O\a—c-—c—4 ON—IM l\-—400\D arm—cm \DN-flo OM04“ \D—tOOO (“NOW h—cON OOOOO‘ #NON U-I _Q U—l C-fl U—I v -n — -—I .—. —I 8 —t 3 >~ 3 >~ 3 >~ 8 3 >~ 3 >~ 5 3 >~ HHHO “up” “yo-:0 OHH'HU “wa _Hflflv 055'”: mmc‘”: Mac”: “#:013qu (00):”: 0.10:”: 030w0 00000 0.3000 3.53””0 Omvmm ~30”0 3 03> 0003> g 03> 33 03> 8003> 3 03> U :8 >- ma. 0 .5 (continued) “moo mo 83a ooodm no 50302096 mo mtc: n 6338; on emf 83* 110 M \D \D 000% O\O\D N'fiN oo o. o. o. o. u _. NNN _ . NNN _ . NNN 53> " N3 N3 3: so; “ SR SR can .228 u :5: 0;: ”a: :3 " Ewe... " tam—no.0 " 03 0.: n .S 53> “ k .3 3% Q8 ~33 " 3: 0; 3 3.5ch u 19 1: 9% :3 u 22 “ mi ”.2 Q: 5:3 “ mg 0.0 06 so; " 30 :.o :5 3:80 " N6 w .w n 5 ram Ouzummm Afizcouv .m Tn 03m... 111 .3va ooodn :3: $2 386 .5 .3E: m 323 29,2 Cetus—00.5 $3360; on cm...“ ..mszm 33m 02:65 “0: 80v 66:35:06 .5"— .cofiEOm 8:300 0%- 9.23m cos-36> 50:35 05 mm: c0328 .28.: £023 365 mfifitovcam mesa £330 mczm x66 6ch a Na 8 8 u 8 w» 8 8 380 .88 .98 878 878 :78 5-8 ” 83: 83K 3.8 2 73 3.8 8:883 ” Bmwvuwwt. % S o: R u 8 8 z: Ma 88 69a .33 872 828 28-8 878 " STE 8.? 878 83: £8 cozusuoi u mama—005.59:— 02-2: onN-am ST: 978 " ofl-ce con-R $78 87% 8:888 86 “ 3mm3ww< RE: com-R Rm-S 878 u 873 08-6 omN-R end-om 8:888 86 “ 33.65.35 $_ 9: 9m 8: n a. w: m: w: 22 378 83-8 ”2.8 278 “ 878 83-8 «2-8 wfl-R 82m 86 " vamwmhww’m N: ”2 § 9: u E m: 5 m: 22 N33 com-R SN-S 5-3 ” 872 82-2 32-8 5.? 82m 86 " 0550:3335 >m=m> H33 33ch “mam u >0=m> 303 33ch “mm-m acorn—om mczm 5330 van meEzucom - macs-20m .00—630 «0 mommy-:86.“ mm meets—om 3mm3ww< cam muvaEhopc— -- .Z-n 63w... 112 in the Benchmark run, the solution values, as compared to those of the Detailed L.P., are greater, #2 to 69 percent. This is partially due to widening the use of PPA area constraints from 70-120 percent of the Current Normal in the Benchmark to 70-130 percent in the Growth run. As the Table shows, crop area solution values as a percent of the Detailed solution vary widely both at the Region and less so at the Valley level. Production shifts widely between Regions and models. Except for those crops whose areas are fixed, sugar cane, bananas, and cocoa, each of the L.P.s usually generated production levels near to the right-hand side constraint. Except for rice, crop production cost in the Regions and for the Valley under both models is less than or equal to the Detailed solution; total production costs are less than the Detailed. This is again a consequence of the positive yield bias in the aggregated models. Objective function values for the three models and runs are shown in Table 5-1 1. In each instance, the aggregated models have a lower optimal solution than the Detailed model. The differences range from $3.3 million in the Intermediate Benchmark which is 4.4 percent less than the Detailed solution to $7.4 million in the Aggregate Growth which is 8.7 percent less. The Theil coefficients may have little explanatory value since the three L.P.‘s have different input data and differences in solution values of the three models would be expected. See Table 5-18. It is interesting to note, however, that for crop area, 9 out of 16 possible (3 Regions plus 1 Valley x 2 L.P.‘s x 2 runs) times these two models have Theil coefficients lo_w3§ than the Intermediate Weighted and Aggregate Weighted L.P.‘s. In other words, using the Theil statistic, it is not at all clear that the two L.P.s with the same yields data as the Detailed model are any better able to correspond to the Detailed L.P. croppingpattern.l l"U" statistics for Intermediate and Aggregate production are much higher than for the weighted L.P.‘s. This stems from not using the Detailed yields as was done in the weighted models. .:ofl:_8 30:00:33:— 53 003: 0:0 33:09:000 down—cm 003300 :33 003: 0:0 weentmaEOUn— 60328 6.33 .ww< :o .933 4.:— 05 3 020> 03:62. E0330 05 Emt 09.3.0 05 fl _0 30:28 002300 05 3 030> _mEcoZ 3:00.50 05 59: _ 0306330 058 E 0w:0:0 05 m_ .< 0053 IEH 2 3:32:80 1.0:... 05 :3 03:33 05.0 113 E 2.. w 0. o- 0 2:3 as.“ .35 805 8: 8~.o 3: 02 .N 8&2»? 0%: 3205320 Rm . N 8n .n n85 Rog 3m ._ N35 30.: an ._ 8823: 0:30 539.0 036 23.0 205 085 28.0 3.3 .085 «2.0 8323: 0:20 0:066:00 20.0 NS ._ ~35 :8 .n :n . N an... .o 03 ._ 09 .N 3&2»? «$5 92.. 3.2 find 83 80.: End :3 8285.85 :8: £52320 «$5 08.: find 3%...” 833 a: 8m; 2.; 3&8»? new .o E 4 8: 9m; 03.: on: $0; 03 ._ 8285.35 ncsm 539.0 S» .o $8 .o 2m .o Em .N an» .o ~85 NS ._ .5 .o 8&2»? NS 4 ”8.: SN; and. 28 .o Etc 83 :0 .o 8285.35 nczm {3:50:00 £86580 3 €885.80 2 >020> S03 .3300 $00 >0=m> $03 .3300 500 meets—om g mcoflaom 0umw0cww< 0:0 03300630“:— 602300 :003300 30:0—0:280:60 I .wdnn 030.— 1M Average signed error of the two aggregated models with respect to Detailed L.P. solutions was strongly negative (23 or 214) for area, on balance unbiased (12 positive and 12 negative) for production, and strongly negative (21 of 2(4) for production cost. See Table 5-19. This reflects the difference in yield input data. Similar bias held for the Aggregate model with respect to the Intermediate. 115 N N N 3 u c o N o u N N N N 35 2.00002 c o o N u N N o N u o o o o 35 03:8: 0N- o.N- N.N- N5- " N.N_ N.NN N.Nw- N.Na " N0- N5- N.NT N5- 85.85203 " u .50 £55320 0N- o.N.. N.N? N; " N.N_ o.NN NJN- 0.: “ N6- N5- N.NT N.N- gaze—”.0? " " :3m 5320 N4- N.N- N.N- Nd " N.N: N.NN N.:- _.N “ N.N- 0N- N6- 0N- .BE.oE_u.ww< ” u :31 x..0E:0:0m 33 " $3 “ Ga? 0 0 N 0 u N a a 3 u 0 N 0 0 35 033002 c o m o u m N N n u o d o o A65 0>Emom N.N- 0.2- N4- Z. n 0.0 N.o 3N- a._N “ Ni- N.NT 3:- 3:- 8:80:20? N6- N3. N5 .3. " N.N- o.NN- N.N: 0.2-" N.N- N.N? 90- N.N- 00:30:35 ” u :3” 02520390 N.N- 0.2- NJ. 0.3- u 0.0 N.N- N._N- N.N " 11:- N.NT 3:- o.:. 8:88;? N0- N67 0.2 N. " To- N.NN- :N 32$ " N.N- N.N? N.N- 0N- 8:30:35 " u :3m 5380 N.N- 5:- N4- :4. u 06 N.N N.N? o.NN " 05- N.N- N.N? N.N- 8:30:20? NJ- N.N- N6 N.N- " N.N- N.NN- N.N- N.0N “ 1NN- N.NN 3:- N5- 025.000.2— " u :30 x:0E:0:0m be n G? u as >0=0> $03 .03.:00 3.00 " >0=0> 50B .03.:00 3.00 “ >0=0> 503 .0550 50m mcofigom 380 5303005 5303.095 00.2 00003230 mcotaom 000w00ww< 0:0 00:00:...005 6020000 :00300m 005 l .me 030:. CHAPTER VI CONCLUSIONS The objective of this dissertation has been to investigate how varying the amount and type of input data affects the information output of an agricultural information system. Using the conceptual framework of CRIES in the Dominican Republic, five systems of the Cibao Valley were prepared. The systems are intended to provide information to national planners concerned with the valley's agricultural production potential. One aspect of this area of concern -- the least costly way to meet crop requirements —- is the focus of this study. The information produced is the result of partial, non-comprehensive analysis. It is designed to meet some, but not all, of the user group's information needs. Conceptually identical, the systems differ from one another only in terms of data inputs. Recall Figure 3-1. The real-world set of problems, the approaches of various disciplinary specialists to simplification, the means to operationalize these simplifications and the analyses undertaken are the same for all five systems. In addition, the intended user group -- policy analysts and planners concerned with the agricultural production potential of the Valley and its three Regions remain the same. The user is not as interested in the details of resource quality and use within the Regions as in the Regional location of important crops, in the Regional use of land for agriculture, in excess resource capacity, and in the ability of the Valley to meet domestic and export crop requirements. Since the Valley is treated as if it were a country, the hypothetical users are equivalent to national planners with avariety of backgrounds and experiences. With respect to 116 117 Figure 3-1, the necessary interaction between the user and other parts of the information system does not occur, and hence, the systems are incomplete. The different information systems are as follows: Detailed - a soil inventory was undertaken to identify #5 soil units (PPAs) classified as phases of soil families; crop yields were determined for each PPA. Intermediated Weighted a soil inventory was undertaken to identify 11+ phases of soil subgroup PPAs; crop yields on these PPAs are weighted averages of the Detailed PPAs that compose each one. Aggregate Weighted 9 phases of soil great group PPAs were identified; crop yields on these PPAs are weighted averages of the Detailed PPAs. Intermediate - crop yields on the 14 phases of soil subgroup PPAs are estimated independently from the Detailed estimates. Aggregate - crop yields on the 9 phases of great group PPAs are estimated independently of the Detailed estimates. The differences in data do result in the production of different information. Although information output varies by system, no attempt is made to say which solution is better or worse. This can only be done by the user. L.P. solution values pg _sg have limited utility due to the simplifying assumptions of L.P. and to the inability of the researcher to meet the four preconditions for maximization, see Chapter 11. Model results, therefore, are used only to show ordinal differences between models. 118 First comparing soil maps, the Aggregate soil map (and then the Inter- mediate) would appear to best meet the user's decision-making needs. The map is legible and the scale could even be reduced before seriously harming legibility. The number of map units, 3, is the same as the number of socio-economic regions that result when the user defines his minimum area of interest. The reliability of soils data is lowest of the three inventories but so is the cost and time involved in surveying, doing laboratory analysis, and mapping. See Table 6-1. It was first assumed that any of the soil inventories could be completed in one year. The author's experiences with agencies in several developing countries suggests that, due to administrative, logistical, and technical bottlenecks, this assumption underestimates the time needed for the more detailed inventories. It seems more reasonable, given the number of map delineations, to triple personnel time to 3 years, recall Table 5-3, in the Detailed inventory, and to increase personnel time 50 percent in the Intermediate inventory to 1.5 years. Per hectare costs would change. See Table 6-2. This change lends further support to selecting the Aggregate map. Second, comparisons to a base solution, that of the Detailed model, are made in two groups in order to estimate aggregation error that results from measurement errors. The first involves the Intermediate Weighted and Aggre- gate Weighted solutions; the second involves the intermediate and Aggregate solutions. As a result of system comparisons, summarized in Tables 6-1 and 6-3, statements about the relative acceptability of the systems by the user group are made. Recall from Chapter II that the small, simple L.P. specified and run at three different levels of soil detail has a high level of specification error. This does not vary between the Detailed, Intermediate, or Aggregate levels. Measurement errors also occur at the Detailed level. Assuming these are not biased, aggregation to the more aggregate levels tends to reduce measurement error as 119 20035980. .05 NNdm 892m 0000-. 30.0.0880 o - 0:33 500-. 0>50E:o.:. 0000-. m0> .E. 2: 8 Ezem o .2 8N3 .0000 2.8 01:03 030:0: 300... 2.2.3 3.0 U—DMF 8% I UUHMEmu—MO 3.09.5 >—nmno:av coo.nn~m .ww< :05 :0000:w 0.053058 3:30:00 A89. n «o 2050.08 :00 m - 0000000 :05 $0. u0£B0Eom 5.:3N. .ww< 0:0 .000 :003003 :o.«0E:o.:. .. 32200: x0.N.E00 500-. . .2 SNdN .ww< :05 :0500:w “0:30:50 5:30:00 .2\NN.NN .N-m 030... 000 u 00508.30 ..:00:3 b30393 08.2mm 30000.5 30.0. .333 30E .m 3 05.00.0850 m. .. 000:2. 03006.53. 08.2 am 282003 .3. :05 :0000:w 00:30:50 .0>0. 3:83:50 m .2 8N.N .0000 $.90. 0.003 030:0: “mos. 0:500: :0m .03... 080 8:258 3 5:32.00 0:3 30:: .30 55:3 23 :o 30:: co.“ -0E:o.:. 2.8 3 33:2 3.:0 30.2 .65 225005.00 no :03E3Z .8302 me: -0300: 030:3 “305.3 03.32. 0.000 :0..0Ew 5.08.an0 30E .0 00:w00 62:26 .3330... :05 :0..0Em 3.:0 3.:3 30:: no 0.2m 0w0:0>< 5:303 :22 5500.30 0500 >:oa:0>:. ..om E0093 E0093 5033 E05. 003303 000w0:ww< 000F303 000.00E:0u:. 0020500 2053 5.00.503. 000.303 000w0:ww< 0:0 €005.03 000.00E:0»:. .0020000 05 .3 0003095 5508.63. I .70 030... 120 30:50:00. ..0>0. >0..0> 00 000:000:. ..0>0. >0..0> 00 000:000:. 0 :o 0 :5 .8008 0 :o N :8 :3 .8: 0:: 0:0 00:0 .0:0.w0~. 0:003 0:0 00:0 .0:0. :00 000:000:. w. 00 n :. ” uw0m :0. 000:000:. w. 00 >30 0:.0> D :030. 00... .. :. 0:.0> D :030. 00... .00:.0> :0.w0~. :00 :05 >0..0> 05 :0. 0:0 00:0 00:0 :05 :000.0 :0::: 0. :0.:.0:00:: :00 00:00:0000::00 ..0:0:0w :. u .<.Z .0.0>0. :0.00:00:: 00:0 :0. :0.00.:0> :05 :0000:w 0:0 0000 : :0.00:00:n. 0:0 00:0 00:0 00 0:000E0:0a :00 0:0.00.:0> : ..0>0. >0..0> 00 5.028 000.000 05 00 00000.0 0:0 .0>0. “ .0:o.w0~. 05 00 :0000:w 0. 0:05:30 00 :0.00.:0> ..0:0:0w :. u ..0>0. ” ..0>0. >0..0> " >0..0> 00 :0.00.:0n>. 0000. u 00 :0.00n..:0> 0000. 00: u 00: 000:000:. 0 00 a c. : 000:000:. n 00 . >.:0 :. " .x:0:.::0:0m. :. 0:00:00 : 0:066:00 u wow. 0. >30 00:0 0.0. u :. 0:00:00 03 0. >30 " 0:0 000:0 .. :00 : 00:0 0.0. 0:0 000:0 u :0.00.:0> 00:w.0:: u .. :00 :0.00.:0> 00cw.0 u .000 .. 0.:0.00.:0> : n:: .000... 0.0:0.0 : .0:0.w0~. 0000. 00: : -0.:0> .0:0.w0~. 0000. " 000:000:. w. 00 0 >30 :. u 00: 000:000:. w. 00 m. :. " .<.Z .00.:: :08 00 003:5: 00 0:0 0.03 000w0:ww< :0. 0000. .00..0000 :00 000000:w 03 0.:03 >0..30..0: 00 .0>0. :0>.w 0 :00 0:3 0000E.000 002 ..00:00 00 0:00 0:0::0 00003:: 00 :0.00w0:ww0 5.3 0000:000 >0E 0:0::0 0:0E0::000.2 .2009: 03:50 00:: 0:3 00.0.0002: 0E00 00 00:09:00. :0::0 :0.000.0.00a0 00 .0>0. :w.: 0 :. 00300: 0.000.: 00:5 ..0 00 :0::E00 0:303:00 d... 05 .0 0:30: 000w0:ww0 .0.n.::.0 0:... .00000 0:000:05 0:0 :0.00:00:n. .0:0.00:3E00 0050:: :0.00:00:n.nn.0:0 .0 03038 00:0 :00 >0..30..0: :30:ch 06:00:80 .5: 3 :30:0 :0.0:.00 0020000 5.3 0050500250 0:05:30 0020000 00 000000: 5.3 :0.00.:0> 00 0::0::< 0:0.0:.0m 0000 8:083:25: 5500.30 0000 0.0022 $3503.05 :00:.... 0:28. -- ..-0 030:. 121 .038=&< 0oz -- .<.z ”082 .2 u 2 x N x m x C 80053.5 3 2% n Amcflwe m x 20.09: N x 2?. m x €000Eflma 3 300555 anu .n u 03 d .30— w u N .. 20.09: N x 9:: N x n.0u0Emhma 9 300535 n n .300 830.695 20 5252005 .005 nob «o m0wcm. 0.8 3000—:0th 05. .fi u N u $000080th m x m_0_uoE N x 2.: N x mco_w0.. 3 309.0me w—m .0303 cozzE ném .5 cm: 030% 5:28 aNm .8 $06 .Am0ES m Ho 3 65. >23 05 um 33 0030.30 300.. ._0>0_ _mco_w0m 05 um 33 002030 30.2 000003 85:: :.nm .5 03.0 000005 8:25 mam 5 fix ._0>0_ >0=m> 05 um 83 0020QO $0.2 $0.50 N no fiNv .0>0— _mco_w0m 05 um 33 €0cwfic3 0020QO $004 ..0>0. >0:m> 05 um m0E$ 2 no 5 0:0 _0>0_ _mc2w0m 05 “W 0033 Eggnog 05 >05 30:03.5. an no an E 603 -020th Low 0033:: 0:0 pmou cofiusvoa .050 00.0 U0pm> -330 ..3 0033 303062. 000 £0.09: 95 05 ..mh0c0w E 1 and 5380 cam meEcuc0m c2028 002900 3 000900.. at? .<.Z 020> .5383 0300030 30:28 8:800 2 0009.0. .<.Z 53 0.03 0mmpc0000m Afiucouv ..i Jaw 030... 122 Table 6-2.-- Revised Soil Inventory Costs Inventory Map Initial Cost Revised Delineations Assumptions Cost Estimates (no.7 (5/ ha.) Detailed 95 2.28 #.45 Intermediate ll» 0 . 70 O . 88 Aggregate 3 0.22 0.22 over- and underestimates tend to cancel out one another. To the extent that measurements at the Detailed level are biased, measurement error will increase along with aggregation. The net effect of aggregation error due to imperfect specifications and to inaccurate measurement is not known a priori but is shown by running the same L.P. model under different conditions. Both specification and measurement errors are frequently encountered in building an L.P. This has been the case in each country in which CRIES has worked and is usually the case in U.S. studies. Primary data generation to determine representative crop production techniques, the extent of crop produc- tion techniques by political region and by soil type, and yield by technique and soil are examples of data that are needed for analysis and that can be reliably established only with statistical survey techniques at several points in time. Statistical agronomic/economic surveys were not undertaken. Crop-produc- tion technique yields by soil type were ordinally estimated by a tropical agronomist. There was no reliable agronomic data from which to make esti- mates nor now was time-series data relating production, climate, soils or political regions available to infer relative productivity of different soils. Since Detailed crop yield estimates were not perfectly measured, their aggregation to Inter- mediate Weighted and to the Aggregate Weighted yield estimates resulted in measurement and aggregation errors. Crop yield estimates in the Intermediate and Aggregate inventories were based on progressively less soils data than in the Detailed inventory. Hence, their yield estimates are subject to greater variance 123 8005050: #6053 N: «o w :o_»0_:0> $0: 0>0: 200oE 000F303 ._0>0_ >0=0> 05 00 ..0>03oI * ._0>0_ _0:2w0m 05 0.0 3:3 on no :N m0wc0: 0020:; 0>0: $002: 002303-:0: 0:0. .200oE 00u:w_03 N 05 .3 :39? :o_u0_:0> Ho m0w:0: 05 8. 802388 :: $6.500 5:00:02: 0:0 00:0 no.6 :3 035 :05 000% :0So 0:0 m_0>0_ 50.0308: .5: m0w:0._ :o:0_:0> .:o0.0_..0> >0=0> 0000008 :o_u0_._0> _0:o_w0m .325: 000503 5:? m< *0._0>0_ >0=0> 0.0 :o_p0_._0> $00: 05 0.0.: 30:05:: 0 no N 3:0 :— * .:o_:.0_._0> _0:o:w0m 0&0: 05 m0: .usu>e_:u=m> 00 :o_u0:._0> $00: 06. m0: 80:03:: 0 no a :— * .:o_u0:._0> _0:o_w0m 0&0: 05 m0: 30:05:: m: :o m: :— w00:0pm:_ m: :0 0 Eco :: .<.Z 0:000:32: 0.m00:0o.m:_ on :o So :0 03:8: :0: no So an 03:8: .<.Z 6:5 :08 Ho 209:0: 05 op 0:0 000w0..ww< 0:0 .6: 300: 0:0 00:0000 05 .5: um0u00:w 0: 0303 523020: Ho _0>0_ :0>:w 0 :3 :5 0000::30 0oz 600002: :o_u0w0:ww0 m0 So _00:00 3 0:00 _0>0_ 00:00.00 0:0. 00 0803:: H: .9800 0:0::050002 6.80%.? 00:5 :0 3:0EE00 :009: d:— 0_:::m £5 :_ :20 53005020 £035 :.:: 3:30:00 56:ch .>:ou:0>:: 0~0w00ww< 0:0 :_ 000.5000 300— ".Cou:0>:: 00:00.00 05 E :o0.0:to::_ £8 Ho 0::oE0 000% 0:: 3 0:0 _0>0_ 00:00.00 06. :0 000.5000 3.2: 30009:: :.:: 3:30:00 :30:ch 22028 0020000 3 «0030.. :33 520205 no pczoE< 3.0; 8:80: 3 0009.0: :33 $3 m0_0_> mcofliom “moU :o_w0.<0p00 .050 3:: :05 \m0u0ESm0 0:0; no.0 5:000:00 0000 £0022 30:60..on 00:3 To 030:. 00m 70 030:. 00m 70 030:. 00m >..ou:0>:: zom E0093 " E0093 ” EBmzm E0: 0u0w0:ww< " 30:00:00:— " 00:00.00 2:00.93 5:05:85 0:0w0:ww< 0:0 .000_00E:0u:: 60:00.00 0:» .3 000209.: 5:08:35 3 .num 030:. 12‘: .mm n N - 2.0.33-5: 0:0 .033: £009: :0 30: N x m::: N x m:0_w0: m x 2000:: N x 3030890: m: ”30:03:: on 03002:}: 302 I. .<.Z "0302 0 .3 030500: 70 030:. 00mo .0 030508 70 030:. 00m: 58:03:... 22: o: 8 a... ”E: m x 30:0 0: "30:05:: on “0:0—:33 0:0; 0: 53 0:030:33 :0:U n (in: o: 00.0: m0_ >0=0> 0:0 _0:0_w0m 50: 30 :03 “00:32:: 032030 «mo—2 .m_0>0_ >0=0> 0:0 _0:o_w0m 50: 30 m03 A00:w_m::: 0320.30 300.: * :22 :20; 05 30 0.95:. m: 30 2 0:0 _0>0_ :0:0_w0m 05 30 00m03 30>30w0: 0:03 >05 30:03:: 3 mo 3 :— .:030:0o:: :03 0803:: 0:0 #500 :030:0o:: :03 0:0 #020 :0:0 :0: 0083 30>30w0: bwcotm 0:03 3000,: 033 05 ._0:0:0w :: ..00:0300 05 03 00:00:0%0::00 :0mo_0 m3o:m 33039200 300:: :0::02 00:80:20 :02: w:300_0:: :.Tud E03 m0wc0: 3:0:033000 :3: 05 $2820 0:9: 08:00:50 0.000 :030:0o:: >0:0> 02:3 *.:03:_0m 0020300 05 03 308—0 0:9: 08:00:00 0:. E000 :030:0o:: :0: :0:0 :0:o_w0: :05:02 :3: 530:0 :3: 010:: :0:0m 8328 0020300 03 3009.0: 53 .<.Z 0:_0> 530:3 0>30030 20320.: 0020300 03 .<.Z ..00:00: 53 m03 0w03:00:0: 3:033:00“: ..D: .3 :30:m :03:_0m .<.z 0020300 53 00:00:0%0::0U :.:28: -- .3 2%: 125 than the more detailed estimates. Similarly, aggregation of the Intermediate map unit—level land use inventory to Aggregate map units would possibly decrease measurement error; disaggregation of the Intermediate land use to Detailed map units would increase both types of error. Aggregation produces errors in each of the aggregated L.P.‘s, but more error in the Intermediate and Aggregate L.P.‘s than in the Intermediate Weighted and Aggregate Weighted L.P.‘s. However, the aggregated L.P.‘s are progressively less costly and less time-consuming to build. The user is interested in Regional location, production, and costs of specific crops. Other things being equal, the Intermediate Weighted model would be preferable to the Aggregate Weighted model since in 13 of 18 instances its solution values were closest to the Detailed solution, the base model for comparison. Furthermore, for one parameter, crop area, total unsigned varia- tion from the Detailed Regional values is only one-third the variation resulting from the Aggregate Weighted solution. This preference must be tentative in view of the very large variations from the Detailed solutions that occur for some crops. For example, area of plantain in the West is 36 percent greater than the Detailed solution value; interplanted area in the East is 62 percent greater. In general, the large values of the Theil coefficient indicated a lack of correspondence among solution values of all three models. Objective function values for the three models are very close. That is, the "most efficient" cropping patterns of the two aggregate weighted models had associated costs of just 4-6 percent greater than the Detailed model. Comparisons of the Detailed with the Intermediate and Aggregate models entailed using "independent" yield estimates in the three models. Given the lower reliability of soils data in the Intermediate and, even lower, in the Aggregate inventory, it is concluded that the yield estimates are subject to progressively more measurement error in the aggregated models. Recall that yield estimates in both unweighted models with respect to the Detailed yield estimates have a 126 strong positive bias. This positive bias has the direct effect of producing a negative bias in solution values for crop area and production cost. The use of these estimates changed the bias of their L.P. solutions (with respect to the Detailed model solution for crop areas) from positive to strongly and consistently negative. The Aggregate model may appear most suited to the user's needs since it is the least costly to generate data for and to construct. It also most often (13 of 18 times) has the least variation with respect to Regional Detailed solutions of crop area, production, and costs and has an Objective Function value 6-9 percent below that of the Detailed model. However, it may not be preferable to the Intermediate model due to its less accurate crop yields. Therefore, preference for the Intermediate model's information could derive from its more accurate crop yields, from the fact that it has the smallest amount of absolute (unsigned) bias with respect to the Detailed Regional and Valley solutions, and third, from its ability to produce an optimum solution 4-6 percent less than that of the Detailed model. In comparison to the ranges of variation around Detailed solutions shown by the two weighted models, the non-weighted models have smaller ranges 21 of 36 times at the Regional level. This should 955 be interpreted as producing a smaller amount of aggregation error. It does show how measurement errors can significantly affect results. Given this sensitivity to yields, one building such a system ought to obtain approximately consistent levels of reliability between parameters. This certainly did not occur in the Detailed system where physical and spatial soils information was more reliable than the yield estimates. It may, or may not, have occurred for the more aggregated systems. A final note on costs supports constructing systems more aggregated than the Detailed. It is reasonable to assume that the variable costs of generating such data at a given level of reliability and of preparing data in the format 127 necessary for the L.P. algorithm would also be at least directly proportional to the number of map delineations. The least cost and time would be incurred in generating the Aggregate (and then the Intermediate) data set. Assuming that the agronomic/economic surveys could begin once the soil mapping was one-half completed, a total time requirement for all surveys, analyses, and publication of results could require 4.5 years (Detailed), 2.3 years (Intermediate), and 1.5 years (Aggregate). Including L.P. analyses and report writing, total study time could then require approximately 5, 3, and 2 years respectively, and costs of $2.23 million, $0.44 million, and $0.11 million, respectively. As stated repeatedly in this dissertation, the utility of any one information system can only be determined by the user. In view of the relatively long time and high cost of the Detailed soil inventory - linear programming system, the ability of L.P. to provide only a limited amount of information on the subject matter under analysis, and the interest of the user in information about just three regions, it seems obvious that the user would prefer one of the more aggregated information systems. A clear choice between the Intermediate or Aggregate system could be made only by the user. He would give priority either to cost-time considerations or less aggregation error. Recommendations Possibly the most general and most important recommendation that can be made is that the resource specialist(s) building the information system be objective. That is, that the system be logically consistent; that its assumptions, relationships between parameters, data inputs, and information outputs correspond, not totally but in important ways, to reality; that the conceptual approach and use of analytical tools be clear; and, that it produce information which will assist the user develop/modify policies relevant to his area of concern 128 or to a specific problem. To accomplish this, it is suggested that the resource team periodically work through steps of a process such as shown in Figure 3-1. Second, it is recommended that this process work be done with the user or his representative. The team should clarify the user's perception of the area of concern and his needs and values in this regard. Discussion of the type of information produced in previous studies of similar resource issues would help the team and user strengthen each of the linkages in the process of Figure 3-1. For example, if an L.P. of the type structured in this dissertation was one of the analytical tools under consideration, discussion with the user of the strengths and weaknesses of L.P. in this subject-matter research would improve the credibility of the team's recommendations. Third, it is recommended that future research regarding the appropriateness of alternative models for analyses of agricultural production potential focus upon the values, perceptions of production potentials, and needs of various kinds of user groups. This should be the focus rather than on means to reduce and/or control aggregation errors. First, cost and timeliness considerations and the ability to make informed statements about only his minimum geographic regions of interest may be very important to the decision-maker. Second, since it will always be impossible to specify perfectly relationships among important parameters -— edafological, agronomic, economic, social, and so forth -- aggregation of large into small models will increase aggregation error. In the face of almost always unreliable data from secondary sources and the possibly prohibitive costs of generating primary data with a known reliability, measurement errors will exist even in large and complex models. Depending upon the amount of measurement bias that exists, aggregation will lead to growing amounts of measurement and hence aggregation error. Finally, even if specification and measurement errors could be avoided, it would be impossible, in practical terms, to meet Day's homogeneity conditions to aggregate without error. 129 A final recommendation is that researchers avoid developing highly reliable estimates for one parameter, e.g., soil area and type, and much less or unknown reliability for other parameters, e.g., crop yield by soil-climate type. Not to do so precludes developing an internally consistent system and one whose results will correspond to reality. System results may be costly in that their reliability is unknown. Given these concerns, the team should attempt to determine, from previous studies or a hypothetical model of the work under discussion with the user, the parameters to which production of information is the most sensitive. Similar levels of reliability should be developed for each of these key parameters. This may mean achieving much lower reliability levels for some parameters than currently occurs in many studies. 10. ll. 12. 13. I30 BIBLIOGRAPHY Agronomy Department, State College of Agriculture and Life Sciences. _S_o;i_l_ Risource Inventories, Proceeding; of a Workshop. Agronomy Mimeograph 77-23. Ithaca New York: Cornell University Press, 1977. Amos, D.F. and Whiteside, E.P. "Mapping Accuracy of a Contemporary Soil Survey in an Urbanizing Area". Soil Science Society of American Proceedings, 39 (September 1975): 937-42 Arscott, Trevor. Personal letter. Columbus, Ohio: Ohio State University, June, 1979. Atchley, Alan. Oral presentation at CRIES staff meeting. East Lansing, Michigan, July, 1978. Aubert, G. "Soil Survey: Different Types and Categories". In Soil Resource Inventories: Proceedings of a Workshgp, Agronomy Department. Ithaca, New York: Cornell University Press, 1977. Banco Agricola; "Planes de Inversion para las Distintas Labores de Cacao a Realizarse Bajo Credito". Santo Domingo, Dominican Republic, 1978. (Typewritten.) Barker, Randolph and Stanton, Bernard. "Estimation and Aggregation of Firm Supply Functions". Journal of Farm Economics, 47 (August 1965): 701-12. Beckett, P.H.T. "Method and Scale of Land Resource Surveys, in Relation to Precision and Cost". In Land Evaluation: Papers of a CSIRO Symposium, edited by George A. Stewart. Melbourne: Macmillan, 1968. Bie, S.W. and Ulph, A. "The Economic Value of Soil Survey Information". Journal of Agricultural Economics, 23 (September, 1972): 285-95. Bleisch, Robert. Organization of American States. Natural Resources Division, Washington, D.C. Telephone Interview, 23 April, 1979. Bonnen, James T. "Assessment of the Current Agricultural Data Base: An Information System Approach". In A Survey of Agricultural Economics Literature, Volume 2, edited by Lee R. Martin. Minneapolis: University of Minnesota Press, 1977. Bonnen, James T. "Improving A ricultural Information". American Journal of Agricultural Economics, 57 December, 1975): 753-63. Boss, Gary et. a1. "Food and Agriculture Models for Policy Analysis". In System Theory Application to Agicultural Modeling, A Proceedings, 14. 15. l6. 17. 18. I9. 20. 21. 22. 23. 24. 25. 26. 27. I31 edited by Alexander Levis and Leroy Quance, Economics, Statistics and Cooperatives Service, U.S. Department of Agriculture, ESCS-07. Washington, D.C.: Government Printing Office, 1978. Brinkman, R. and Smyth, A.J. (eds.), Land Evaluation For Rural Purposes, Summary of Expert Consultation Wageningen in 1972. Wageningen: Inter- national Institute for Land Reclamation and Improvement, 1973. Buckwell, Allan E. and Hazell, Peter. "Implications of Aggregation Bias for the Construction of Static and Dynamic Linear Programming Supply Models". Journal of Agricultural Economics, 23 (May, 1972): 119-134. Buol, S.W., Hole, F. and McCracken, R.J. Soil Genesis and Classification. Ames: Iowa State University Press, 1973. Burrough, RH. and Beckett, P.H.T. "The Relation Between Cost and Utility in Soil Survey". Journal of Soil Science, 22 (September, 1971): 359-394. Bylinsky, Gene. "EDP Managers Put On Business Suits". Fortune, November 6, 1978. Caplan, Natham et.al. The Use of Social Science Knowledge in Policy Decisiolns at the National Level. Ann Arbor, Michigan: Institute for Social Research, 1975. Chappelle, Daniel E. "How Much is Information Worth?". In Proceedings: Resource Data Management Symposiug, edited by John Moser. West Lafayette: Purdue University Press, 1976. Chappelle, Daniel E. "Precision Required in Allowable Cut Determinations on the National Forests: An Inquiry into the Nature of the Appropriate Loss Function". East Lansing, Michigan: Michigan State University, 1975. Cline, M.G. "Objectives and Rationale of the Cornell Study of Soil Resource Inventories". In Soil Resource Inventories and Development Planning: Proceedings of a Workshop, Agronomy Department, Ithaca, New York: Cornell University Press, December 11-15, 1978. Comprehensive Resource Inventory and Evaluation System; Economics, Sta- tistics and Cooperatives Service, U.S. Department of Agriculture. "Land Inventory and Crop Totals". East Lansing, Michigan. July, 1977. (Type- written Working Draft). Cummings, Ralph Jr. "Minimum Information Systems for Agricultural Development in Low-Income Countries". New York: Agricultural Development Council Inc., 1977. Day, Richard H. "On Aggregating Linear Programming Models of Produc- tion". Journal of Farm Economics, 45 (November, 1963): 797-813. Dent, 3.8. and Anderson, J.R. Systems Analysis in Agricultural Manage- ment. New York: John Wiley 6: Sons, 1971. Egbert, Alvin C. and Hyang M. Kim. "Analysis of Aggregation Errors in Linear Programming Models". American Journal of Agricultural Economics, 57 (May, 1975): 292-301. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. #1. 42. 132 Eisgruber, Ludwig M. "Developments in the Economic Theory of Informa- tion". American Journal of Pmcultural Economics, 60 (December, 1978): 901-5. Eisgruber, Ludwig M. "Micro- and Macro-analytic Potential of Agricultural Information Systems". Journal of Farm Economics, 49 (December, 1967): 1541-52. Eswaran, Hari. "Evaluation of Soil Resource Inventories". In Soil Resource Inventories and Development Planning, Proceedirgs of a Workshop. Pre- pared by the Department of Agronomy. Ithaca, New York: Cornell University Department of Agronomy, December, 1978. Fisher, Walter D. Clustering 6c Aggregation in Economics. Baltimore: Johns Hopkins, 1969. Frick, George E. and Andrews, Richard. "Aggregation Bias and Four Methods of Summing Farm Supply Functions". Journal of Farm Economics, 47 (December, 1965): 696-700. Fromm, Gary and Schink, George. "Aggregation and Economic Models". International Economic Review, 14 (February, 1973): 1-32. Grumfeld, Yehuda and Griliches, Zvi. "Is Aggregation Bias Necessarily Bad?". Review of Economics and Statistics, 42 (February, 1960): l-13. Haverly Systems Inc. Magen Linear Programming System. Denville, New Jersey. 1978. Heady, Earl O. "Programming Models for the Planning of Agricultural Sector". In The Theory apd Design of Economic Development, edited by Irma Adelman and Erik Thorbecke. Baltimore: Johns Hopkins Press, 1966. Heady, Earl 0. (ed.). Economic Models and Quanitative Methods for Decisions and Plannim in Agriculture, Proceedings_ of an East-West Seminar. Ames, Iowa: Iowa State University Press, 1971. Heady, Earl O and Srivastava, Uma. Spatial Sector Programming Models in Agriculture. Iowa State Univ. Press, Ames, Iowa: Iowa State University, 1975.. Heady, Earl O. and Candler, Wilfred. Linear Programming Methods, Ames, Iowa: Iowa State University Press, 1958. Herfindahl, Orris C. Natural Resource Information for Economic Develop- ment. Washington, D.C.: Resources for the Future, 1969. Holcomb Research Institute. Environmental Modeling and Decision-Making. New York: Praeger Publishers, 1976. Johnson, Glenn L. "Contributions of Economists to Rational Decision- Making Process in the Field of Agricultural Policy". In Papers and Reports of the XVI International Conference of Agricultural Economists. Oxford: Agricultural Economics Institute, 1977. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 133 Johnson, Glenn L. A Generalized Simulation Approach to Agricultural Sector Apalysis. East Lansing, Michigan: Michigan State University Press, 1971. Johnson, Glenn L. "Opportunities for Systems Scientists in Modeling Agricultural Systems". In System Theory Applications to Agricultural Modeling, A Proceedings, edited by Alexander Levis and Leroy Quance. Washington, D.C.: Government Printing Office, February, 1978. Johnson, Glenn L. "The Roles of the Economist in Studying Problems Involving Energy and Food", Staff Paper 1974-23, Dept. of Agricultural Economics. East Lansing, Michigan: Michigan State University, 1974. Johnson, Glenn L. and Rossmiller, George. "Improving Agricultural Decision-Making: A Conceptual Framework", in Agricultural Sector Planning, Department of Agricultural Economics. East Lansing, Michigan: Michigan State University, 1978. Johnson, W. M. "Soil Classification and the Design of Soil Surveys", In _S_o_il Resource Data for Agricultural Development, edited by Leslie Swindale. College of Tropical Medicine: University of Hawaii, 1978. Kohout, Jose. Sector Agropecuario de la Republica Dominicana Informe Puliminar, Banco Interamericano Desarrollo, Washington, D.C.: Banco Interamericano de Desarrollo, 1975. Knox, Ellis. Personal communications. Columbia, Maryland: Soil and Land Use Technology, December, 1978 and April - July, 1979. Lekachman, Robert. A History of Economic Ideas, New York: Harpter as: Row, 1959. Leuthold, Raymond M. "On the Use of Theil's Inequality Coefficients". American Journal of Agricultural Economics, 57 (May, 1975): 344-6. Levis, Alexander H. and Levis, Ilze S. "Toward 3 Conceptual Framework for Policy Analysis Modeling". In System Theory Applications in Agricultural Modeling, A Proceedings, edited by Alexander Levis and Leroy Quance. Washington, D.C.: Government Printing Office, 1978. Levis, Alexander H. and Quance, Leroy (eds.). System Theory Applications to Agricultural Modeling, A Proceedings. Washington, D. C.: Government Printing Office, Economics, Statistics and Cooperatives Service, U. S. Department of Agriculture, 1978. Martin, Lee R. et al. (eds..) A Survej of Agricultural Economics Literature Vo_l._ 2. Minneapolis: University of Minnesota Press, 1977. Miller, Thomas A. and Heady, Earl O. "Analysis of Aggregation Error in Supply Functions Based on Farm - Programming Models". Agriculture and Home Economics Experiment Station, Research Bulletin 578. Ames: Iowa State University, 1973. Myint, Hla. "Agriculture and Economic Development in the Open Economy". In Agriculture In Development Theory, New Haven, Connecticut: Yale University Press, 1975. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 134 National Academy of Sciences. Supporting Papers of World Food and Nutrition Study, Profile 4 -- Technical Land Classification Systems. Washington, D.C.: Government Printing Office, 1977. National Academy of Sciences. Supporting Papers of Worldeood and mitrition Study, Vol. 111, Study Team - Information Systems", p. 173- 237. National Science Foundation. "Workshop on Modeling Large Scale Systems at National and Regional Levels. Washington, D.C.: Brookings Institu- tion, February 10-12, 1975. Naylor, Thomas H. "Policy Simulation Experiments with Macroeconometric Models: The State of the Art". American Journal of Agricultural Economics, 52 (May, 1970): 263-71. Nicol, Kenneth J. and Heady, Earl O. A Model for Regional Agricultural Analysis of Land and Water Use, Agriculture Structure, and the Environ- ment: A Documentation. Ames: Center for Agricultural and Rural Development, 1975. Ogando, Jose. Secretaria de Estado de Agricultura, Subsecretaria de Recursos Naturales. Proyecto SIEDRA, Interview, 4-5 May 1979. Olson, Gerald W. "Land Classifications". In Search for Agriculture. Agricultural Experiment Station, Vol. 4, No. 7. Ithaca, New York: Cornell University Press, 1974. Orcutt, Guy H., et. a1. "Data Aggregation and Information Loss". American Economic Review, 58 (September, 1968): 773-786. Organizacion de Estados Americano. Plan de Accion para el Desarrollo Regional de la Ling Noroeste. Washington, D.C.: Organization of American States, 1977. Reconocimiento y Evaluacion de los Recursos Naturales de la Republica Dominicana. Washington, D.C.: Union Panamericana, I967. Organization of American States. Survey for the Development 2 the Guayas River Basin of EcuadoA An Integrated Natural Resources Evalua- tion. Washington, D.C.: Organization of American States, 1964. Paris, Quirino and Rausser, Gordon. "Sufficient Conditions for Aggregation of Linear Programming Models". American Journal of Agricultural Economics, 55 (November, 1973): 659-66. Parsons, Kenneth. "Logical Foundations of Economics Research". Journal of Farm Economics, 31 (November, 1949): 656-686. Rein, Martin and White, Sheldon H. "Policy Research: Belief and Doubt". Policy Analysis, 3 (Spring, 1977): 239-271. Rose, J. "Maryland Soil Survey Under Pressure". Journal of Soil and Water Conservation, 33 (May/ June, 1978): 105-111. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 135 Rossmiller, George E., et. al. "Values and Policy Choices in Agricultural Development", In Agricultural Sector Planning. Department of Agricul- tural Economics, Michigan State Univeristy. East Lansing, Michigan: Michigan State University Press, 1978. Schmid, Allan A. Property, Power, and Public Choice, East Lansing, Michigan: Department of Agricultural Economics at Michigan State University, 1978. Secretaria de Estado de Agricultura Oficina Nacional de Estadistica. Republica Dominicana en Cifras. Santo Domingo, Dominican Republic, 1978. . Diagnostico -.---.--.......-----.’ ...... del Sistema de Mercadeo Agricola en Republica Dominicana. Santo Domingo, Dominican Republic, 1977. __________ , Centro de Desarrollo Agropecuaria, Zona Norte. "Costos de Produccion del Maiz y Frijol en la Region Norte". Santiago, Dominican Rebpulic, October, 1978. (Mimeographed). ._ _______ , Division de Administracion Rural, Regional Nordeste. "Costos de Produccion de los varios cultivos". Santiago, Dominican Republic, April, 1979. (Typewritten). Shapiro, Harold T. "Is Verification Possible? The Evaluation of Large Econometric Models". American Journal of Agricultural Economics, 55 (May, 1973): 250-58. Sheely, Stamus J. and McAlexander, R.H. "Selection of Representative Benchmark Farms for Supply Estimation", Journal of Farm Economics, 47 (August, 1965): 681-75. Singh, [.1 "Book Reviews of Models for Spatial Agricultural Development Planning and of Spatial Sector Programming Models in Agriculture", American Journal of Agricultural Economics, 58 (November, 1976): 768- 71. Smyth, A.C. "The Objectives of Soil Surveys of Various Intensities". In fig Resource Inventories, Proceedings of a Workshop, Agronomy Department Mimeograph 77-23. Ithaca, New York: Cornell University Press, 1977. Steele, J. Gordon. Soil Survey Interpretation and Its Use, Soils Bulletin 8. Rome: Food and Agriculture Organization, 1967. Stobbs, A.R. "Soil Survey Procedures for Development Purposes". In Ne_w Possibilities and Techniques for Land‘Use and Related Studies, edited by Ian H. Cox. Berkhamsted, Herts., England: Geographical Publications Ltd., 1970. Stovall, John G. "Sources of Error in Aggregate Supply Estimates". Journal of Farm Economics, 48 (December, 1966): 477-80. Swindale, Leslie D. (ed.). Soil Resource Data for Agricultural Development. Hawaii Agricultural Experiment Station, College of Tropical Agriculture. Honolulu: University of Hawaii, 1978. 86. 87. 88. 89. 90. 91. 92. 93. 136 U.N. Food and Agriculture Organization. Diversificacion y Aumento de la Produccion Agricola en el Valle del Cibao, Informe Tecnico 11: Estimacion de los Costos de Produccion de Algunos Cultivos. Santiago, Dominican Republic, 1974. . Diversificacion y Aumento de la Produccion Agricola en emaile- del Cibao. Informe Tecnico 22: Estudios de Perspectivas de Desarrollo Agropecuaria. Santiago, Dominican Repub- lic, 1974. . Diversificacion y Aumento de la Produccion Agricola del Valle del CibaoLInforme Tecnico 6: Mapas Semi- Detallado y Aptitud4de los Suelos del Valle del Cibao. Santiago, Dominican Republic, 1974. U.S. Department of Agriculture, Soil Conservation Service. Revised Soil Sung! Manual, Agricultural Handbook 18 Revi_sed, Fifth Draft. Washing- ton, D.C.: Soil Conservation Service, June, 1975. . Soil Taxonomy, Agricultural Handbook 436. Washington, D.C.: Government Printing Office, 1975. and Michigan State University. Honduras Land—Resource Base-Report. Working Report published by the Comprehensive Resource Inventory and Evaluation System (CRIES). East Lansing, Michigan: Michigan State University Press, 1980. U.S. Environmental Protection Agency, Office of Research and Develop- ment. A Guide to Models in Governmental Planning and Operations. Washington, D.C.: Government Printing Office, August, 1974. Vernon, Raymond. "Comprehensive Model-Building in the Planning Process: The Case of the Less Developed Economics", Economic Journal, 76 (March, 1966): 57-69. APPENDIX 138 Table A-1. -- Measured Areas of Soil Map Units Detailed Map Unit Intermediate Map Unit Aggregate Map Unit No. (kmz) No. (kmz) No. (kmz) 1 71.9333 .A 282.2585 I 302.6215 2 37.8478 B 1008.8568 H 1526.1653 3 54.9364 (2 242.7887 HI 457.5544 4 30.3693 1) 23.9850 5 109.8743 E 571.1855 6 607.4416 F 33.8481 7 98.1878 (3 58.4022 8 87.1817 9 23.5748 10 29.2654 11 37.7887 12 6.4423 13 8.0943 14 50.8096 15 37.2069 16 13.6180 17 29.2592 18 29.1648 19 11.2257 20 162.4731 21 15.5636 22 19.7731 23 18.4843 24 70.5966 25 19.4060 26 49.9989 27 176.2529 28 45.6754 29 84.6362 30 41.4802 31 100.9671 Total Area 2179.5303 2221.3248 2286.3412 139 not... (mm .33 no flow m_ 22 oEsmm. owmucoouom «coacom 85322 m299. m van < 6.6QO co. oo— oo. oo. oo— oo. oo. oo— oo. co. co. oo. oo. oo— 1 l l l I o. S cm I. l l 1 l 1 och cmwam l ..- I l ..- 6 n o. l l 1 l l 1. mcmcmm ow cm on c. on 6 . n . mm 1 1 1 l .i l 535$ 11 l l l ow o . o . l ..- cm 1 l l l ouumno... 1.. .1 l l l 1 l 1 I .1 o . l .1 o. moooU o. n l 0. cm 0. o. 3 cc. 1 l 6.. O6 1 836.66. o. n l l ..- ..- ..- 1 l 0. cm 3 n o. 6.6—6:92 l on on em 0. on on mm 1 l l n.. m cm 02m 60 6o .0 6n. .6 6m 6.”. .... .6. .o .6 6.. 6.6 .< 620 6.6.6.6. 36.—66:663.... 63:26.80 036.0... Ho 63 moumEflmm 22... .663 paw—6.9.0 (mm ouflvoEEBE .i .mu< 03m... 28:56:08 .6865 5:6 8930a 6836 .538... 5 88 666. .58 *3 .88.. 66cm .6. 0:26 86:56. 688 16:53.65 mo oucmtan_ 63.55 on... 6866.3 666 .656 65606. $6.6m .836on 686.6 6.60. .68.... 666.2 .58 666. .5656... 6.6.9.6 "6. 68.5 86.6.0885. 05 no 6.0.56an8 9:. .389 8.508635 569% mm :63 6.6 .6605 6:26 .6665 ..6—macaw: 6.5.56.6 U .656 .m .< 65:: ..6—2 3585.356 I49 6.6 .6 666 666 666 666 666.. 666.. ..6 66 1 6.6. 66 1 6.6. .6. 1 6.6. 6686 ..6 66 1 6.6. 66 1 6.6. 66. 1 6.6. 86.86.. 686.6 ..6 66 1 6.6. 66 1 6.6. 66. 1 6.6. 6.6.66... .66 .6 1 6.6. 66 1 6.6. 6.6 1 6.6. 58 666 66. 1 6. .6 66 1 6..6 .66 1 6. .6 88> 1 1 1 66 1 1 66 1 1 66 6:6... .8666... 666 66. 1 66 66 1 66 666 1 66 8662886. .65 .65 .65 .66. .65 .65 .66. .65 .65 .66. amp—f. mute. “cvuhvm mmv6< mv6< pCUqum mmvh< wok/6. accouvm QOEU Bauer—xx QOLU m_mscc< GOLU m—macc< ww—macc< «O town—Datama— 666.66 66..6 666.6 66. 666.6 6.6.6 66. 6.6.6. 666.6. 66. 6.6.6 666 1 6. 666 1 6 666.. 1 6. 6.66:6... 666.. 6 1 6 6 1 6 666.. 1 6. 6680 666.6. 666.6 1 66 666 1 6 666.6 1 66 8:. 666.6 666.6 1 66 666.6 1 66 6 1 6 8366.. .66.66 1 66. .6 1 1 666.6 1 1 6.6.6. 1 666.680 66.60.. 666 .6 .. 666.. 1 1 666.. 1 1 66. .6 1 6..6. .65 .65 .65 .66. .65 .65 .66. .65 .65 .66. .68... 66.9.0 6.6.6.650 tam—6.9.0 66.50 cam—6.50 tam—no.0 66.50 cam—6.50 new—6.50 no.5 :65 6.52 638... 66.8... o>$u< we “COP—om wo ucvuuvnw HO HCUULQQ 6.6 6... .< <9. 35.85635 6.165 3585.35 3 6.86.6. 6.50 mo 83.68.36 I .o_1< 033. 150 Table A-IO. -- (cont'd.) B1 Percent of Active Crop Cropland Cropland Crops (96) (Ha.) (Ha.) Idle - 20 , 177 - Active Cropland - 80 , 709 - Annuals 90 - 72 , 638 Cocoa 10 - 8 , 071 100 100,886 80,709 Distribution of Annuals Annuals Crop Percent Area Areas (96) (Ha.) (Ha.) Intercropped 30 - 21 , 791 Planted Alone 70 - - Yucca 21.0 - 15,254 Corn 17.5 - 12,712 Plantain 10 . 5 -- 7 , 627 Sweet Potatoes 10. 5 -- 7 , 627 Beans 10.5 - 7,627 100 72,638 72,638 (continued) Table A-10. - (cont'd.) I51 C1 Percent of Active Crop Cropland Cropland Crops (96) (Ha.) (Ha.) Idle - 4, 856 - Active Cropland - 19,423 - Annuals 10 - 1 , 942 Tobacco 90 - 17 , 481 100 24,279 19,423 Corn 12 . 5 - 243 Beans 12 . 5 - 243 Yucca 12 . 5 - 243 Sweet Potatoes 12.5 - 243 Plantain 12 . 5 - 243 Tobacco 12 . 5 -- 243 Rice 12 . 5 -- 243 Sugar Cane 12.5 - 241 100 1,942 1,942 D1 Percent of Active Crop Cropland Cropland Crops (96) (Ha.) (Ha.) Idle - 480 - Active Cropland — 1,919 -- Pasture 100 - 1, 919 100 2,399 1,919 (continued) 152 28:53:08 6...66 666.6 666.6 66. 666.6 666.6 66. .66.66 666.66 66. 66. . . 666 1 6. 666 1 6. 1 1 1 866...... 666 .6 666 1 6. 666 1 6. 666.6 1 6. 83.66.. 666.6 666 1 6. 666 1 6. 666.6 1 66 6660 .6666 666.6 666 1 6 666 1 6 666.6 1 6. 666666 666 . 6. .66 1 6. 666 1 6. 666 . .. 1 66 6.6.66... 666.6. 666.6 1 66 666.6 1 66 666 .6 1 66 8.6. 666.66 1 666.6 1 1 666.6 1 1 .66.66 1 666.668 66.8.6 666 . .. 1 666 .. 1 1 .66. . 1 1 666 . 6 1 6.6. .65 .65 .65 .66. ..65 .65 .66. .65 .65 .66. .30... 66.50 6:26.50 6:26.50 66.50 tam—no.0 9.6.6.50 66.50 tam—no.0 tam—6.50 6.0.0 5.5 66.2 o>...u< o>$u< ~38... Ho uCOqum «o HCUUCOQ mo acouhvm 66 66 .6 <6... 3.6.8.535 .6.66861 .6.... 6....66 153 Table A-10. -- (cont'd.) F1 F2 Percent of Percent of Active Active Map Unit Crop Cropland Cropland Crops Cropland Cropland Crops Total (96) (Ha.) ( Ha.) (96) (Ha.) (Ha.) (Ha.) Idle - 339 -- - 339 -- 678 Active Cropland - l, 354 -- -- I, 353 -- 2 , 707 Rice 10 - 135 80 - 1,082 1,217 Tobacco 20 - 270 - -- -- 270 Plantain 50 - 677 10 - 135 812 Pasture 20 - 272 10 — 136 408 100 1,693 1,354 100 1,692 1,353 6,092 (continued) 154 mm. a» ..w 11 an ..n 1.. 1 n. m 1 1 m 1 1 1 l n . o. 6:66m n. m .1 1 w 1 1 1 1 n . o. 83306. .866 n. m 1 1 6 1 1 1 1 n . o. 53:6... ..N n . 1 1 m 1 1 1 I n . 6. 500 mm m. 1 1 . . 1 1 1 .1 o . .N 68.; 1 1 1 1 1 1 1 1 1 66 666... 6366... 0.. ..N 1 1 w . 1 1 .1 1 on 8808.35 ..6... ..6... .66. .6.. .66. ..6... .6.. ..6... ..6... .6... 666.... 66.... 58.6.. 666.... 66.... 58.6.. 666... 66.... 58.6.. no.0 6.62.62. no.0 6.62.96. no.0 6.62.5. 66.635... .0 8.55365 666.6 .66 .66.. 66. 666.. 666.. 66. 666.6 666.6 66. 666 . 6 666 1 66 6.6 1 66 666. . 1 66 6.6666... mm . 6w 1 o. ..6 l n .... 1 1 6800 mm . ..w 1 o. ..6 1 n 1 1 1 8.... 66 . .6 1 1 1 666 1 66 666 . . 1 66 6.366.. 666.6 1 .66 1 1 666.. 1 1 666.6 1 666.680 66..66. mm. . . 1 SN 1 1 mom 1 1 $6 1 6..6. .65 .65 ..65 .66. .65 .65 .66. .65 .65 .66. .30... 68.0 2.6.8.0 2.6.8.0 .850 2.6.8.0 9.6.8.0 68.0 9.6.8.0 5.6.8.0 no.0 .....D 8.2 638.6. 638... 638... .o 58.6.. .o 58.6.. .o 58.6.. m 0 N0 .0 <6... 36:85.35 66..66... 1 .6.-.. 6.66.6 155 90:558. .666. 66.2 5 5.600085... 2.05680 0.. 0.500 0.... 5 36.5559... .60..6> o6...0 .o 50508:. co 866.. . .8502 ... 850.2 .3 060. ...n< .360082 .6853. 46.3. 8.063.555... .o 8.520 6.3.8.5... 0.0 8360 on 5.30.08 .0058 .66 66.6.6.3 .. ... ...... 6.66. ..66 6.. .6338... 6N00.0 n 60.3 . “5.6.8.... 66 n .35.... . 6 .6.. 6.666 0w...m 1 .. ...... 0.60. .56 06.603 1 . “0.300... .0n 6.6.0... 56 380 .6352... 66..6 6..66 66.6. 66.86. ....z 66.66 66.66 66.66 6.686. .680 66.66 66.66 66.66 .6.. 66.66. .60.... 66.6.. 06 .0 06... 06.6 .603 25> 56.03.. 66.66 66.66 66.66 .80 .68.. 66.6 66.6 66.6 8:60.. 66.6 66.6 66.6 66.68.56.856: 06.0 06.0 06.0 .05....3 .o co..60..nn< 66.. 66.. 1 6.60.6666 .6 62.666666 66.6 66.6 66.6 .6. 66.6666. 66.6 66.6 66.6 6.666660 666.0 66.6 66.6 66.6 62.66.... 66.6. 66.6. 66.6. .86.. 66.6 66.6 66.6 ..6. .6.66. 6..6 1 1 66.2.... 06.0 .1 1 .000. .3505. N00 66.0 .1 55062.2. 050.3865 66.. 66.. 66.. 6.6.36 65.6666... 66.6 66.. 1 6.66-6. .66.....6... 66.6 .. 1 6-6-6 66.6 66.6 1 66-.. 66.6 86.0.6.6: 66.6 III 66.6 66.6 .6.66 6..6... .6666 06 .. . wwd mm... 32.5 06.0. 06.0. 16N.6 8.66.68... .56.. ... 8:66.... .. 8:56.. 6... 8:66... 5.6.8.. 355655352860... 8.56... ..6. 55.028... .0 6600 00.x 1 .. .1< 0.6.6... 156 Table A-1 l. -- Tobacco Annual Cost of Production, By Method of Production* (Pesos/Tarea) ACTIVITY METHOD I METHOD 11 Land Preparation 2.00 2.00 Inputs 5.35 5.81 Plants 5.00 5.00 Insecticide (Azodrin) - 0.46 Water tax 0.35 0.35 Labor 22.67 25.97 Cleaning drains 0.50 0.50 Planting 3.00 3.00 Application of insecticides - 0.30 Weeding 1.50 3.50 Pruning 3.00 4.00 Harvesting operations 14.07 14.07 Irrigation 0.60 0.60 Total Cost 30.02 33.78 Average Yield (qq) 1.50 2.00 Farm Price ($/ q) 25.00 25.00 Gross Income (g) 37.50 50.00 Net Income ($) 7.48 16".2—2 *Annual costs and yields per hectare are: I - $480.32 and 24 qq.; II - $540.48 and 32 qq. (continued) 157 0635.80. 6.2:... 6.66 6:6 66.6666 - E “22:5 6.66 6:6 66.6666 - = 52:5 6.6 6:6 6666.6- . 8.6 2306; .66 6.26.. 0:6 3600 .35... A9605. 000.. 36:5 5 06666.98 6. 0.6; $133558 6:00 6. 0566656.. 0:6 0556.6. 6..66.. 0N 0.6.66 .650 6:00 2.6.560 56555336.... 66.6.5606 65565.65 0:6 565253366 506. 66030:... E E 1'66 . 6. 658:. .62 00.00 00.60 00.6N 65o05 660.0 00.06 00.06 00.06 16.253. 65.6. 5.6.... 00.~ III 00.. III 00.0 262.5. 0.6; 606.6>< 06.06 3.00 06 . .. .600 .396 66.6 66.6 66.6 66:66.... n . .. 00.. 06.0 :o..06..ou 0:6 056656: 00 . N 00 .. I 05:36. 000 06.. 00.6 .6.. 6.60.5620 .o :o..60..6.6.< 00. .. 00.0 0. .0 .3 050663 00. . 06 .0 6 6.6 0556.66,». on .0 on .0 1 0556.6. 0:6 66.2. 050.62 0~.0 0~.0 .1 656.6. .0 5.606.565 06.0 06.0 0~.0 656.6. 055:6. on .0 06 .0 1 656.0 05:66.0 06.0w n0..~ 06.0 .0061. mm .0 mm .0 1 .363 5.660.... n 6 . 0 n 6 . 0 11 35360.0. .8500 00.0 00.0 11 26.05 56.6c0< mm . 0 0N .0 .1 ..o .6.3.:0..0< n 0 . 0 n 6 . 0 1 60.0.05... 00.0 00.6 1 .010N103 .66.....66. on .m 06 .. 1. 60.06562 0N.. 0~.. 00 .0 .6.6.56; .6 9.065. 656.6. 00 .6. 06.0 00.0 3:05 06.6 66.6 66 .. 8.66.666... 066.. ... DOIHME .. 001.52 .2. 00:60.2 >....>....U< .66.6...\6om66. . #:5603096. .o .600 .6:::< 5656.6. .1 .. .1< 6.6.6... 158 Table A-ll. -- Banana Annual Cost of Production, by Method of Production* (Pesos/Tarea) ACTIVITY METHOD 1 METHOD II Land Preparation 1.50 3.00 Inputs 0. 60 14. 89 Plants - 1.20 Nemacide - 3.50 Fertilizer (16-28-0) -- 8.00 Fungicide -- 0.95 Agricultural oil - 0.38 Adherant (Agrel) - 0.06 Control (Sigatoka) - 0.45 Irrigation water - 0.35 Labor 9.40 23.15 Cleaning drains - 0.50 Pruning plants 0.20 ‘ 0.40 Disinfection of plants - 0.20 Making holes and planting 3.25 6.50 Replanting 0.10 0.50 Weeding (3) 4.50 9.00 Application of chemicals - 2.05 Pruning 0.75 1.50 Harvesting and collection 0.60 1.15 Irrigating - 0.60 Propping - 0.75 Total Cost 11.50 41.04 Average Yield (stalk) 60.00 144.00 Farm Price ($lstalk) 1.50 1.50 Gross Income 90.00 216.00 Net Income 78.50 174.96 *Annual costs and yelds per hectare are: I - $184.00 and 960 stalks; II - $656.64 and 2,304 stalks. (continued) 159 Table A-1 1. -- Cane Cost of Production - Esperanza* Pesos/ Ton Pesos/Hectare Cane Cultivation 1. 85 146 .15 Harvesting 2.22 175.38 Maintenance 0 . 50 39 . 50 Field irrigation overhead 1.04 82.16 Field transport 1 10 94.01 Depreciation—transport 0 .12 9 . 48 6.92 546.68 *Per hectare yield is 79 metric tons. Source: Bookers Agricultural and Technical Services Ltd. CONFIDENTIAL REPORT, Appendix I, p. V 3.2 Estadio de Rehabilitacion y Expansion de la Industria Azucorera 1975. Corn Annual Cost of Production, by Method of Production* (Pesos/Tarea) ACTIVITY METHOD I METHOD 11 Land Preparation 3.00 4.50 Inputs 0.30 1.17 Seed (Frances Large) 0.30 0.30 Insecticide (Dipterez) - 0.54 Fertilizer - 0.33 Labor 4.15 7.12 Planting 0.80 0.80 Weeding 1.75 3.50 Application of insecticides " and fertilizer - 0.82 Harvest and sacking 1.60 2.00 Total Cost 7.45 12.75 Average Yield (qq) 2.00 2.90 Farm Price ($/qq) 6.02 6.02 Gross Income 12.04 17.45 Net Income 4.59 4.71 *Annual costs and yields per hectare are: I - $119.20 and 32 qq.; II - $204.00 and 46.4 qq. (continued) 160 Table A-11. -- Red Beans Annual Cost of Production, by Method of Production* (Pesos/Tarea) ACTIVITY METHOD 1 METHOD I1 Land Preparation 2.00 2.00 Inputs 4 . 00 4 . 83 Seed 4.00 4.00 Babozida Karmeta - 0.45 Insecticide (Dipterez) - 0.38 Labor 6.60 9.75 Planting 1.85 1.85 Weeding 1.75 3.50 Pesticide application - 0.40 Harvest 3.00 4.00 Total Cost 12.60 16.58 Average Yield (qq) 1.50 2.00 Farm Price ($/qq) 27.03 27.08 Gross Income 40.62 54.16 Net Income 28.02 37 58 *Annual costs and yields per hectare are: I - $201.60 and 24 qq.; II - $265.28 and 32 qq. Yuca Annual Cost of Production by Method of Production* (Pesos/Tarea) ACTIVITY METHOD I METHOD 11 Land Preparation ** 2.00 4.00 Inputs - 0.66 Insecticides (Endrin) -- 0.23 Insecticides (Azodrin) —- 0.43 Labor 7.60 10.10 Planting 0.60 0.60 Weeding*** 2.00 3.00 Application of Chemicals — 0.50 Harvest 5.00 6.00 Total Cost 9.60 14.76 Average Yield (qq) 9.00 12.00 Farm Price ($/qq) 6.00 6.00 Gross Income 54.00 72.00 Net Income 44.40 57.24 *Annual costs and yields per hectare: I - $153.60 and 144 qq.; II - $236.16 and 192 qq. **Cutting of past crops in 1; cutting and harrow in II. ***21n1, 3 in 11. (continued) 161 Table A-11. -- Sweet Potatoes Annual Cost of Production, by Method of Production* (Pesos/Tarea) ACTIVITY METHOD I METHOD 11 Land Preparation 2.00 3.00 Inputs -- 0.40 Insecticide (Trithion) - 0.40 Labor 10.00 12.20 Planting 2.00 2.00 Weeding and tilling 3.00 4.00 Insecticide - 0.20 Harvesting and on-farm operations 5.00 6.00 Total Cost 12.00 15.60 Average Yield (qq) 8.00 15.00 Farm Price ($/qq) 4.00 4.00 Gross Income 32.00 60.00 Net Income 20.00 44.40 *Annual costs and yields per hectare are: I - $192.00 and 128 qq.; II - 249.60 and 240 qq. (continued) 162 Table A-1 l. - Cocoa Annual Cost of Production, by Production Method* (Pesos/Tarea) ACTIVITY METHOD 1 METHOD II First Year -- 22.30 Clearing brush and trees - 12.00 Fertilizer (10 lbs) - 0.80 Plants (40) -- 4.00 Land preparation for planting -- 3.30 Fertilizer application (2) - 1.00 Disease control -- 1.20 Second Year -- 15.00 Selective thinning - 3.50 Chapeo (3) - 6.00 Fertilizer (20 lbs) - 1.60 Disease control - 1.00 Fertilizer application (2) - 1.00 Reseeding (9 plants) - 1.40 Reseeding new shade - 0.50 Third-Fifteenth Year Rate 13.90 13.90 Weed control 7.50 7.50 Thinning 2.00 2.00 Fertilizer (30 lbs) 2.40 2.40 Fertilizer application (2) 1.00 1.00 Disease control 1.00 1.00 Total Costs 13.90 51.25 Average Annual Costsa 13.90 14.53 Average Yield (qq) 0.5 0.70 Farm Price ($/qq) 91.75 91.75 Gross Income 45.88 64.23 Net Income 31.98 49.70 *Annual costs and yields per hectare: I - $222.40 and 8.0 qq.; II - $232.48 and 11.2 qq. aDoes not include interest on capital. 163 Table A-ll. -— lnterplanted Annual Cost of Production (Pesos/Tarea) Crop Cost When Planted Percent of Cost When Alonea Tarea When lnterplanted ($) Inter lanted ($) 96) Beans 12.60 37.5 4.73 Plantain 11.50 20.0 2.30 Corn 7.45 15.0 1.12 Yucca 9.60 12.5 1.20 Sweet Potatoes 12.00 10.0 1.20 Tobacco 30.02 5.0 1.50 Total 100.0 12.05b a"Low" production method only. bThe cost/hectare is $192.80. 164 Table A-12. -- Distribution of Production Methods by Intermediate Map Unit (96 of Crop Area) Crop Production Intermediate Map Unitc Method ABG CF DE Rice Low 20 25 50 Medium 40 40 25 High i0. 2 25. 100 100 100 Tobacco Low 30 30 30 Medium .29 _79. _79 100 100 g 100 Plantain Low a b b Medium 60 60 60 High 10 _49 _99 100 100 100 Banana Low 40 40 b Medium .99. .99. .199. 100 100 1 0 Sugar Cane Low 100 100 100 Corn Low 30 30 E Medium _7_0_ 7_0 100 100 Beans Low 30 30 I; Medium _7_0_ _70_ 100 100 Yucca Low 30 30 E Medium lO_ _Z_Q_ 100 100 Sweet Potatoes Low 30 30 3 Medium _70 _79_ 100 1 0 Cocoa Low 80 E I; Medium _2_0_ 100 aPlantain low is only used when intercropped with other crops. bNot found in these map units. CA, B, and G roughly correspond to the East region, C and F to the Central, and D and E to the west. 165 ... <00 3606.00... .0 6.60 m. .../6.06. 060360 6 .6:0...006 :6 600.05 623 266 6.266586. .656 6 6:6. 2 .2 <6. 86.665666. .6 :66 .... .3656 6 666. :6 <6: 62.3606 666.62 65.6650: 65.6650: 65.6650: 5.5.60 on 00 on 5.63360 360 3.0.06 .065\50..6 5.0.06 .065 3.0.06 .065 5.6063. 6:0: 6:0: 6:0: 05000.". 6.03.605. 6.03.605. 603.605. 60656.0 1 1 1 06.06060 .36? 6.6.6..6>< 30.6 56> 30.... 56> 30... 36> ......066560 50.650: 533:0: 533:0: 6665:30 66. 66. 66. .56. 66.666 6. £660 .2 ... 1 6.056... 603.00 6-6 6-6 6.6 .66. 666.6 3.600600... 0.6.00.6? 3.60.6600... 0.6.00.6? 3.60.0600... 56:00500 .60.05.0 2.00 .U 0. 6.336056... .6:::< 606.6>< .55. ..65.6~. .6:::< 606.63.. 603.3 603.3 603.3 6060656.. :0 5.3.600 1 66 66 .66. :6: 66.2 .6 80.666... S... .3560 .. 3606.00... :6 6666.60 2 6.6.6653... 6<66 £66 66:3 666.0 .6 2606:6660 6.6266 1 .6.... 636.6 166 Table A-14. - Transportation Costs, by Region (Pesos/ Unit of Output)8 Supply Requirements Regions Regions East Central West Santo Domingo East 0.00 0.05 0.25 0.45 Central 0.05 0.00 0.20 0.50 West 0.25 0.20 0.00 0.70 aThe tariff structure on which these costs were developed is based on truck transport; cargo weight is in quintales. For plantains which are measured in "fingers" (thousands of fingers are "millars") and bananas which are measured in stalks, no specific tariffs could be found. These tariffs were therefore arbitrarily applied to the units of production used in this study for plantain (millar) and for banana (stalk). Table A-15. -- Benchmark Crop Requirementsa Crop Unit Requirements Rice qq. 2,700,000 Tobacco qq. 465,000 Plantain millar 400,000 Banana stalk 3,350,000 Sugar Cane mt. 300,000 Corn qq. 460,000 Beans qq . 250,000 Yucca qq. 1,775,000 Sweet Potato qq. 1,130,000 Cocoa qq. 40,000 aBased on current cropping pattern and yields. 167 Table A-l6. -- Percentage Distribution of Crop Requirements (96 of Total Production) Crop West Central East Santo Domingo Total Rice 10 30 20 40 100 Tobacco 0 100 0 0 100 Sugar Cane 50 0 0 50 100 Corn 5 20 5 70 100 Plantain 10 18 28 44 100 Banana 10 18 28 44 100 Beans 10 18 28 44 100 Yucca 10 18 28 44 100 Sweet Potatoes 10 18 28 44 100 Cocoa 2 3 5 90 100 Table A-17. -- Percentage Distribution of Crop Requirements, Assuming Population Shifts (96 of Total Production) Crop West Central East Santo Domingo Total Rice 5 30 15 50 100 Tobacco 0 100 0 0 100 Sugar Cane 50 0 0 50 100 Corn 4 20 4 72 100 Plantain 8 17 26 49 100 Banana 8 17 26 49 100 Beans 8 17 26 49 100 Yucca 8 17 26 49 100 Sweet Potatoes 8 17 26 49 100 Cocoa 2 3 1 5 90 100 168 Table A-18. -- Allocators Of Detailed PPAsa to Intermediate PPAs, Aggregate PPAs, and Regions Detailed Portion of Detailed PPA to: Int. PPA Agg. PPA Region (96) No. (96) No. (96) No. 011 100.0 A1 100.0 11 100.0 E 012 100.0 A1 100.0 11 100.0 E 013 100.0 A1 100.0 11 100.0 E 021 100.0 A1 100.0 11 100.0 E 031 100.0 A1 100.0 11 100.0 E 041 100.0 A2 100.0 12 100.0 E 042 100.0 A1 100.0 11 100.0 E 051 31.4 A2 31.4 12 31.4 E 051 68.6 B1 68.6 111 68.6 E 061 100.0 Bl 100.0 111 92.7 E 7.3 C 071 100.0 B1 100.0 111 90.3 E 9.7 C 072 100.0 B1 100.0 113 90.3 E 9.7 C 081 18.7 A2 18.7 12 18.7 E 81.3 B1 81.3 III 81.3 E 091 100.0 Bl 100.0 111 90.0 E 10.0 C 092 100.0 B2 100.0 112 90.0 E 10.0 C 101 100.0 A3 100.0 13 100.0 E 111 100.0 E3 100.0 1113 100.0 W 121 100.0 Bl 100.0 112 100.0 W 131 100.0 C1 100.0 112 100.0 C 141 100.0 52 100.0 1112 97.6 W 2.4 C 151 100.0 Bl 100.0 112 100.0 C 161 100.0 C1 100.0 112 18.1 W 81.9 C 171 100.0 C1 100.0 112 45.0 E 55.0 C 181 100.0 Cl 100.0 112 45.5 W 54.5 C 191 100.0 C1 100.0 112 3.3 W 96.7 C 201 100.0 Cl 100.0 112 22.6 W 77.4 C 211 100.0 E3 100.0 1113 100.0 W 221 100.0 D1 100.0 1113 100.0 W 231 100.0 E3 100.0 1113 100.0 W 232 100.0 E3 100.0 1113 100.0 W See footnotes at end of table (continued) 169 Table A-18. -- (cont'd.) Detailed Portion of Detailed PPA to: Int. PPA Agg. PPA Region (96) No. (96) No. (96) No. 241 100.0 E3 100.0 1113 100.0 W 242 100.0 53 100.0 1113 100.0 W 251 100.0 E3 100.0 1113 100.0 W 261 100.0 E3 100.0 1113 100.0 W 262 100.0 El 100.0 1111 100.0 W 271 100.0 E1 100.0 1111 100.0 W 272 100.0 El 100.0 1111 100.0 W 281 100.0 El 100.0 1111 100.0 W 282 100.0 E1 100.0 1111 100.0 W 291 100.0 E1 100.0 1111 100.0 W 292 100.0 E3 100.0 1113 100.0 W 301 100.0 F1 100.0 113 100.0 C 302 100.0 F2 100.0 112 100.0 C 311 100.0 G1 50.0 13 87.8 W 50.0 113 12.2 C 312 100.0 G2 50.0 13 87.8 W 50.0 113 12.2 C 313 100.0 G3 50.0 13 87.8 W 50.0 113 12.2 C aDetailed PPA measured areas have been expanded (or contracted) to be consistent with the measured areas of Intermediate PPAs. 170 Table A-19. -- Allocators of Aggregate PPAs to Regions (96) Aggregate L PPA East Central West Total 11 100.0 -- -- 100.0 12 100.0 - -- 100.0 13 70.0 - 30.0 100.0 111 99.2 0.8 -- 100.0 112 3.9 80.6 15.5 100.0 113 - 65.3 34.7 100.0 1111 - -- 100.0 100.0 1112 - 2.4 97.6 100.0 1113 - - 100.0 100.0 Table A-20. -- Cartographic Relationship Between Map Unitsa Intermediate Map Units Detailed Map Units A 1-5,8,10,fl B 3,5-9,12,15,1_,2_0,_1_ c 13,16-20 D 22,21 E 11, 14, _1__,_1_9,_0, 21,323-29 F 30 c 31 Aggregate Map Units 1 1-5, 8, 10, 31 II 2, g, 5-9, 12, 13, 15-20, 30, 31 III 11, 111, _1_o_, lg, 19,20, 21, 2_2, 23-29 aDetailed Map Units not taxonomically in the more generalized Map Units are underlined. 171 Table A-21. - Comparison of Measured and Summed Map Unit Areas Map Unit Area Difference Intermediate Detailed Percent of Intermediate Measured Summeda Area Measured Area Map Unit (sz) (sz) (sz) (96) A 282.2585 238.7384 43.5201 15.4 B 1008.8568 955.5242 53.3326 5.3 C 242.7887 253.8351 -11.0464 - 4.5 D 23.9850 19.7731 4.2119 17.6 E 571.1855 569.2122 1.9733 0.3 F 33.8481 41.4802 - 7.6321 -22.5 G 58.4022 100.9671 -42.5649 -72.9 Total 2221.3248 2179.5303 41.7945 1.9 Map Unit Area Difference Aggregate Detailed Percent of Aggregate Measured Summeda Area Measured Area Map Unit (sz) (sz) (sz) (96) 1 302.6215 239.7384 62.8831 20.8 11 1526.1653 1350.4964 175.6689 11.5 111 457.5544 5_8____8. 9853 -131.4309 -28.7 Total 2286.3412 2179. '22015 107.1211 4.7 0.3102 2179. 5303 aSums of Detailed map delineations to the Aggregate level. bError accounted for by cartographic errors. Intermediate level and to the 172 Table A-22. -- Comparison of First and Second Measurements of Planimetered Areas Difference Delineation Area as Area as Percent of Number First Measured Second Measured Area First Measured Area (sz) (sz) (sz) (96 Detailed 26 49.9989 49.8974 .1015 -0.20 22 9.8580 9.8740 .0160 0.16 20 15.1716 14.9955 .1761 -l.16 21 1.2472 1.2258 .0214 -1.72 11 2.1532 2.1383 .0149 -0.69 2 0.4368 0.4316 .0052 -1.19 31 4. 3720 4. 4520 . 0800 1. 83 23 5. 7583 5.7549 .0034 -0.06 27 15.4717 15.6283 .1566 1.01 6 3. 8552 3.9001 .0449 1.17 31 8.5464 8.5185 .0279 -0.33 10 21.6482 21.7365 .0883 0.41 20 13.1632 12.9723 .1909 -1.45 17 13.1560 13.0450 .1110 -0.84 6 49.0721 49.1864 .1143 0.23 6 16.1515 16.2016 .0501 0.31 Average of Absolute Differences - - - 0.80 Intermediate C 5. 6922 5. 7352 . 0430 0. 76 A 59.1494 57.1212 2.0282 -3.43 E 9.3119 9.2575 .0544 -0.58 G 15.7029 15.8036 .1007 0.64 B 6.6525 6.6707 .0154 0.23 F 8.6361 8.6304 .0057 -0.07 B 86.2626 86.7439 .4813 0.56 Average of Absolute Differences - - - 0.90 Aggregate 111 211.5547 211.6197 .0650 - .03 111 205.1875 205.3168 .1293 - .06 11 37.6836 37.7212 .0376 - .10 111 4.7692 4.7697 .0005 - .01 II 474.0572 464.5413 9.5159 2.01 11 38.7266 40.2757 1.5491 -4.00 11 265.1561 268.0198 2.8637 -1.08 Average of Absolute Differences - - - 1.04 aMap delineations were planimetered on the 1:50,000 base map which was in fourteen map sheets. Table A-23. -- Estimated Costs of Three Cibao Valley Soil Inventories 173 Detailed Intermediate Aggregate Assumptions: Area to Inventory (ha.) 250,000 250,000 250,000 Work Days/Year 250 240 240 Area/Soil Scientists/Year (ha.) 25,000 100,000 400,000 Scale of Photography l:20,000 1:60,000 1:70,000 Scale of Topographic Map 1:50,000 1:250,000 1:250,000 No. of Photos to Cover the Area 275 70 Field Time (months) 8 8 8 Time to Publication (months) 12 12 12 Published Scale 1:100,000 1:250,000 1:1,000,000 Budget: Soil Scientists ($1000/mo.) 120,000 (10) 36,000 (3) 7,200 (0.6) Soil Scientists ($ 440/mo.) 52,800 (10) 14,400 (3) 3,168 (0.6) Drivers (5 350/mo.) 42,000 (10) 12,600 (3) 2,520 (0.6) Laborers ($ 200/mo.) 24,000 (10) - 1,440 (0.6) Laborers ($ 150/mo.) 18,000 (10) 5,400 (3) 1,080 (0.6) Secretaries ($ 400/mo.) 9,600 ( 2) 9,600 (2) 2,880 (0.6) Draftsman (S 400/mo.) 4,800 ( 1) 4,800 (1) 2,880 (0.6) 271,200 90,000 21,168 Materials and Equipment:a Aerial Photos ($6 each) 1,650 (275) 420 (70) 180 (30) Topographic Maps 18 ( 6) 25 ( 1) 25 ( l) Stereoscopes of Mirrors ($700 each) 700 ( 10) 210 ( 3) 70 ( 1) Abney Levels ($40 each) 40 ( 10) 12 ( 3) 4 ( 1) Pocket Stereoscopes ($35 each) 35 ( 10) 11 ( 3) 4 ( 1) Altimeters ($150 each) 150 ( 10) 45 (3) 15(1) Munsell Cards ($60 each) 600 ( 10) 180 ( 3) 60 ( 1) Posthole Diggers ($35 each) 350 ( 10) 105 ( 3) 35 ( 1) Picks ($7 each) 70 ( 10) 21 ( 3) 7 ( 1) Posts ($6 each) 60 ( 10) 18 ( 3) 6 ( 1) Knives for Soil Scientists ($25 each) 250 ( 10) 75 ( 3) 25 ( 1) Cartons (Balsos) ($25 each) 250 ( 10) 75 ( 3) 25 ( 1) Bags for Samples 50 25 15 Other (496 of Materials/ Equipment) 200 _29 _20 4,423 1,272 491 Rental of Vehicles 60,000 18,000 3,600 Operation and Maintenance 10,000 3,000 _6_0_2 70,000 21,000 4,200 See footnotes at end of table (continued) 174 Table A-23. -- (cont'd.) Detailed Intermediate Aggregate Analysis of Samples ($15 each) 150,000 33,750 20,250b (10,000) (2,250) (2,250) Report Publication (2096 of Costs) 74,3439 29,204 9,222 Total Cost ($) 569,900 175,300 55,400 Cost/Hectare ($/ha.) 2.28 0.70 0.22 Source: Personal communication from soil scientist, Jose Ogando, project SIEDRA, May 17, 1979, as based on Clasificacion de Suelos para Proyectos de Desarrollo de las Tierras de la Republica Dominicana, FAO, Rome, 1977. aStereoscopes (estereoscopio de espejos) costing $700 each with useful life of 10 years; 10 Abney levels (niveles Abney) costing $40 each with 10 year life; 10 pocket Stereoscopes (estereoscopio de bolsillo) costing $35 each with 10-year life; 10 altimeters costing $150 each with 10 year life. b$9 each. C1596 PLEASE NOTE: Page 175 is missing in number only per author. No text is missing. Filmed as received. UNIVERSITY MICROFILMS INTERNATIONAL. 176 ooann noma— son.~ ..6666 0...6. 61.66. 0.000 0.660 06mm. 0.00 06- 06.: 0600 @600 36.7.4 ..0NN 010- )onro ~o~n 060.0 oomhn .6... .1... .100: oowwa 06$ rot¢ ..ofix a. c. 01000 0.0.. QoVn Open 0.00 01.06. a..nc 010:0 ~60~mn nonn— OoOnN nooqfle 0..~¢. 00:00. oonac 0.000 ~60- econ 0630. oo- N.NO ..0h .oo~ 6..66. 6..66 6.6.6 6.65. no?! 060: 0.00. cone. 6.666 6.666 ..n 6.6. 6.66 06.N. 06.4. 0.00n. morcn 0.066 0.000 0.00. ..~.~ toflufl! p.000. p.000. 06m00 0.000 no.0. 0100 00- 0.00 monoNu 060- 000.0 noocw .600 heNt 6.500 «p.00 .000 .600 6.666 6.666 06- a. comp. aon~u N.NN N0N~ cup—m. 0.000. ¢.~m~ 4¢~0p03n 002.!00 th40000 0010 11666.11 .6.. 666.266 .646 .6.) 6.6666666 11066.11 11600.11 :0: 4.6300060 1 6.600: 06.50.63 6.606.00< 0:6 .06.:0663 606.063.660. .06..6.60 6;. .0 630.0 :0..6..006:6.H 00.0 0663060 660:6.60060 ll .6N1< Gun-h 177 ’. N.on Noam 0.: 0.: .003 m.n N. 0.0N. 0.0N. a..0N m..0~ 0.¢. 00:. 11666.11 .<¢.zw. a. (A. .13.... n..m 0.Cfl 0.. N.¢.) ~..)¢ a... mo~0 0.0.n n...m. 0..n. n..~. 0.0.. 0.~:¢ 3..nw 3.3.. a..~ o..n 3.0. .Ad. 0.n.. 0.m.. Oonw o.fl~ o.~ .wyx 0.~¢~ 0.~¢~ «.Oa c.0m 11:6o. . 44.0.00m 002.200 0.2¢w .mw: 418.200 .mcw .00. w2<00 4.<.w0 00.20.w3 w¢< w04w.> w20.0w¢ .2w2w¢.00w¢ 5080 A.v.u6000 II .QNI< 0.0.9 178 0.OJ.N o.~0 n.30V 0..O€N c.acn¢ m.¢nm: n.n.. ..n N.AVON 0.0NOV n.c ..a.00. n.0nn0 N.Noo. #..N.m O..oo.. ..n.0~ 0.0..0 ..Joe o.MNo~m 0..~.a. 0.0.0. r..qcm. 0.4001004 ..9.0N ..0« 0.Vfi¢ ¢.J.NN 0.q00¢ m..7mt ..00. 0.0 v.0wow 0.0~0~ m.¢ O..n3.. 0o0¢-. ..n0n. ..3000 0.30.0. ..a.0~ o...0. n.0«m 0.000NM fi.-~m. e.«.0. N.¢n0m. ulnanum 000.!I 0.4—002:0.2. .00) 0.?ch 6.5: 0.939 aorn.~ 3.~05¢ o..:fi¢ ..:3. :.a m.c~o~ 6..~e~ ...Ncc. ~.¢..0 n...a. 0.0V0¢ o..000. n.00m~ ?.~.- 0.5M. 8.0.6.6 h.VmDfi. mom3o. 00000:. 0.4.4.00 0.V00 ¢.¢. o.0¢ 0..?¢ ..mon 0.... m.m. 0.onn¢ ..nflmc 0.00: c.00N ..o¢ m.m- 0.0nm $.00 n..fl¢ 0.0m 0.0ewq ~.¢.c. 6.... c.9no. 0.4001034 0.~nn ..o 0.90 :..c¢ 0.00: 7..nn ..N. n.0.n¢ n.c.n¢ 0.m 0.00o 0..0N n.5n 0.~m. 0.030 ~.00. m..0¢ m.0¢ 0.0:wn ~.nm¢. ..o.. ..0.0. It»..20 000.3! 0.4.00Ixu.2. 20..4fi0010 0.~rn Ooh no.0. ..fl¢¢ ...Jfl m.nnfl 0.~. ..0Nfi¢ :.o..¢ m.¢ 0.000 ~.0.N n..o m.n0~ Gonna 0.0.. n.¢.¢ 0.9N 0.0¢~n 0..¢o. m..n. ..o.o. 004.4.00 0.nnom. m.no¢ 0.0.0. 0.0.0n. 0..n~0 n.0000 n..0~ «.0 0.0300 «..00n 0.0 0.5000N c.000c. o.00.~ 0.0.00 m.~..- nonnon o.¢a¢.. ~.¢.. ..0maon 0.0o¢¢. 0.noo. «.mwon. 0.4000004 ~.n¢00. 0.0.. Q.0N.~ 7..c.~. 0..n~0 0.qn00 m..o. n.~. 0.0:om «..00n n.~0o0~ ..~.¢0~ 0.0oow 0..omd m.0~00~ ..0000 0.0000. m.~n0. ...noow 0.0000. 0.000. 0.0000. II004—00III 0.4—00210.2. 40x4 o.~0.m. o...n nonmaN c..nc~. 0.0n~0 O.Nn00 m..0. m.N. ..Oooe m..00n Nun. NohwoeN ..0V0n. 0.000N ooam~0 00.000N coomnt 0.000e. ..NOO m..0.0~ n.00ec. 0.0.0. ..n.0~. 004.4.00 44.0.0Dm .mwx 440.200 .040 .00. 2:00 44.0.0Jm .nw: 440.200 .040 ..x. 0240 14000 44.0.000 .003 441.200 .040 .244.m. 424240 44.0.000 .003 440.200 .040 .0444.2. 2.4.2440 44.0.000 .mwl 440.200 .040 .00. 000400. 44.0.000 .003 440.200 .040 .00. 00.1 4.4.00 00.10.03 044 0040.» 20.000l0080 63¢ gauche I c.0000 00.00.03 oua00u00< new .vuunw.03 uuw.v~lumu:. .vo..au00 wzu 00 mumoo can .co.uu:voum .o0u4 coooruum ouauuouu.0 II .n~.4 0.019 179 .99.. IO... .0... l0... n..rwn 0.0.. N.c0n 0.0000 0.000V 0..0 0..0. ..300. m.a~.~ a... o.oq~ ...... 3.omon ..n.. n.non 0.30:4 ..0.«~ 0.00 0.00. 0.00.N 0.4000334 00000 006.0 0.9.. .0090 2.?00: a... :..nr ..raaa >.>na~ o.m. ...e. a.~am. 3.00.0 0.0N 0.00N N.0N0. 0.0000 0..0 h...0 0.0000 0.00NN ..0N 0.00V 0..~0. lannum 000.ll 4.4.00f14_z. .nau .00.. 00.91 060.0 60).. 3.3..fi ..:V —..II ....J..0 0.333% ..N. 3.0}. JoNrnd c.0n0~ 0.0MN a.me~ 0.V00. 0.0000 ~.Vn 0.000 0.00.0 0.04.4.0: 0.00? 0010' 0.0.0 60).. 0.... .5... 00.00 00.00 N.N0 0.0 0..0 0.000. 0.00 ..00. 0..... 0.00.~ 0.00 0.0.N 0.N00. 0.000 N... ....u 4 .400 1.394 000.9 00.00 009.0 .000. .06.. .040. .0000 .00.. o.~¢ 3.. ...n a.onm. ..n. ~.¢o4 o.m._. 0.0n.~ 0.0N N.Nom 0.000. 0.000 0.~m 0.0JN II0..ZD 000.!0 0.4.04t00.z. 23.—430300 .69.. 0.9.0 0.... .0... 600.0 09.00 0.0.. .0000 fiofiQ ~. 0.0 0..“ 0.onn. 0.30. ..«.N fi..00. 0.000 0.. m..0 ...mw 004.4.00 0.00.0. 0..«m. 0..nm n..o.0 0.000.~ 0.00. 0.000. n.n.QnN 0.4.00 0.00. 0.0.0 0..000 0on0.0 ..00N 0.000. 0.000. 0..000. 0.000 n..o.. ~.00oo. 0.0000 0.00. m...0 c.0000 0.4000334 0.002n. 0..0na 0.0m: 0.0:00 ..0000N 0.N- 0.~00~ 0..000~ 0.V.n0 0.00 ~.0~0 0.0000 N.0000 ..0N. 0.0.0. 0.000. 0.0000. n.-~ ..00NN ..00.c. 0.00.0 ..0.. 0.000 0..~0. unw¢<.amxu- up..:mtgu~2_ 4u1< 0..00.. 0.0000 0.0.0 ..N.00 0.0000N n..0. 0.0.00 0.00omw Non.n0 0.00 0.000 ..0000 0..000 0.000 ..N... 0.0000 n.0wmh. 0.N0. 0.000N 0.0000. 0.0.00 0.~0 ..00.. 0.000. 004.4.00 44.0.000 .008 440.200 .040 .004.00:. 410.040 44.0.000 .003 440.204 .040 .004.001. 00.244000.2. 44.0.000 .003 440.200 .040 .00. 40000 44.0.000 .003 440.200 .040 .00. 000.4.00 .0030 44.0.000 .003 440.200 .040 .00. 4000» 44.0.000 .00! 440.200 .040 .00. 02400 4.4.00 00.:0.03 004 0040.. 20.000I0000 ..0.u000. II .0NI4 0.0.. 180 oo~¢noa ~.a~9~n :.oao_= ..... ..... ..... ¢.~3a-~ ~.afi—N- o.~n-- dC—ab ~.a-qm ~o7¢6~n nc~eoan ..... ..... ..... ~c~¢~o~ ~o~¢~o~ o.-~o~ pawn mocaNéq ~o¢0nnu ¢.n~m*. ..... ..... ..... noco¢~n «.mmm~m conconn atapru N._~ann no¢-0n no~1qrn ..... ..... ..... ~030n¢- ~oacmv‘d oowaow—u pmcw aduoh oz lInJaaz ocean! tin—~23 ODD—II Ilwa<~0w£ll «_dowxaoq g_<_:utxy_£_ and—«.43 n.43w1334 4_<_2gtfiu.2_ a;4—<_4: &— wZD—cwz bzwtwauaamu c319 .vouuuuon wzu we mao~m :oHuMUuOchmuh mouu cwmauwm uwucouuuuwa II .cNI< oaauh 182 ’0 3.. ~au! non N. soo¢~ ~o¢¢d n.o- moc- No¢~ Noeu Incas.-- 44aanm ¢cnm OoOfi «ow cows.— 0.07m acoua mows ~ca~n o.~¢n_ ~.~mo ~.~ad ~._~. :.oon noraw Dowfid oo- modw Go¢= ~54; fl. ‘. comm— comm— Oom— Coma mod hwy: can N. ~O ~o¢ ~.n- «on- aoamn comma «oat node IIaoc.II 4<;_zuu oma_¢_ua noun ddpopoam mumn 362-836 :bzdw haw) ddzbzwu oo~ hmdw .00. (Ouau nosoo— 4(bD—nan oooam ouz-too O~z «oumfl Adhcpoaw oownu ouz—too OhZCm mom~ paw) so~a 4<1bzwu ao¢e bmdw .00— mz un4313m a..c: :.ssn :.:— o.n~ _own sound cuwnd «one not: ~.3~O_ ~o¢n- 3.03: bmdu monnm ¢.am_ dosuu «ooufle ~.3oc~ homom— Oonoa Donna Noodw :.mn coda Nooo monu— monu— mo.0¢~ ooNO— OoN~7 no~mn .mwr Mc—o ~oo~ 0.0m act~¢ moo—e moan— Non~< IIoco_II 4.3—zuu cad—«.mo Q. :0 «.93V monON «oON OoON poo—o— ~.0ao- _m mza~ow¢ watwanaouc 1310 as: uu«:m\;u)ouu I cameo: voucmuoz auwwmumw< can .kuswuoz ouuuvoaLUQau .vuaquuoe may no mzoam ceauuuuonmcmue coho cmosuom muucquuuun II .uNI< Gun-h 184 60¢ hwy! ~o¢¢~ ~o¢¢d mun—N boo-N No¢~ Nova Iuaoa_-I 44Q13m c... 3.0m :. (‘J Donsdu :.«oo hows Moca OINAn oowoau ~ofl¢o~ o.¢¢. ova: noamm no¢o~ oo~¢~ Oodw nova :.ns hwca ‘. mono— nomad com— com— a.— m.— uwwz donqw _on—N oocmn Gonna Icoaa.II dag—z»; om4_«pac «cemo~ #0000 ¢o~u Oonn Occas— soMOOu oomma comma mosmw ao~¢u moNN ho? Ooah pmdw Adha—aam Caz-ISO thcw haw) 4<¢hzwu pm mza—cwx _zwtwa—aawa nsxu A.v.u=oov II .NNI< CHAIR $035208 185 w~.o mm.o ow.o Nw.¢ mm.o wn.o NN.o ~n.o 8000 2.0 ww.a- mm.o mm.nww n—.o Nw.~m nm.o N.wNN mvoumpom wovam nN.o fiw.mnw mn.o mu.wmn._ Nm.o am.on nm.o :o.mmm 88> m_.o we.” mn.o wa.m~ N_.o w..— n_.o aw.d mammm N_.o :w.n _~.o fi¢.w o~.o No.¢ om.o mm.a EoU _m.o no.ow¢ m~.o uo.¢m wN.o om.mw~ mfi.o mm.ma mch ..mwsm w_.o aw.mmu.o_ wn.o 0......ow mm.o om.m~m.o_ a~.o o..amm.k mcmcmm mn.o nm.w mm.o mw.d dm.o mm.— ¢N.~ :~.o—~ mm.o ow.wm um.o nu.n~ w_.o mm.~ ¢N.o ~m.a Emu—BE 8230... ...o N~.a__._ m¢.o Nw.NN~ nm.o uo.won wm.o mm.an mic mwém «cam co_um_..m> «0 .mmvoU vucmwhm> _m 533»; no .380 oucmtm> :u cofimtm> mo .uvoU vucmtm> flu coflmtm> no .380 oucmtm> —m 533...; no .380 00:22; m<_ cofimtm> w0 .3000 vucmtm> ~¢‘ coflmtm> no .3000 vucmtm> fi< (rum BEBEEE~ Em; vawEB 3.38.53:— ozuuoamum 05 £55 moumEme Ev; 3:300 mo cozmtm> Ho «58580 vcm mucmtm> .I w~I< v33. 186 > .I. cofimtm> no «0223000 q H u GUCMCM NA> I .>V N . > 60000:? 0: 0.5 0.05 .0000; 000 20m 0020002005 £000 ..3 :8 00:30.0 0:0 3:0 0>mc n0 smack: E o o c o o and o o o o o oniam o o o o o o o o o o o o and 0.x..mm 36 80.2.3 no .080 wmdmm 85:; mm o cofimtm> “0 .3000 o 0ucmtm> Nm 30.0008 II w~I< 030% 0000:3500 187 0 0 0 0 0 00 .0 0 0 00.0 00.0 53020.» 30.3000 0 0 0 0 0 01000.0 0 0 00.00 00.20; 00:0_._0> 3: 00.0 010 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 :030?0> 30.3000 00.0 01000 «0.000 00.00 0100 00.0: 0100.3 00.3 3.00 00.03 00:0_._0> 0: 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.3 00.0 00.0 :030_..0> 30.3000 00.0 00.000 00.000; 00.00 00.0— 00.000 00000.00 00.00 00.3 00.000 00:0_._0> N: 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 :030_..0> 30.3000 00.0 00.030 00.000; 00.00 3.0 00.00 00000.00 3.0 00.00 00.000 00:0_._0> :— 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 :0_00_._0> 30.3000 004 00.000 00.000 00.00 00.00 30.000 00000.00 00.0 00.0 00.00 00:0_._0> 0— 00.0 00.0 00.0 00.0 2.0 0 00.0 00.0 00.0 00.0 :030_._0> 30.3000 00.0 00.0 00.000 00.0 3.0 0 00000.0 00.0 00.0 00.000; 00:0_._0> ~— 000 00.0 00.0 00.0 00.0 00.0 00.0 _0.0 00.0 00.0 :030C0> 30.3000 00.0 0.000 00.000 00.3 00.0 00.00 00400.0 004 30.0 00.00 00:0_._0> : (mm 00000 003300 000.; 0:000 500 0:00 0:050 53:05 000000... 003 030M000»? 00030 ..0me 20; 002303 0300000010 030000000 05 :33? 03063.5 20; 002300 30 :030_..0> 30 #:2033000 0:0 00:0_..0> .i 0min. 030,—. 188 00.0 00.000 00.0 00.00 3.0 800:; S .080 8.20 85:2, a: 0 :0_«0t0> 00 40000 0 00:0_._0> NE 00.280 -- 0~-< 030:. 0 0 0 0 0 0 0 0 0 0 0.05 0200002 008:. 0 0 0_ 0 0 2 0 0 0 0 0.05 030000 00E: .. - 0.00 0.00 0.00 0.00 0.00: .. 0.00: 0.00- 00 0.00 0.00 0.000 0.00 0.00.. 0.0— 0.00 0.000 0.00 0.00- 00 0.00 0.00 0.000 0.00 0.07 0.00 0.00 0.00 0.000 0.00.. _U u 0.00 0.00 0.00 0.00.. 0.00 0.00.. .. 0.00- . 00 0.00 0.00 0.00 - 0.07 0.00 0.00 .. 0.00s - _0 .. u u u u 100- u .. 0.00.. 0.00 0m .. - u u u 0.00- n u 0.00 100.. 0.0 9 .. u u u n 0.3 n - 0.00 0.0 0.0 one. u u u .. .. 0.00 n .. 0.00.. 0.00- 00 0.0 0.00 0.00 0.0 0.00 0.0- 0.00 n 0.00 0.00_ 00 0.00 0.00 0.00 0.0 0.00 0.00 0.00 0.00 0.00 0.00 E 0.3 0.00 0.00 - 0.00 0.00 0.00 0.00 0.000 0.07 0< 0.00 0.00 0.00 0.0- 0.0- 0.00 0.00 0.00 0.00 0.0— 0< 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.0_ 0.0. 00 <00 000.00. 00000 0030000 0000> 0:000 :30 0:00 0:0:0m 500:0—0 0000000. 0000 00050—0000:— 00030 00000 0000Efi0m 20; 0000000 0:0 000505.00:— :00>$0m 005 u- 00-< 030... a o o o _ 0 N o 0 0 m 005 200002 8:5 s 0 0 0 n 0 0 0 0 0 0 35 03:80 85$ - - .. - .. 0. a- - - 0.2- 0.2 a: - - - .. - 0.00- 0.2: - 0.00 0.9.... NE - - - .. - 0._~ - - 0.00 000 :0 w 0.: 0.00 NS 0;: 0.: 0.: 0.00 0.: 0.3 ~.R- m: 1 0.: 0.00 0.00 «.00 N.N 0.: 0.00 0.0- 0.2 0.00 N: 0.00 0.00 0.00 0.00 0.00 0.00 0.00. 0.00 0.00 0.00 0: 0.0.0 0.3 0.00 0.2 0.00 0.: 0.8 0.00 0.5 0.00.. a «.00 SN 0.00 0.2- 9a- 0.2 0.00 0.3 0.00 0.0 0 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0: 8 <00 .0000. 00000 00000000 000.; 0:000 :000 0:00 0:0:00 59:20 000000 0. 0000 0000000010 00030 00030 0000:0200 E0; 00:00.00 0:0 00000000. :003000 020 nu ~0u< 0.000. 191 0.000 4.004 0.000 8.3N4m aoso¢4 0.444N 3.030 3.000 0.00V 6.0: co~0 0.0: «.00 50:0 ~00! 309m $.JN 4.30 0.00 ¢.0~ 00300 00000 0.004 0.004 uneaadn: 04¢.z00 0.000x000 0.3N¢ 3.M~a 0.04 ..Na 3.0N ~. *0 0.704 0.9¢~ Von-q 093004 Noan4 6.000 0040 0.00M 0.004 3.304 co~enq 3.40:4 3.000 0.030 0.000 N.4m4 0.3: O.N~ ~.?fi Mona 0.00 m.v9~ c.30N 0.NVO haw: 0.30 0.0! 0.000 n.:¢m 0.00V mo0nN II3004II 44:0200 0—4—50200—24 mza—oux >4103w ~..a¢ 0.000 0.”. 0.0. a..~ Q. a. $033— Dock— 0.0N ¢.¢m ¢o¢fl 3.0N04 0.3004 0.000 pw44 0.000 9.004 4.04N 0.0—no 0.0004 0.0004 0.000 0.009 0.404 0.00 0.00 0.00 moonw4 0.30N 0.0—c noOON 00w: #0000 «.000 4.00 uuo=0~uu 4000200 004.<~00 noetm o.-0 m. ”O 0.004 0.004 N.NN ~0N~ «.0-04 0.0004 cosnw 004w 44b3b03m 302—233 3024» ~00: 44¢~Zwu 004w ~30- 2034 44030030 302—:33 0—240 uwwx 4420209 004w .0!— w24u C4030 44030030 002—136 30240 paw) 4410200 hm4w 0144—0. 424240 44000030 Dcz—ZOO 90240 pmw) 44¢wau bm4w .1444—t. 20402440 44h0~¢3m 002—200 00240 haw) 44¢hzwu 004w ~93— 0004030 44000030 302—too D~Z4m paw) 44¢~zwu hm4w .304 wu-¢ embrouws ~32 w¢4 064wu> wzo—ow¢ hawtw¢~3ow¢ menu can suaazuc00 I 04000: 00000000< 0:0 .uuaavoauaucu .00040000 000 00 03000 :0000000000009 0000 0003000 00000000040 II .0004 000.9 192 ~.4 V.4 3.Nn 3o~n ¢.o¢ :.oo o.~ ~mwz :03V4 n.3N4 ~.¢o4 ~.¢c« :-::..uu 44 o.~o4 l‘ocaall 441—249 u—4—fiutiup2— «TD—34¢ >4133m ~.am c.0n ~.3v «.3no N.~9o 0.3m Noam ¢.c~n oomcm4 3._¢h o.~m4 ).-4 c.o~m 3.m34 3.344 .m44 ‘. 3.m44 3on44 mo¢4 mo¢4 m.~ m.4 any: o.m N. h. ho¢ m.o3~ mooav c.0~n mum moo—n II3334II 44¢meu 3w4—wa3 moo4m ~.~0¢ Io~ eoo—n aomn¢4 3.4oh c.~m~ 3.~o¢ ¢.44~ 3.3—4 m.m~ o.o~ ~m4w 44h0b33m 332—!33 3pz4w bmwx 44¢~Zuu pm4w goa- 4ouou 44~3~n3w Dal—too 3h24m paw) 44¢~2wu pm4w .004 ww3b4hos hwwxm 44—3»33w 332—!33 3hz4m hum: 44¢hzwu um4w .33. 4093» 44p3h03w 332—!33 3b24m haw: 44¢hzwu hm4m .30— m24wo awpzuuwx ~32 w¢4 m34w—> mzauowx »Zwthuaawz ufldu A.v.ucoUv II .an4 04448 193 :.:acw ..va.n :.acc~ 0.~nm 0.~mm o.~mn m.n¢-~ m.m:¢~. c.~0~m_ adhapaam ..o¢ ~.0n :.no ..0_ ..~ o.~ ..m- 0.m- o.~.n »mwj ~.0o~ o.o.~ :.::: :.0n ~.:n «.~o. ccaa: n..0n~ nonm0~ aaapzmu comsn_ ~.mm«. a.an.w :.~_n ~.c:¢ ..nee m.mco.~ ~.0o¢c— ¢._me~_ hmc. ..mm- m.~nc m..n_ 0.-o_ ..0o_n m.o~o. 4<¢pzwu o.~a>- ~.o_mm. o.oone. o.d~n~ o.on:. ~.o.od ~oo_oe. o.~.0- ~.n.o- ~m tlmamwa 000.II Ila—.20 000.3! Ilwc<~uwxli w— Innawma 000~II wimp—z: 0:0—It olwxdpuszI u—45u109< w—<_0wtxu—z_ auaudpua wbdoaxao< 0_<_c~t1;_z_ cud—«p»: u~ mza-uwc thth—Dawx cacu cam Sancho I ammvo: wuumwumw< van .ouauvoahuucw .voawauoa osu mo mzbam :ouuouucnmcouh menu comauwn auucwuouuua In .enl< OdAIH 197 Nod no~ hag: as Com «onw— dcnvd No¢o— ~o¢o~ nuaaa.u| 4<¢hz4u u. mzo—uwz bzwtwd—Dawu couu A.v.uaoov II .emu< OHAOH 198 noo— _.o~ monmfi domnu comwN moaNdm oo~o¢u coda‘N Donoa soamo oom- town cons oomo noeo mace Nonox :oNod Noaoa hwy: Qoon «oar Nova No:— ...-I .0 ~00 acme moOM Noo~ Ounce somae —omm~ «amm— II000.II 40aq0m n.no: ¢.~an c.n~ 0.o~ ..- a. a. wo—ON Nommw nod: no_¢ coca—N soowo— Doom: away monnfl #oon— «oh-N Oo0~n¢ soone— homcnu canoe Deena Non—N town no—o N.oo con-— a.n4~ Sonata OoNOu O.-o manna bmwx ”cud no- ace-Q outdo magma NoQN— IIoooqII d¢¢_zuu zed—«bu: coon: ¢.~1n #90— —.V~ Q. ”. moQON moron QoON QOON soOuo— nonaod pwaaaam ~.om 9.3“ Nocw oosdou {0:06 ooman ¢.m~a~ ~ofi¢o~ a.- n.nna «oven o.~¢. ¢oc~ 9.3— o.m~ _wem ‘. moaou wooed m.~ 00m N. h. soc don-N ~.m- comm“ 000mm «out node IIooouII 4<¢hzwu owd_ 4(n0u33w Doz—zoo D»Z(m haul 4<¢bzwu bm ”ZS—aw: bzwthuaouc Aacu A.v.ucOuv II .nnI< CHAIR