Date This is to certify that the thesis entitled Statistical Sampling of Agricultural and Natural Resources Data: A Comparison of Aerial Sampling and Area-Frame Sampling Elements with Special Reference to Lesser Developed Countries presented by Gerhardus Schultink has been accepted towards fulfillment of the requirements for Doctor of Philosophydpgreein Resource Development Oct. 23, 1980 0-7639 .. ~ liljllllllzllllllllllfllllllllMilli OVERDUE FINES: 25¢ per day per item RETURNING LIBRARY MATERLAES; Place in book return to mum-1:3 charge from circulation l‘iiC-Lh ~. Y \~ . u -C ‘ winch” (/w’. rlnyuflua‘ -. . a V . (7:73 o s; an 1 ((361905 OCTZZZDBS STATISTICAL SAMPLING OF AGRICULTURAL AND NATURAL RESOURCES DATA: A COMPARISON OF AERIAL SAMPLING AND AREA-FRAME SAMPLING TECHNIQUES WITH SPECIAL REFERENCE TO LESSER DEVELOPED COUNTRIES By Gerhardus Schultink A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Resource Development 1980 // D (5; A“? 92/4. ABSTRACT STATISTICAL SAMPLING OF AGRICULTURAL AND NATURAL RESOURCES DATA: A COMPARISON OF AERIAL SAMPLING AND AREA-FRAME SAMPLING TECHNIQUES WITH SPECIAL REFERENCE TO LESSER DEVELOPED COUNTRIES By Gerhardus Schultink The development of agricultural and natural resource management programs and related policies for lesser developed countries (LDC's) requires the availability of reliable information on the status and extent of the resource base. This focuses attention on the quality of the spatial information which is largely determined by factors such as spatial and temporary accuracy. The generation of this information involves important cost-accuracy trade-offs, especially if primary data collection efforts are introduced using remote sensing technolo- gies. The major objective in this research is to evaluate the potential of low-level aerial photographic sampling to effectively capture re- liable resource inventory data. This includes the integration of these data with multi-level inventory procedures, such as satellite and high-altitude remote sensing as well as ground (sampling) surveys, in an effort to supplement secondary data and to improve the compre- hensive quality of the resource information data base needed to effectively support natural resource monitoring, management, planning and policy analysis. This research compares aerial sampling and area-frame sampling procedures to assess cost/accuracy trade-offs and to evaluate the Gerhardus Schultink appropriateness of the techniques, given various topographical con— ditions, information needs and technology levels encountered in a typical LDC setting. In this process the design of aerial sampling procedures is discussed and two selected sampling designs and their associated sampling intensities are evaluated. Two modified, syste- matic, aerial sampling designs are tested for the capacity to provide accurate land cover/use estimates for a selected test site of inten- sive, mixed agriculture in a tropical region. Cost and accuracy of the aerial and area-frame sampling procedures are discussed in com- bination with the cost estimates for a nation-wide sampling survey. Light aircraft survey techniques and sampling survey designs aim at providing an application-oriented framework to create an in-country operational capability to conduct agricultural and natural resource inventories using aerial sampling techniques. Results of the Dominican Republic case study indicate that four of the five major land cover/use categories representing more than five percent of the surface area of the test site in intensive agri- culture could be accurately predicted (with a 95 percent confidence level) using a sampling intensity as small as 4.8 percent. Therefore, it can be concluded that aerial sampling techniques, using a modified- systematic sampling design, could provide a viable, cost-effective resource inventory alternative, especially if access condition, time limitations and the state of technology pose significant constraints. This situation is frequently encountered in many developing nations. The research identifies the need to improve the data specifica- tions to support short- and long-term resource planning and policy Gerhardus Schultink development. These specifications should include the type, spatial detail, accuracy, timeliness and anticipated cost of data capture ef- forts over a period of time, as well as the institutional framework needed to support these activities. This identification process should include the selection of the appropriate technology for data capture, storage, analysis, and information retrieval. Selection criteria should consider the geographical, budgetary and technological con- straints as well as the priorities for development identified by the individual countries. ACKNOWLEDGMENTS I am grateful for the assistance and encouragement received from a number of people. Without their special efforts this work could not have been completed. Special thanks is due to Dr. Shelton who guided me through the major stages prior to his untimely death on August 20, 1980. Professors Barlowe, Ramm and Farness served on my committee and they provided me with many valuable comments and editorial assistance. The support of the Comprehensive Resource Inventory and Evalua- tion System (CRIES) project staff is gratefully acknowledged. The CRIES project involves funding from and participation by the U.S. Agency for International Development, the U.S. Department of Agricul- ture, the National Aeronautics and Space Administration and Michigan State University. In addition, the support and assistance of the fol- lowing institutions is gratefully acknowledged: l. The AID Mission in the Dominican Republic; 2. La Secretaria del Estado de Agricultura--SEA--(The Secre- tariat of Agriculture); 3. Sistema de Inventario de Evaluacion de Recursos Agricolas-- SIEDRA--(Agricultural Resource Inventory and Evaluation System); and 4. La Fuerza Aerea Dominicana--(The Dominican Air Force). Finally, I want to thank my wife, Joanne for her support and the sharing of my feelings of frustration and accomplishment during the cruCial last stages of this work. ii TABLE OF CONTENTS ACKNOWLEDGMENTS ......................... Chapter INTRODUCTION ....................... Research Context, Funding and Development ....... Problem Statement ................... The CRIES Project ................... Spatial Information System Component ........ Area-Frame Sampling ................ The Potential of Remote Sensing .......... The Need for Strata Refinement ........... Study Objectives ................... Major Study Goals ................. Research Approach ................... County Description .................. Description Study Area ............... Test Site Description ............... CONCEPTS OF REMOTE SENSING - BASED NATURAL RESOURCE INVENTORY SYSTEMS ................ Remote Sensing Systems: A New Perspective in Integrated Natural Resource Inventory Systems . . The Need for Natural Resources Information ..... Contact Versus Non-Contact Data Gathering Techniques .............. Data Collection Using Remote Sensing ........ Multi-Level Remote Sensing ............. Current and Future Spaceborne Systems for Natural Resources Inventories; Landsat and Space Shuttle . . Characteristics of Imagery and Imaging Systems . . . The Landsat System ................ The Large Format Camera .............. Sample Survey Considerations Using Light Aircraft Sampling Techniques ............ Resource Planning Unit (RPU) Inventory ....... Sampling Resource Planning Units .......... Sampling Considerations .............. Selection of the Sample Size/Sampling Error Sampling Error Associated with Area Measurement Using a Dot Grid Method ............. Area-Frame Sampling Methodology ............ Important Considerations in Area-Frame Sampling . . iii ii Sampling Frame Methodology as Applied in the Dominican Republic ................. 7O Multi-Stage Sampling: An Integration of Landsat and Aerial Sampling Data .............. 76 III. DOMINICAN REPUBLIC CASE STUDY: METHODOLOGY ........ 80 Landsat Data and Derived Information ......... 80 Information Extraction ............... 8l Refinement of Landsat-Based Strata ......... 8l Aircraft Survey Data ................. 87 Imagery Acquisition ................ 88 Ground Truth Acquisition .............. 88 Development of the Classification Key ....... 90 Imagery Interpretation and Area Calculations . . . . 9l Quarterly Survey Data of Area-Frame Sampling ..... 93 IV. RESULTS AND ANALYSIS ................... 99 Comparison of Area Statistics Derived from the Aerial Survey and the Area-Frame Sampling Survey Relating to the Sample Segments ............... 99 Error Estimation of the Aerial Survey Inventory . . lOO Interpretation Error ................ lOl Measurement Error Associated with the Use of the Dot Sampling Method ............. lOZ Crop Inventory and Error Comparison for Plantains for the Sample Segments #14224 and #l6224 . . . . 104 Development and Analysis of Linear Sampling Procedures Based on Aerial Surveys ......... 106 Development of the Single Flight Line Sampling Design ................. l08 Sampling Statistics and Evaluation of Two Sampling Designs ....... ' ........... ll4 V. CRITICAL ANALYSIS OF ALTERNATIVE SAMPLING PROCEDURES FOR AGRICULTURAL AND NATURAL RESOURCE INVENTORIES ..... l26 A Comparison of Area-Frame and Aerial Strip Sampling Procedures ................ l29 Requirements and Limitations for the Application of the Two Sampling Procedures ......... 129 Design Task Elements and Information Quality Aspects ................. l3l Cost/Accuracy Assessment of the Two Inventory Techniques Aerial Sampling and Area Frame Sampling . 136 Applicability of the Two Sampling Procedures ..... 142 VI. SUMMARY AND CONCLUSIONS .................. l46 Summary .............. I .......... l46 Needs for Further Research .............. 147 Conclusion and Recommendations ............ 148 iv APPENDIX 1 ............................ 15] APPENDIX 2 ............................ T65 BIBLIOGRAPHY ........................... T70 N—l #00 (DNOTU'I #00 N O N N N __a -—l ._a —J ._a _a_| N -' O KO CD \l 01 UT boo LIST OF TABLES Gross National Product of the Dominican Republic by Activity . 6 Distribution of Family Income Among Urban and Rural Population ........................ 8 Annual Labor Requirements of Major Agricultural Products . . . 9 Yield Trends fbr Principal Agricultural Products Based on Three-Year Moving Averages: 1962-64; 1968-70; 1974-76 . . 10 Land Capability Classification ................ 26 Land Capability and Conservation Requirements ........ 27 Crop Calendar, Moca Test Site, Eastern Cibao Valley ..... 31 Typical Imagery Scales for Multi-Level Remote Sensing Systems .......................... 40 Characteristics of Remote Sensing Imagery .......... 45 Landsat Spectral Bands .................... 47 Example of Resource Planning Unit Description in the Dominican Republic .................. 54 Strata Definitions Used in Sampling Frame Design of the Dominican Republic ................. 71 Sample Unit Allocation by Stratum and Total Area ....... 73 The Selection of Samples Used in Past Surveys in the Dominican Republic Employing the Area-Frame Sampling . . . 74 Imagery Scene Number and Dates Used in the National Land Cover/Use Inventory of the Dominican Republic . . . . 82 Land Cover/Use Classification Categories of the Original National Landsat Inventory ............ 84 Total Land Area (km?) by Land Cover/Use Category as Identified in the National Landsat Inventory ....... 86 Total Enumeration for the Moca Test Site by Classification Categories ................. 92 Total Enumeration of Classification Categories for Two Sample Segments Based on Aerial Survey Information . . . . 93 Crop Inventory for Survey Segment #14224 Included in the Moca Test Site, December 1978 and March 1979 ....... 94 Crop Inventory for Survey Segment #16264 Included in the Moca Test Site, December 1978 and March 1979 ....... 96 Matrix with Percentages of Commission Errors, Interpretation Accuracies, Overall Performance, and Omission Errors Associated with the Interpretation of Color Infrared Film, Scale 1:30,000 ................... 103 Area-Frame Sampling Data for the Two Sampling Segments (Plantains) ................... 105 Selection of Initial Sampling Size to Determine the Actual Sampling Size and Intensity ............ 113 Scale and Coverage Matrix for Selected Small- format Camera Systems Used in Low- altitude Photography ...... 115 Data Summary of Sample Parameters for Two Sampling Designs, Various Sample Sizes Intensities and Associated Land Cover/Use Categories ................... 117 vi 27. 28. 29. 30. 31. 32. 33. 34. Estimates of Population Percentages for Sampling Intensities of 9.52%, 14.26%, and 19.3% ................ 118 Estimates of Population Percentages (95% Confidence Level for a 4.76% Sampling Intensity .............. 120 Correct (C) and Incorrect (I) Prediction of Population Means for Five Major Categories .............. 122 Absolute Error Percentages for Major Land Cover Use Estimates Based on various Sampling Designs and Intensities ...................... 125 Some Design and Information Quality Aspects of Area—Frame and Aerial Sampling Procedures .............. 132 Correctly Predicted Population Percentages for the Four Crops Included in the Test Site Study ........... 137 Cost for the Initial Coffee Survey and First Crop and Livestock Survey ..................... 140 Comparison of 35mm and 70mm Camera Systems (1978) ...... 153 vii -_l O cow 10. ll. 12. 13. 14. 15. LIST OF FIGURES Gross Flow Chart of Sampling Research ............. The Electromagnetic Spectrum: Wavelength, Frequencies, and Band Designations. (Adapted from Tomlinson, Geographical Data Handling) ................ A Portion of the Electromagnetic Spectrum. (Adapted from Johanson, Remote Sensing for Planning Resource Conservation ....................... Selection of a Systematic Random Linear-sampling Procedure for Resource Planning Units ................ Area-frame Sampling Stratification of the Dominican Republic ......................... Outline of the Moca Test Site. This Area Conforms with the Count Unit of the Area Sampling Frame and Contains 16 Sample Segments ...................... Landsat Mosaico of the Dominican Republic ........... Aircraft Camera Slide-mount with 70mm and 35mm Camera Systems Installed ......................... Schematic Outline of Various Samples Centered Along a Single, 4 km. Long, Flightline. The Layers Represent Captured Surface Information for Sample Intensities of Approximately 5% Through 20%. The Four Sample Sizes are Assumed to Consist of a 4 km. Long Strip, Composed of 16 elements, (Photo Plots) Each ..................... Schematic Outline of Sample Test Site and Four Flightlines Associated with the Modified, Systematic Sampling Design. Intervals Between Flightlines and Frames are Constant; Starting Points are Randomly Selected for Each Flightline .................... Outline of Samples Representing a Simulated Aerial Survey to Estimate Crop Composition of the Test Site ....... Nikon 35mm Camera, Large Motorized Back for 250 Exposures, Battery Pack and Intervalometer ........ Hasselblad Cameras (70mm), 500 EL/M, 150mm Lens, Medium- Size Back of up to 60 Exposures (15 Feet of Film), Motorized Film Transport with Battery Pack Included, Adapter, Intervalometer, Cable Release for Manual Operation (10 Feet) and Standard Hasselblad Camera with 80mm Lens and Small Back (12 Exposures) ................. The Relationships Between Camera Distances and Flying Height for Vertical and Oblique Aerial Photographs ........ Camera Mount for Light Aircraft. MSU Remote Sensing Project Configuration for 35/70mm Camera System, Lower Platform . . viii 22 37 38 57 72 75 83 89 107 109 112 155 155 156 164 CHAPTER I INTRODUCTION Research Context, Funding and Development The Congress has stated that the objective of U.S. foreign as- sistance programs is ”to assist the people of less developed countries to acquire the knowledge and resources essential for development and to build the economic, political, and social institutions which will meet their aspirations for a better life with freedom, and in peace."1 This statement focuses the necessary attention of aid programs for lesser developed countries on the major concerns and the primary needs of the more than 50 percent of the world population living in rural areas: food, shelter, and production potential of the agricultural sector. Any operational program aimed at promoting active participation of the rural poor and small farmers in increased food production and resource development requires substantial information about the ex- tent and the quality of the resource base, the various components of the production base, available and capable management, and the know- ledge of technological options for development. Such a realistic assessment requires detailed knowledge of the agricultural sector: 1National Academy of Sciences, 1977, Resource Sensing from Space; Prospects for Developing Countries, Washington, 0.0. 1 its current status, its potential and the existing linkages with other sectors of the economy. The research reported here was conducted with the foreign as- sistance context expressed above, specifically in connection with a cooperative effort involving U.S. agencies and Michigan State Univer- sity to develop this knowledge. The inventory and analysis procedures described as part of the research were conducted in connection with an agricultural study of the Dominican Republic. These efforts aim to adapt existing resource inventory methodology and to develop new techniques by the United States Department of Agriculture (USDA) and National Aeronautics and Space Administration (NASA) to improve agri- cultural sector planning systems and to expand matters addressed by sector analysis techniques. Their work was initiated as a result of efforts by the Economic Research Service (ERS) Foreign Development Task Force of the USDA resulting in a draft proposal in 1975: to explore over a one-year period the possibilities of estimating technological and economic potentials for agricultural production in developing countries. The specific proposal objectives listed were: a. To determine the availability and quality of specific infor- mation about soil resources . . . and other important physi- cal information such as water, weather, and climate . . . and to develop a system for organizing and handling such data. b. To assess the availability and quality of current land use information as it relates to soils of different productivity. c. To evaluate the prospects for acquiring information on present and prospective type, cost, and extent of resource development measures such as irrigation and drainage. 1TA/AGR/ESP, 1976. A proposal for a comprehensive land and water inventory and evaluation system for agricultural planning. To evaluate the availability and quality of yield-nonland input production function relationships. To specify those economic, social, and institutional con- straints which should be considered in assessing possibili- ties and costs . . . To examine alternative methodologies for evaluating pro- duction potential . . . To develop a detailed plan of work for undertaking a study This initial proposal resulted in negotiation with the U.S. Agency for International Development (A10) and subsequently in limited funding for the development of a plan of work. After several revisions of a draft plan of work, a final document1 was produced in 1976 listing the following project goals: 1. To assist developing countries to develop their capacity to identify and analyze the consequences of alternative policies, programs, and prospects for agricultural and rural develop- ment in terms of their own multiple economic and social goals. To improve the information and the analytical basis for mak- ing decisions on agricultural and rural development strate- gies, policies, and investments. To expand the number and enhance the capability of develop- ing country planning personnel to construct and use such an information base and analytical system. The document listed the project purposes: 1. To select and apply techniques for collecting, classifying, collating and documenting data on a country's land and water resources, land use, production inputs and expected outputs, production costs, technology options, and insti- tutional constraints. To establish a system, using existing data management tech- niques and analytical processes, for evaluating these data. To demonstrate the analytical capabilities of this system and to test the reliability and usefulness of the results. 1 Ibid., 1976. 4. To develop procedures for linking the source data and analytical system into a sector analysis. 5. To internalize the use of the techniques developed as part of the project and to integrate the system with sector analysis activities in-country. Current AID-funded sector analysis work as applied in several lesser developed countries (LDC's) aims at providing policy makers with the necessary analytical and monitoring capabilities to develop policy instruments in meeting specific country objectives. This analytical capability is to provide insight into the status of the agricultural sector and its functional relationship with other sectors of the economy. Fer this reason agricultural sector modeling techniques in- clude the location of available natural resources, as distributed among alternative enterprises, in order to provide an optimal spatial relationship between producers and resources. The modeling effort aims at achieving certain objectives (food, income, etc.), given cer- tain constraints and model assumptions. One of the critical elements in sector analysis with the modeling approach is the use of realistic constraints and data, representing the relationship between land re- sources and their environment context. It is specifically this rela- tionship which identifies the production environment available to meet a final demand. The Dominican Republic was chosen as one of the test countries for such a comprehensive resource inventory and analysis effort, since a sector analysis project was being implemented. The Comprehensive Resource Inventory and Evaluation System (CRIES) project which re- sulted, involves funding from and participation by the U.S. Agency for International Development, the U.S. Department of Agriculture, the National Aeronautics and Space Administration and Michigan State University. Participation of USDA is covered under PASA #AG/TAB-236- 15-76 and NASA under PIO/T 931-0236.01-3l78632. Participation between the Economic Research Service (ERS), U.S. Department of Agriculture, and Michigan State University is covered by Cooperative Research Agree- ment No. 12-17-07-8-1955. Two other countries, Costa Rica and Nicaragua, were chosen for similar prototype development during a three-year period from 1976 to 1979. A subsequent fourth country project was initiated in Honduras in 1980. Some of the techniques were applied to agricultural studies in Syria during 1978-79. The work is currently being extended to additional countries. However, the Dominican Republic was the first completed project and provided the opportunity to conduct the research reported here. Problem Statement The Dominican Republic case study illustrates the basic problems this research addresses. Agriculture remains, in spite of the recent growth of other sectors in the economy, the most important sector in the Dominican Republic (see Table l). The direct contribution to the GNP is currently about 20 percent, with additional values added through forward linkages in sugar cane processing, cotton spinning, vegetable canning, etc. Exports of products with an agricultural origin amounted to approximately $500 million (US) in 1976, or about 68 percent of the total merchandise exports. More than 50 percent-of the population employed in the sector experiences widespread underemployment and poverty. Rural incomes are not only lower than those of the urban population but also spread more unevenly. 0n the average, urban in- TABLE 1 Gross National Product of the Dominican Republic by Activity 1964 i 1967 1970 Agriculture: Crops 16.9 15.8 n.a. Livestock 6.6 6 6 n.a. Forestry & fishing 0.6 0.6 n.a. Total agriculture 24.1 23.0 22.4 Manufacturing 16.5 17.9 17.9 Commerce 17.6 16.4 16.9 Public administration 13.1 11.9 10.6 Transport & communications 6.0 6.8 8.2 Rental property 6.2 7.0 6.7 Construction 4.4 4.7 5.4 Banking & insurance 1.7 1.3 1.7 Mining 0.9 1.3 1.2 Utilities 1.1 1.2 1.2 Other services 8.4 8.5 7.8 Total 100.0 100.0 100.0 NOTE: Numbers are percentages. SOURCE: Neil, Thomas E., et a1., 1973. Dominican Republic. Area Handbook for the come per capita is estimated to be four times that of rural income. According to the Secretariat of Agriculture (SEA), approximately 64 percent of rural families earn an income at or near the rural poverty line of $35/month (OR) or less than 20 cents per day. This figure probably does not include the full estimated value of subsistence food production (Table 2). The underemployment condition of the rural labor force could be improved by means of a technical assistance program to encourage the production of labor intensive crops that produce a higher cash flow per hectare such as tobacco, fresh vegetables and industrial crops like cotton (Table 3). The major crop with the highest yield per hec- tare is sugar cane. This crop accounts for 60-65 percent of the GNP during the last decenium and for 32 percent of the country's exports in 1978, with an annual production of 1.4 million metric tons. A great variety of other crops are grown; their yields vary greatly (Table 4). The yield variations makes it difficult to optimally use the cur- rent processing capacity for sugar can and difficult to develop short- and long-term agricultural policies aimed at satisfying final demand."2 An additional problem expressed by government agencies is the in- creasing surface erosion and runoff in the catchment basins of hydro- electric powerplants. The Dominican government has recognized the need 1Garcia, C., 1979. Personal correspondence. Director of Plan- ning of the State Sugar Council, SEA. 2General Secretariat of the Organization of American States, 1969. Survey of the Natural Resources of the Dominican Republic. OAS, Natural Resources Unit, Washington, D.C. TABLE 2 Distribution of Family Income Among Urban and Rural Population _A_. Monthly Percent of Families Average Monthly Income Income (0R8) Urban Rural Urban Rural 0 - 50 29 64 36.0 37.3 50.1 - 100 24 28 79.0 70.5 100.1 - 300 33 8 168.1 135.8 over - 300 14 - 762.3 - SOURCE: Secretariat of Agriculture, Diagnostico y Estrategja del Desarrollo,Agropecuario,_1976-1986. TABLE 3 Annual Labor Requirements of Major Agricultural Products A. Crops ' Man-days per hectare Tobacco 130 Plantain 110 Potatoes 85 Yucca 80 Coffee - 8O Sugarcane 70 Cacao 55 Beans 50 Corn 48 Ricea 45 Peanuts 35 B. Livestock Man-days per animal Dairy Cattle 16.5 Beef Cattleb 4.5 Other Animals 4.0 SOURCE: "Generacion de Empleo Productivo y Crecimiento Economico,“I ILO, Geneva, 1976, Tables 54 and 55. aAverage of rain-fed and irrigated rice. bThe countrywide average stocking ratio is 1 animal unit per hectare. 10 TABLE 4 Yield Trends for Principal Agricultural Products Based on Three-Year Moving Averages: 1962-64; 1968-70; 1974-76 Yield Expected SEA Target with Good 1962-64 1968-70 1974-76 1977a Managementb Sugar 70.9 58.6 60.5 - 80.0 Rice (Paddy) 1.6 2.5 2.1 2.4 1.8- 3.6 Red Beans 0.8 0.9 0.9 1.0 1.4- 2.2 Sweet Potato 8.7 9.3 8.1 8.5 14.4 Corn 1.4 1.8 1.3 1.5 l 8- 2.2 Plantain 4.7 4.4 5.9 7.3 n.a. Cassava 8.4 10.5 7.9 8.0 10.8-14.5 Cacao 0.6 0.5 0.5 - 2.2 Coffee 0.6 0.6 0.5 - 2.2 Tobacco 0.9 1.2 1.0 - 1.4 Pidgeon Peas n.a. na.a 1.8 - 4.3 Peanuts n.a. 0.8 0.8 - 2.2 NOTE: Agricultural products in metric tons per hectare. SOURCE: National Statistical Office; Central Bank; SEA; USAID. aAs stated in the SEA Operative Plan, 1977. bBased on findings of a study team from the International Fertili- zer Development Center which visited the Dominican Republic in 1975. For sugar, based on recommendations of Bookers study. 11 to protect the remaining forest resources and to make this issue a vital element in new resource management programs.1 Several regional studies, such as the DELNO project,2 have em- phasized the interrelation of natural, human and economic resources to improve the utilization of the various production factors in an effort to minimize negative impacts on the resource base. A comprehensive and nationwide resource planning orientation was often jeopardized by lack of spatial information. These facts illustrate the critical need for a continuous and timely capability to gather reliable data on the quality and quantity of the natural resource base. This need is particularly crucial in many developing countries since, in the majority of the cases, the ex- tent and condition of the arable lands, forests, rangelands, and water resources have not yet been adequately assessed. In a time of world food and energy shortages and of spreading environmental degradation due to soil erosion and environmental pollution, it is vitally impor- tant to carry out these inventory and monitoring efforts. The nature, scope, and detail of the required resource informa- tion vary greatly from one country to another. These depend on the current quality (comprehensiveness, detail, and timeliness) of the 1General Secretariat of the Organization of American States, 1969. Survey of the Natural Resources of the Dominican Republic. OAS, Nat- ural Resources Unit, Washington, D.C. 2Secretaria General de la Organizacion de los Estados Americanos, 1977. Republica Dominicana: Plan de Accion para el Desarrollo Re- gional de la Linea Noroeste. OAS, Unidad Technica del Proyecto DELNO, Washington, D.C. 12 data base, the country's development objectives and critical needs for short-term policy development and evaluation. The need for spatial information, as defined above, poses a problem. It relates to the fact that the institutional framework needs to be created to transform spatial data into relevant information for resource management, policy analysis, planning and implementation. This case study addresses one element of this process, the ef- ficient data gathering to procure timely, reliable data to be used by the decision makers (the agencies and institutions involved in resource management and planning). Relatively new techniques are introduced (remote sensing) and aerial sampling procedures are designed, evaluated and compared with an alternate sampling survey approach; area-frame sampling. The CRIES Project The Comprehensive Resource Inventory and Evaluation System (CRIES) project encompasses a cooperative effort between the United States Department of Agriculture and Michigan State University. The major objectives of the CRIES project are: (l) to develop a con- sistent approach to land resource classification adaptable to many countries, and (2) to provide the training and technical assistance to classify and inventory resources, evaluate crop adaptability and pro- ductivity, and develop food strategies in participating countries. The land classification system developed provides the elementary spatial information framework to store, retrieve, update, and cross reference natural resource data as a basis for land use analysis, re- source planning and evaluation. This system involves two land resource 13 components, the resource planning unit (RPU) and the production po- tential area (PPA).1 The RPU is the basic land resource classifica- tion unit: a mapped delineation of physically and environmentally homogeneous land strata used to identify areas in the resource inven- tory system. A PPA is an unmapped delineation within a RPU used to conceptually relate agronomic and economic data to a specific land resource area. RPU's and PPA's are primarily based upon two major underlying taxonomies--soil and crop climate. Soil resources are stratified according to the USDA's Soil Taxonomy,2 allowing for predictions of responses to management practices. Crop climate zones are differen- tiated based on temperature, day length, number of wet seasons, annual precipitation, wet season rainfall, and presence or absence of frost.3 Spatial Information System Component A computer-based spatial information system is used by the CRIES project to store, retrieve and map resource data. The system is grid based; the relevant information is stored in its original form, based on the predominant resource category found within a predefined grid c611 (e.g lkmz). This allows for cross tabulation of data, like land cover/use breakdown by administrative region or RPU. Additional data sets can be added based on the expressed information needs. Derived lUSDA/MSU, 1930. CRIES Report. Special report #1 (draft). 2Soil Taxonomy, A Basic System for Soil Classification for Making and Interpreting Soil Surveys. Soil Conservation Service, USDA, Ag. Handbook No. 436, December, 1978. 3"Crop Climate Taxonomy, A Second Approximation,“ Science and Education Administration, USDA, Unpublished Staff Report, 1978. 14 spatial information might include elements such as water availability, erosion, development potential, crop production costs, etc. to evaluate different planning alternatives and long-and short-term land use policies. Area-Frame Sampling The area-frame sampling concept, used in agricultural inventories, is based on the assumption that reliable estimates (95% confidence level or better) can be obtained for certain population parameters (agricultural statistics) with a relatively small sample at generally less than 10 percent of the cost for a complete enumeration. This in- dicates the usefulness of applying area-frame sampling techniques in lesser developed countries to provide initial estimates at the national level on crop acreages, on number and type of livestock, and more.]’2’3’4’5 1Huddleston, Harold F. et a1., 1979. Use of Landsat Classified Pixels for Estimating Annual Livestock and Crop Inventories. Paper presented at the Third Conference on the Economics of Remote Sensing. Incline Village, Nevada. 2Hanuschak, George, et a1., 1979. Obtaining Timely Crop Area Estimates Using Ground-Gathered and Landsat Data. USDA/ESCS Technical Bulletin #1609 Washington, D.C. 3Cardenas, Manuel, et a1., 1978. On the Development of Small Area Estimators Using Landsat Data as Auxiliary Information. USDA/ESCS Washington, D.C. 4Huddleston, Harold F. 1976. A Training Course in Sampling Con- cepts for Agricultural Surveys. USDA/ESCS, SRS #21 Washington, D.C. 5Craig, M., et a1. 1978. Area Estimates by Landsat: Kansas 1976 Winter Wheat. ESCS, USDA. 15 Area-frame sampling refers to the use of small land areas as the sampling units and probability sampling as the method of selecting these sampling units with a known probability. In order to accomplish this, a design is used which initially stratifies the country in "homogeneous" strata in which the various sample segments, and the sample units contained within them, are delineated on a map base or existing aerial photography.1 These area-sample boundaries must also be identified by the enumerator in the field and the desired informa- tion obtained for the farms or household within the sampling units. In the next stage this sampling information is expanded to provide estimates at the national level. The area-frame sampling methodology just defined can provide the basis for a program of dependable, current statistics on agricultural production, land use, livestock inventories, farm size and number, etc. Various participating countries in cooperation with the U.S. Agency for International Development (AID) are applying the area-frame sampling design to generate the appropriate agricultural statistics. New AID efforts try to extend this technology to other LDC's with an 1US/AID/DR, 1971. Probability Area-Frame Sampling for Data Col- lection. Internal Document US/AID Mission to the Dominican Republic, Agricultural Development Division, Santo Domingo, Dominican Republic. 16 increasing reliance on modern remote sensing techniques (Landsat and aerial survey data) to aid in stratification and primary data collec- tion. Similar efforts are carried out in other parts of the world.“2 The Potential of Remote Sensing From a series of developments of remote sensing techniques (main- ly military oriented), the launch of Landsat-1 in 1972 by NASA repre- sented a major technological advancement in the application of space- born data acquisition to civilian use (see Chapter 2 for technical de- tails). It became possible to record electromagnetic radiation in four different wavelength bands and to display the "spectral signature” of the earth surface on imagery and digital computer-compatible tapes for further analysis. The_synoptic view Landsat scenes cover an area of 185 x 185 mm with a resolution element of approximately 79 x 56 meters (.4 hectare of 1.1 acre is the size of the picture element of "pixel"). They provide a near orthographic perspective well—suited to general natural resource inventory and mapping efforts. An added Landsat-l advantage is its capability to provide repetitive coverage every 18 days. That time interval was reduced to nine days with the launch of Landsat-2 in January 1975. 1Richard Hooley, et a1., 1977. Estimating Agricultural Production by the Use of Satellite Information: An Experiment With Laotian Data. American Journal of Agricultural Economics November 1977, pp. 722-727. 2Mukai, Yukio and Shoji Takeuchi, 1979. Estimation of Primary Production of Vegetation in Agricultural and Forested Areas Using Landsat Data. Remote Sensing Technology Center of Japan, Tokyo. 17 SeVeral studies indicate that air- and space-borne remote sensing makes a critical, positive contribution in many disciplines to the development of our natural resources.1 Because of its usefulness Landsat and other remote sensing data have been applied in numerous cases but especially in LDC's where a limited resource data base necessitates the generation of reliable land cover/use data for the evaluation of development strategies.2’3’4’5 The Michigan State University Center for Remote Sensing, for example, is involved in such efforts in Syria and the Dominican Republic, where nationwide land cover/use information was generated using Landsat data. The Need for Strata Refinement Other Landsat-based inventory studies have been carried out for many parts of the world. In many cases, it was concluded that inter- 1National Academy of Sciences, 1977. Remote Sensing from Space; Prospects for Developing Countries. Washington, D.C. 2Nossin, J.J., 1971, Promotion and Training Aspect of Integrated Surveys. Proceedings of the Fifth International ITC/UNESCO Seminar. ITC, Enschede the Netherlands. 3Broek, J.M.M. Vander, 1969. Integrated Surveys for River Basin Development. Proceedings of the Fourth International ITC/UNESCO Sem- inar. ITC, Enschede the Netherlands. 4Lafortune, Robert, et a1., 1978. Landsat Applications to Land Use Mapping of the Cul de Sac Plain of Haiti, Proceedings 12th ERIM Symposium on Remote Sensing of the Environment, Ann Arbor. 5In INPE, 1978. Use of Landsat Data to Identify and Evaluate Areas of Sugar Cane. Instituto de Pesquisas Espacias, Sao Jose Brazil. 18 pretation accuracies for Level I information (urban, forest, intensive agriculture, range, water) could be expected above 95 percent."2 Level I land cover/use classification defines, to some extent, the practical limitations of the system, based upon the pixel size. Areas exceeding .4 hectare in size are not always sufficiently resolved to allow for accurate visual or computer-aided image interpretation. Another aspect is the lack of sufficientspectraldiscrimination result- ing frbm the variation between and within land cover/use classes. This is especially the case in the small scale farming systems in the tropics. For these reasons it is practical to define major land cover/use (Level 1) categories but not to provide a further breakdown of these categories. It is for example feasible in some cases to differentiate between deciduous and coniferous forest cover (with a decrease in accuracy), but not to differentiate between major vegetation associa- 3 However, a further refinement of in- tions within these categories. formation (crop types, percent irrigated land, percent clearcutting, etc.) can be essential to natural resource inventories, efficient management practices and agricultural sector analysis. This requirement 1Anderson, James R., et a1., Land Use Classification System for Use with Remote Sensor Data, Geological Survey Circular 671. 2A1drich, R.C., N.X. Norich, and 11.9. Greentree, 1978. Forest Inventory: Land-use Classification and Forest Disturbance Monitoring. From: Evaluation of ERTS-l Data for Forest and Rangeland Survey, USDA Forest Serv. Res. Paper PSW-llZ, 67p. Pacific Southwest Forest and Range Experiment Station, Berkeley, California. 3See also, Holmes, Quentin A. and Robert Horvath, 1980. Procedure M: An Advance Procedure for Stratified Area Estimation Using Landsat. Paper presented at the 14th International Symposium on Remote Sensing of the Environment, San Jose, Costa Rica. 19 can only be fulfilled with the aid of medium-scale (l:20,000 to 1:50,000) aerial photography, or detailed, costly ground surveys. The Level I data derived from Landsat can be compared with data developed in the agricultural sector analysis approach of the CRIES project. Data including soils, climate, etc., are classified on the basis of micro-climatological conditions and ecosystem and soil as- sociation factors to form the more or less homogeneous units or (sub) strata (referred to in area-frame sampling). These strata, derived from ground survey methods, are referred to as resource planning units (RPUs). The Landsat-derived strata and the RPUs should coincide at some aggregate level of mapping. In order to improve the detail of spatial information as it re- lates to current resource information, the capability to more ac- curately assess the environmental variables within these categories must be improved. Currently this is being done by means of a costly, time-consuming process, area-frame sampling, with, in a practical sense, unknown estimation accuracies. The research reported here aims to improve population-parameter estimates by using aircraft surveys or aerial sampling techniques and to assess the cost, time and ac- curacy of these techniques in a specific, small-area agricultural in- ventories, in order to refine satellite-derived stratification. Study Objectives The primary objective of this study is to examine the feasibility of conducting random, systematic sampling procedures using aircraft surveys to derive area statistics on agricultural or natural resources. Within the context of this research, special attention is given to multi-stage sampling: the refinement of area statistics derived from 20 space-borne remote sensing systems (a two— or three-stage sampling procedure). Recommendations are made on the developed procedure, its methodology, the sampling size, estimator accuracy and costs. The procedure specifically addresses typical resource information problems encountered in developing countries, such as the lack of a sufficient data base for economic development, insufficient aerial photographic coverage (scale, timeliness and areal extent), and lack of topographic data or other secondary data sources. Major Study Goals Specific goals of the study are: a. The development of appropriate sampling procedures to refine area statistics for strata derived from Landsat satellite imagery or from ancillary information sources, such as resource units (RPUs) used in the CRIES agricultural sector analysis approach. b. The evaluation of the utility of light aircraft surveys and/ or aerial sampling techniques to provide the required information, to develop a low-cost, low-technology, data-acquisition methodology for developing countries. c. The evaluation of the theoretical and empirical sampling errors inherent alternative sampling designs. d. The comparison of the aerial sampling data with area-frame sampling information and the associated potential to estimate crop acreage, given certain accuracy/cost parameters. e. The evaluation of elements relating to non-sampling errors associated with the aircraft survey technique and the area-frame sam- pling design. 21 Researcthpproach The focus of this research is to develop light aircraft sampling procedures to obtain detailed land cover/use information and thus im- prove general resource management practices and provide relevant in- formation for agricultural sector analysis. Figure 1 outlines the sixteen major steps for evaluation of the proposed aerial sampling procedures and a comparison with the area-frame sampling techniques. The selected test site was based on the intensive agriculture category, as delineated using visual analysis of Landsat imagery, the description of RPU's, and the initial stratification used in the de- sign of the area-frame sampling. Aerial photography was acquired over the test site during March 1979 and interpreted based on photo inter- pretation keys developed during the fall of 1979 from the available CIR photography and ground truth data. The land cover/use information was transferred to a l:12,500 enlarged base map, produced from a 1:50,000 topographic map. Area statistics for the various crops with- in the test site and sample segment boundaries were calculated using a dot grid with a density of 1024 dots per square kilometer (l dot = .0977 hectare). These were compared with the area statistics derived from the quarterly area-frame surveys (two segments) in an effort to compare the relative accuracy of the area-frame methods in generating crop inventory statistics. Next (Stage 14) the expansion factor (N/n) was used to predict the total crop area statistics for the counting unit (test site) based on the crop information accumulated during the ground survey methods used in the area-frame sampling. Stages 15 through 18 progress from the generation of sampling statistics and the estimation of land cover/use composition of the test site to a final comparison and evaluation. 22 STRATIFICATION OF INTENSIVE AGRICULTURE 1 FROM LANDSAT DELINEATION OF RESOURCE PRODUCTION UNITS BASED 2 ON SECONDARY DATA STRATIFICATION OF INTENSIVE AGRICULTURE USED IN AREA SAMPLING 3 FRAME DESIGN + T ACQUISITION OF AERIAL PHOTOGRAPHY OVER 5 TEST SITE 1 GROUND TRUTH DATA 6 FOR CROP INVENTORY l DEVELOP IMAGE INTER- 7 PRETATION KEYS 1 TEST SITE SELECTION USING 1, 2 and 3. COR- RESPONDING WITH A COUNT UNIT OF THE AREA SAM- 4 PLING FRAME IMAGERY INTERPRETATION 8 FOR TEST SITE 1 TRANSFER DATA TO 9 BASE MAP 1:50.000 1 CALCULATE AREA STA- ACQUISITION OF GROUND SURVEY DATA FOR APPROPRIATE 11 SAMPLE SEGMENTS 1 COMPILE CROP AREA STATISTICS FOR TWO 12 QUARTERLY SURVEYS 1 13 SEGMENTS COMPARE AREA STATISTICS DERIVED FROM SAMPLE i PREDICT AREA STATISTICS FOR TEST SITE USING 14 AREA EXPANSION FACTOR TISTICS BY FIELDS 10 AND SEGMENTS 1 DEVELOP SAMPLING STRIPS 15 STATISTICS 1 ESTIMATE STATISTICS a | COMPARE EMPIRICAL AND THEORETICAL 1 17 SAMPLING ERROR .7 FOR TEST SITE USING 16 SAMPLING INFORMATION FIGURE 1. Flow Chart of Sampling Vv COMPARE ACCURACY AND COST ASPECTS OF AERIAL SAMPLING WITH AREA 18 SAMPLING FRAME 1 Research 23 During the next phase a simulated strip-sampling approach was conducted using various sample sizes. These sampling strips are based on simulated aerial photographic coverage derived from aircraft sampl- ing procedures at a selected image scale (approx. 1:30,000). The sample sizes selected were approximate samples with an increment of 5 percent, starting at 5 percent up to 20 percent. Also a sample size was selected which is approximately equivalent to the ration n/N that is used in area-frame sampling in order to compare the results with the ground-based survey. Based on the various sampling sizes, estimates were made regard- ing the crop composition of the test site. Empirical and theoretical sampling error were determined using the total enumeration derived from the original inventory and the sampling design. The last stage consists of an evaluation of the cost/accuracy aspects associated with the various sampling procedures in order to make a recommendation on the feasibility of conducting an aerial sampling approach in a typical agricultural inventory in the (sub) tropics. County Description The Dominican Republic occupies the eastern two-thirds of the island of Hispaniola, the second largest island in the Caribbean; the western one-third is occupied by the Republic of Haiti. Hispaniola is located between Cuba, on the west, and Puerto Rico, on the east. The relief features of the Dominican Republic are dominated by four parallel mountain ranges which extend in a northwesterly direc- tion, located mainly in the western part of the country, and a single range of low mountains extending from east to west in the eastern 24 part. The southeast is dominated by a lowland plain suitable for in- tensive agriculture. Soil and climatological conditions vary greatly as a result of the complex topography. The total area is 48,442 km2 with a population of 4.8 million (1975) and a rate of growth of 3 percent over the period 1970-1975. This amounts to a population density of 99 persons/km2 or 200 persons/ km2 of arable land, since almost 50 percent of the country's land mass is cultivated. Other population characteristics (1975) are a crude birth rate of 45.8 and a crude death rate of 13.0 p39_mflllg, A rela- tively high infant mortality of 104.0 239 mille exists. The adult literacy is increasing (51 percent in 1972), as is the GNP per capita ($720.00 in 1975). In an economy mainly based on agriculture, 54.6 percent of the total labor force were employed in that sector in 1970. The major crops are sugar cane, coffee, cacao and tobacco. Total agricultural production has expanded since 1960 at an average annual rate of about 2.5 percent. This expansion resulted mostly from an increase in the area under cultivation. Since almost all the cultivable land is currently in active agricultural production, future growth in agricultural output will have to originate from improved yields, through better management and appropriate land use policies. Although food production has outpaced population growth at an average annual rate of 4.2 percent over the last fifteen years, a large deficit still exists in the country's nutritional needs. This deficit is explicitly 1The International Bank for Reconstruction and Development/The World Bank, Dominican Republic. 1978. Latin America and the Carib- bean RegionaTTOffice, Washington, D.C., U.S.A. 25 referred to by the International Food Policy Research Institute;1 calculations indicate that the total production needs expressed in nutritional equivalents of milled rice are about 620,000 metric tons compared to the total production of food crops of 260,000 metric tons. This difference is partially bridged by an import of 260,000 tons per year, leaving an unfilled nutritional gap of about 100,000 tons. The World Bank study2 indicates that the country's natural re— sources are adequate for the country to become self-sufficient in food production, taking into account diet requirements, and assuming that appropriate agricultural sector policies are followed. Table 5 illustrates a land capability classification and as- sociated production capacity ratings based on information of the Na- tional Statistics Office (DR) and a survey conducted by the Organiza- tion of American States. The figures show that less than 50 percent of the total land area can be classified as having some agricultural production capacity, assuming that careful management practices and conservation measures are used to sustain the long-term carrying capability of the resource base (see Table 6). Description Study Area The study area is located in the eastern Cibao Valley in the North-Central portion of the Dominican Republic. The Cibao Valley represents one of the best agricultural areas producing a variety of crops. The western part is currently being developed by means of 1 ZIBRD/The World Bank, 1978. Dominican Republic. Its Main Economic Development Problems. Latin American and Caribbean Regional Office, The World Bank, Washington, D.C. Ibid. 26 TABLE 5 Land Capability Classification Class Ha. % Cgm Production Capacity I 53,700 1.1 1.1 Excellent for cultivation II 235,000 4.9 6.0 Very good for cultivation 111 312,200 6.6 12.6 Good for cultivation IV 363,900 7.7 20.3 Limited or marginal for cultivation V 607,100 12.7 33.0 Pasture - no erosion hazard VI 561,100 11.8 44.0 Pasture - erosion hazard VII 2,516,100 52.7 97.5 Forest VIII 120,200 2.5 100.0 Wildlife 4,769,300 100.0 Totala SOURCE: Modified after National Statistics Office; and OAS Survey of the Natural Resources of the Dominican Republic. aDoes not include 58,800 ha. in islands, lakes and other un- classified areas. 27 TABLE 6 Land Capability and Conservation Requirements Class Land Capability and Potential Use Conservation Requirements II III IV Cultivable lands, suited to ir— rigation, with level relief and with important limiting factors. High productivity, given good management. Cultivable lands, suited to ir- rigation, with level undulating or smoothly hilly relief. Limit- ing factors not severe and can be compensated through moderately intensive management practices. High productivity, given good management. Cultivable lands, suited to ir- rigation but only with very profitable crops. Level, un- dulating or smoothly hilly relief. Rather severe limiting factors. Moderate productiv- ity, given intensive management practices. Possible crop range restricted. Lands of limited cultivability, not suited to irrigation ex- cept under special conditions and with very profitable crops. Chiefly suitable for pasture or perennial crops. Level to hilly relief. Severe limiting factors. Require very inten- sive management practices. Low to moderate productivity. Lands not suitable for culti- vation, except for ricegrowing. Suitable chiefly for pasture. Very severe limiting factors, particularly in relation to drainage. High productivity for pasture or for rice, sub- ject to very intensive manage- ment measures . Require only good manage- ment practices. Require moderate conserva- tion measures. Require intensive conserva- tion measures. Optimum capability is for tree crops that require little tilling work. Optimum capability is for pasture, without restric- tions. 28 TABLE 6 - Continued Land Capability and Conservation Requirements Class Land Capability and Potential Conservation Requirements Use VI Lands unsuitable for cultiva- Optimum capability is for tion, except for mountain crops. forest and pasture, with Suitable chiefly fer fbrestry restrictions. and pasture. Very severe limiting factors, particularly steepness, shallowness, rock- iness. VII Uncultivable lands, suitable Optimum capability is for only for forestry. forest with severe restric- tions. VIII Lands not suitable for culti- Recreation and wildlife vation. Suitable only for use areas. as a national parks and wild- life areas. SOURCE: World Bank, 1978, Dominican Republic. Latin America and the Caribbean Regional Office, Washington, D.C. 29 infrastructural improvement and a new irrigation project. The crops produced here include rice, soy beans, tobacco, coffee, and some citrus. The eastern portion of the Cibao Valley includes significant tobacco production, north east of Santiago, and mixed, intensive, agriculture to the south. Here the fields become somewhat smaller and include a greater variety of crops. Test Site Description The selected test site is located in the east central portion of the Cibao Valley. It is an area around Moca consisting of level terraces and pediments over water-deposited sediments.1 The climate is generally moist with mean annual rainfall in the 900-l,500 millimeter range.2 The period January through March is considered a distinct dry season. The mean annual temperature is 25° - 27°C, nearly uniform 3 is estimated to have con- throughout the year. The native vegetation sisted of savannah and subtropical broadleaf forest types. The area is extremely productive and well suited for intensive agricultural land use. The major crops consist of plantains, yucca, sweet potatoes, kidney beans, coffee, cacao, tobacco and corn. The terrain is flat with gentle slopes. The climate supports rainfed crops; however, some irrigation can be practiced with increased yield. The topsoil is very 1Arens, P.L. et a1. 1976. "Republica Dominicana: Diversificacion y Aumento de la Produccion Agricola en el Valle del Cibao. CNIECA, San Cristobal, RD. 2CRIES, 1977. Land Resource Base Report 77-1. USDA, A10 and Michigan State University (draft). 3Holdridge, L.R. 1967. Life Zone Ecology, Revised Edition. Tropical Service Center, San Jose, Costa Rica. 3O fertile and contains a high percentage of organic material. Table 7 summarizes the planting/harvesting period for the major crops in the area. 31 ounnop ouumnoh Saunas» am osmm2\xumpm onom\vma mmpzm58wnm: memos aucuwx .P=> mspommmga mpeumm mmoumpoq ummzm mmumamn amoeoaw muz> m>mmmmu .am pogwcmz ocmucmpm :pmucmpm .am mmzz umo >oz boo mum w=< 4:6 23¢ > omnwu :cmummm .wpwm pmm» moo: .Lmucmqu aogu n 39: 32 cowgmq mcwpmm>cmx ------ cowsmn mcwucmpm .mcmowcveoo mowpnaamm .omcwsoo opcmm .9 .oz Fmgzm cowumgummcwsc< mace-umuwpaza .mcmu_cmsoo mowpnaamm up no mo>mupau mocamp< on cowouzcocm mp cm mopssm mommcmcH m moucwwswucmm mop mcaom mcasmwm mu mooam Fm cm ownsmo Poe opumwm pow owuspmm .ofimF .mcagpzuwcm< mu cumumm an m-cmamgumm "mumzom -------- ------ --------- ~sz ceou mxos.mmm .m um: >oz Poo. mum w:< 4:6 zzn > omnwu :smpmmm .muwm pmm» moo: .cmucmpmu noeu uozcwucou - n m4m<-EMISSIVE REGIONS-O- ULTRAVIOLET RADIATION _.Iummumunnum-mum-111mm — _ 0.4 FIGURE 3. RADIATION IIIIIUIUUIIllllllllllllllllllllllllllllll REFLECTIVE INFRARED RADIATION lllllllllllllllllllllllllll VISIBLE In“ I 0.7 1.0 10 20 HAVELENGTH (microns) A portion of the electromagnetic spectrum. (adapted from Johanson, Remote Sensing for Planning Resource Conservation). 1111111 lllllLLll 100 39 The sequence from high to low detail is associated with the following general characteristics: a. an increasing area coverage per unit of time during data acquisition efforts, which reduces inventory cost per area unit and, more important, information accuracy; b. a decreasing reliance on remote sensing as the primary and unique data source, and therefore, an increasing reliance on field data, lower level systems, sampling data or secondary data sources. Selecting the optimum system or combination of systems to be utilized in resource inventory efforts depends on two basic parameters, data acquisition cost and required level of accuracy. In general, one should select any system or combination, which, given budget constraints and accuracy standards, increases the aggregate utility of resource in- formation for efficient management practices. Currently, operational spaceborne remote sensing systems for practical applications are limited to the Landsat series, in operation since the early 70's. The Landsat system, in combination with the large format camera, that will be carried as a part of the payload of future space shuttle missions, are discussed in the following section. Spaceborne systems can be considered the first stage of a comprehensive remote sensing-based inventory effort. Variations in image scale, associated with a multi-stage remote sensing approach, are summarized in Table 8. 4O ooo.o~"H-ooo.mup .ooo.oH-.ooo.H cepmwm .mcwmcm-mHmcmm .mcgoncw< >H ooo.omuH-ooo.o~”H .ooo.om-_ooo.oH vmmcmgoonsau .coamHn .mchcm-cqu .mcconew< HHH ooo.o~HuH-ooo.oonH .ooo.om-.ooo.om conHaaoca you .cmmH .Hm-mm .N-z .mcconcw< HH mmm.mom.muH-omm.wm~.HuH HExV cam-mom mengm mumnm .mcconmomam H ooo.mmm.muH-ooo.ooo.HuH st ommv mmHHE cum ammucmb .mcconmumqm H ooo.omw.NHH-ooo.ommuH HEx mmov mmHHE chm anzxm .mceonmueam H mmHmum Hgmmmew mmvzuHuHm :oHpmcmqo :oHnggummu EgomumHm mmmum HmuHazH Hmowaah mamumxm mchcmm mpoemm Hw>mH-HHH22 so» mmHmum HgmmmEH HmuHaxp w m4moo Hmcon o.~n mHu ooo.omuH -mc .coHazHommc mumcmuoe o.om~ mHu\Hou coo.o~HuH Hue ooo.oov HummmEH peacoch .mHmom mumgmuoe op HHmsm H.¢em m: ooo.omeuH Ex mH manquHm gown ou m>HaHumame .coHuaHommc mumemvoe .3mH> 8H» www.mHm . z omo.o~H «HQ\Hou omm.mmH.HnH Ex wow u LEH .mHupacm mumam .mcwaams uHumEmgu .coHpaHommc mumcmuoe-on .3mH> UH» emm.e Hog ooo.ommnH HHS oHNV -aoesm .oHaom HHaEm zeo> Hom.o_ m: ooo.omm.~ H Ex mme zeomaeH eaHsgm .mchamE qumEmsu .mamu HmuHmHv .mmwcm>ou m>HuHumamE mqumpmum .coHuaHommc 38H .3mH> UH» HHS oumv -aoesm .oHaom HHeEm seo> www.mH -mmz ooo.mem.m H ex owe xeomaeH panacea muHumHempumemgu mama rrH.HE .cmv maxu mHmom musuHuHm zgomwumu mumsm>oo waned mmmEH mmmEH HmcoHpmcmgo ngmmsH mcwmcmm muoEmm mo muHamnguomsmcu m m4m coo.NH H sweep apo> op omeep H.m zoo a... zou Hucaou quoHLma H.m~ mHu ooo.mmpH .eoppsHoaoe gape op m.~m Zea ooo.oe H Hemmeep oppeoeoe .oHeom opaeoeoz p.NH pep\zou wsmcd mmmeH mmeEH Hmcoprcmao HummmsH mchcmm mposmm mo mqumHmeumcmcu nwzcwpcou - m mpmcmz . mnmp nucmz Aa.e~v mNe Am.va woe Ae.e v on A~.e_v mew A -- v o Ao.e V we an.m v mop Am.mmv mam Am.e V cop Am.wpv com Am.e V oer Aa.e v we Ae.e V No, A_.m V cm Ao.e_v NNN Ammcmuumzv umaaocucmucw umucmpn mmmsm» . Ammcmuumzv noeu mchwm .umucmpa mmmsMP . Ammgmuumgv nmaaocucmucw maven» umumm>emz - Ammcmpumgv aosu mpmcmm as» we mmmsmp umumm>cm= . meme eeeeeuao Accumuv mouou Amemuv mmewou Aocmumpav mcwmucmpa Amumummv mmoumuoa ummzm Amus>v m>wmmmu Ammpngumnmzv comm seeeex aNVezv ccou noeo meccmemm aocu Pascc< apeeeee> meme eueez ace mam_ caesaueo om m4mszm Low >gopcm>=~ aoco 95 eeeae _ u a; amec e .Aeese_=ueem< to eeeeeeeeuamv cmx . Aomumuv Amemuv mooou mmeeou Aocmumpav mcwmpcmFa Amumpmmv mmopmpom Amuz>v Ammpmaguwnmzv m>mmmmu :mmm seeeex ANeezv ccou aogu meccmcmm noeu szcc< e_eeeea> .muwm pmmh moo: we» cw emuzpucH emmepe ucmsmmm au>e=m toe agopcm>cH mumF zugmz new mum, LmnEmumo uwzcwucou 1 cm m4memz . 96 mmmp nose: -- -- A_.m,V mom -- -- -- MmeeeeueeV cm aocucmuce .uwpcmpa mmmgm» . -- -- Ao.m~V mes -- -- -- AmeeeeueeV aosu «Pmcem .cmucmpa mmmcmp . -- -- -- Am._V mm Ap.m V mm_ -- Ao.FV op AmeeeeueeV cmggocucmucw mmmgmu uwumm>cmx - -- -- Am._NV mam Am._V om Am.mmV aam -- Ao._ V op AmeeeeeeeV aoeu opmcwm ecu mo «mmmcmp vamm>emx . meme teaseueo AccumuV AmemuV AocmumpaV AmumummV Amua>V Ammpmsgumnme ANszV mpnmmcm> omomu mmmeou mcwmucmpa mmopmpoa m>mmmmo :mmm ccou ummzm amccwx mayo meccmgma aosu pmzcc< mump cute: new mNmF cmnEmumo .muwm pmm» moo: we“ cw umuspucH eomope ucmEmmm xu>czm toe xgoucm>cH aocu Fm m4meoz AccumuV AmmmoV AocmucmFaV AmumummV Amua>V Ammpmzsowasz ANszV omumu mmmeou mcempcmpa mmoumaoa m>mmmmu :mmm ccou uwmzm sucvmx aogu meccmcma auto szcc< mum, eueez new mem_ eeeeeueo .muwm ummp moo: on» cw umuapocH emmm_e acmEmmm >u>ssm eom zgopcm>cH aocu easeeeeeu - _N msmm c< .uwpnsamm cmuwcweoo mew c_ mmwcoucm>cH Fmszupzuwcm< .owmp .mwgmugmx new xcmupzzum ”mumaom .maoso maowcm> Lee mmwumczuum cowpmumcacmucw use chommmc we“ acopm mm=Fm> umcmpsmucsr ~.m_ o.om w.FF w.~e m.~P a. meowmmeEou “.mm m.~ n. m m.N camp .ppzu emzuo up m.mm ~.me m.- m.mm o.m o.m~ maogu emcee up m.mm N.m N.o m.mm mmoumuoa ammzm op o.mm a.mc o.m o.op o.op a.mm o.o¢ ooumno» uF wouz> a.mm m.op ¢.F e.n m.m a.mm m.F mp m.¢m m.m m.m m. m. m.em mcwmgcmFQ

o mm me

c=m Pewsm< mocmwsmaxm mucmwcqum x mmsmuom: N mmgmpum: & mmcmuum: pcmwuweeam cmaeEwV sue; mcwumcacowcH AmcwmpcmFaV AmcwmucmpaV AmcwmpcmpaV a e2; 3533 2: $3; 5%: cmpmmmemm< com Pepe» newsmwm mpasmm acmEmmm m—qum AmcempcmpaV mpcmEmmm mchaEmm 039 one to» name mempasmm osmcdiemg< mm mamou x x x x x x x x x x x x x <1m=ou wame Nance Neeem NNe_m Neaem Neamw Nmem_ Nmemp Nee_, Nwmw Neee Nave NmNN e«ummm0mummM m4 U mmeme Nem_e Noomm Nmpwe makes Nemem Nmoww NNeom ~a_ap Nm~m_ Nemop News mean e«uwmmomukwm ”Seam mammm mmmem mecca aeeem mmmmw macaV mmmmp omee_ emmm omen mpam omen mow, oeamsmmmm m «muzou new mpoum umzcwucou . mm m4m chzpmz w.N~ c.0m a.mp m.mp Afiwp w.- F.mp «.mp m.~F m.m_ new? , umum>mupau cmcpo 0.0 m.F m.P N.o m.~ ¢.o .. i- i- u- mqoco emzpo e.m ¢.¢ N.m F.N v.m ¢.N n.m P.N w.m m.~ mmoumuog ummzm 0.“ m.P w.p m.o m.o P.o o.o F.o i- i- ouomnoh m.mm a.mp w.up ¢.¢P a.mp «.mp m.mp m.ep a.mp ¢.m mou=> m.m~ ~.~N ¢.m~ ¢.o~ m.¢~ m._m m.mm a.mm e.o~ e.w~ mcwmucmFQ xm x xm M xm m xm x xm x seemeeeo an: ee~.a smo.mp sem.e_ smm.m Nee.e \eesao aces prmcwch AuwmcmecH Aummcmch xpwmcmucm zuwmcmpcm mcwpasmm mchaEmm mchQEmm mchaEmm mcepn5mm mpopa ouoza op muopa opoga em mpOFa amaze me muopa omega mm mpopa ouoca op mmcmpucmwpm Lace meepucmwpd Opacwm mmwcomoamu mm:\cm>ou ccmm cmpmwOomm< use mmwpemcmucH mmem mFQEmm mzowem> .mcmwmmo mcwpasmm 03h com mcmmemcma mPQEmm eo secessm meme om memn cowumw>mu cowpmw>mu eteeeeam eteeeeem 0e000000 N.m u m.0 u N.0 u LOLLw Lott—m LOLLm mszomnm weapomnm Op=FO0nm :00: cam: :00: 0._ N.F H N.0 0., e._ M 0.0 0.0 0.F M P.0 ~.m Faeeeeeenem :0 0.0 .... 5: 0.0 0.0 H 0.5 72 mg .H 0.3 0.2 00.28 N.“ 0.0 H 0.0_ 0.0 F.0 M 0.0_ “.0 0.0 H 0.0, N.0 eeeeeeeme> Peeseez 0.0, 0.0 H m.0_ 0.0. 0.0 H 0.~P 0.0. ~.0 H 0.NV N.0N eeep 0000>e0_=0 eaeeo 0._ 0.0 M N.0 N.P 0.0 M 0.0 0._ -- 0._ mecca eeeeo 0.F 0., M _.N 0., N.F w 0.N 0._ 0., M F.N 0.0 meeeeeea 000:0 N.o 0.0 H 0.0 0.0 P.o H F.o 0.0 N.o H P.o 0.F eueeeee N.0 m.m H 0.0F N.N ~.e H a.mp m.0 “.0 H m.0_ 0.0. euc=> 0., m.0 w a.mm m.m 0.0 M m.Fm 0., 0.0 m 0.0m 0.0m neeeeee_a m-.. “a 0 V.“ m-.. K 0 Va m-.. K 0 V.“ a Lorzm Fm>prcm Lot—gm Pm>LmHCw LOLLG _.m>Lmu:.P manpomn< mucmcmecou muzpomn< mucmcmwcou wuzFomn< mocmcmecou some >Mszmucw wcwpgsmm xwwmcmucw wzwnmaem xuwmcmucwmcmanmm cove eeeueea mo.0_ .0eeueea 0~.0F eeeueaa ~0.m -epaaea we mmmuwmcmucH one? 0am.m_ 0:0 c.&0N.0P namm com mmwmmpcmocm RN m4mm— mucmfl wu—EOU Nmmm .Amin V memos cowum~aaoq czocx ea mecccm weapomnm umumV000mm use mc0_ ucmmpe umpompmm zpsoucmg a co vwmmm ”whoz nmzcwpcou 1 mm m4m00000 00000 0.0 0.0 0 0.0 0.0 -- 0., 00000 00000 0.0 0.0 0 0.0 0._ P.0 0 0.0 0.0 00000000 00030 0.0 0.0 0 0._ 0.0 -- 0.0 0000000 0.0 0.00 0 0.0_ 0.0 0.0 0 0.0 0.0_ 0000> 0.0 0.00 0 0.00 0.0 0.00 0 0.00 0.00 000000000 0- 00 0 0 0- 0.0 0 0 00000 00000 0000000000 muapomn< 00>cmucH mucmnwmcou muapomn< 00>cmucH mucouwecou cmmz 0000 me0 0000.000000 0000000000 0000000000 00000000 0000 .Eoucmm 0300 .000000000 apeoucmm mchem .xuwmcmch -003000 mcepasmm ucmucma 00.0 0:000500 uzwucma 00.0 Apm>m0 mucmcwem zuwmcwucH mcw_050m 000.0 0 Lee 00 000V 000000200000 0000000000 mo 000050000 mm ugm<0 121 .Fmp.w N u .o u 00 e0 .00>0p 0020000200 00mm .Axin V 02005 2000000000 2:020 020 00 020220 00000000 0000000000 020 .00200 0200—0 2000 020000020 000500 00005m0000 200202 0 .0200 020000 000200 00000000 00200202 20 00000 "0002 0.0 u 0.0 u 200000>0o 200000>00 02002000 02002000 0.0 u 0.0 u 20220 20220 0000000< 0000000< 2002 2002 0.0 0.0 m 0.0 0.0 0.0 m 0.0 0.0 0000200000: 0. 00 0 .0 0- 00 0 0 20220 20220 0002000000 0000000< 00>20p20 0020000200 00:0o00< 00>2002H 0020000200 2002 0000 0000 0020.000000 0000000000. 00000000P0 00000000 0000 1200200 2000 .000020020 00200200 000200 .000020020 -002000 02000200 0200200 00.0 m 2000500 0200200 00.0 0000020020 m2wpasmm 000.0 m 200 A00>00 0020000200 000V 000000200200 2000000000 mo 000050000 _000000000 - 00 00000 122 TABLE 29 Correct (C) and Incorrect (I) Predictiona of Population Means for Five Major Categories Single Flightline Four Flightlines Four Sampling Intensities Single Sampling Intensity Test Site Categories 4.76% 9.52% 14.26% 19.03% 4.76% Plantains C C C C C Yucca C C C C C Other cultivated land I I I I C Natural vegetation C C I I C Forest I I I C I 695% confidence level These results clearly indicate -- as does the smallest absolute error for all crops combined, associated with the second 4.76 percent sampling intensity -- the improved accuracy of a multiple flight line sampling approach. When sampling unknown populations the second sampling design is preferred, since it results in error reduction. Another element which plays a role in the sampling procedure is based on the capture of in- fbrmation of various sampling plots within a single flight line: the aspect of independence between the individual plots. The standard error formula M’Il-f)5x2/n used in constructing the confidence inter- vals is based on an assumption of n independently selected elements (photo plots). If a positive correlation between the photo plots 123 within the clusters (the l6 elements of the strip sample) exists, then this formula would underestimate the actual standard deviation. This in turn would produce estimates with confidence intervals too narrow to have the prescribed confidence level of 95 percent. The independence of the photo plots contained within a single sampling strip or cluster can be tested by computing the intracluster correlation coefficient p.1’2 In this case p was calculated for plantains based on four clusters each containing l6 sample units. The following formula can be used (Jessen, l978): <~—.-‘-> D- —-W_- where M2 2 .1.1(Yij'§ .].1 1 : =.I :1: J: TS NMl M Z M - 2 2 (Y.. - Y.) 1 j=l 13 1 N(M-l) M2 1 number of photo strips number of photo plots per photo strip mean square of the total mean square between photo strips mean square within photo strips i' = the observed value for the jth photo plot within the ith 3 photo strip = mean per photo strip = mean of all photo plots -ou meas Romy mgmpme mmom * mmom to game mumm x mnmo ”25 own; .55 mm x mm .umngcmc mcwuczoe meumam .uczocm -cgzu 3o_m .umgwacmg “cwsamacm Pneumam .mmwpmpwumm umpwswo .mego pneumam «HQ Lo ommm m2 meocsuooxm mxwp .mPFog magm— cw gopou Lo mHu .Agopouv mppog mgzmoqu NP Lo» noow ._mu:msmcma5mw .mpnmwcm> .mcmcwmgu uwmwumam mwcwzcmc .umumUWFqsou ago: .3opm cam umpmow~mn .Eou .cwmuno op ppzuwmmwu magma can mum>cwm .xpymou >Pm>wumLmQEou .oooxmm Lm>o .mxmuws cmpm « Nnmp Lo Apmmmv coon w some mEE omnw .55 cm « mm .gozopm mppoc mczmoaxm omm .mmmmmooga new wwwuwpmumm ccmccmpm mcwma ccaogmcgsp xuwso .oom.Pm mo emcee Ezswcws mw> mpnmpwm>m xpza cw .uoom EFW$ «Ho copou vcm 3\m cw mppog mgzmoaxm omuom Lo; pamppmoxm .gmwz .umcwacwc mcwcwmgu vmpwewp .mpaswm xpw>wympmm .ummm vcm Ammo apm>wmemm .mpnmpwm>m apwcmmg magma vcm mow>cwm .pmoo 30F zpw>wumgmasou .ooo.~m Lmucs mcmp wcmucmum saw: “mom ooo.op um mmmgm>ou mms< meLow xgmmmsH mcwmmmuogq EFT; sowpwaach>m spec sowpwnaepoc Pacowoacaao mmmm ucm mcwcwmcp pmcomumgmqo mcvmamm Emumzm mumpasoo umou mpmewxogaa< 25 on SE mm mcaummd AwNmFV mEmumxm mLmEmu 55 cu can 25 mm mo comwgmasou cm m4mvm quwpcmuH .mpnvmmoa cowumanqu noum N\F xpumoz .Anwummc covpmgnwpuuv .umm oo-\_ o» a: mcwsgo ..umm oom\~ umpnpwmmm: .ucoumm can mmsmgw om cu u: gonna .ucoummxm5mcy _ umpnmemm: .mcwaams new acmme> Low mmquanwmmoa coupon co msmm .mpnmmmoa compoau -ocawg coon mmxme Epwm mmmm m: we cowuapommc acmppmoxm .cowgumwoca gumcwu no» Loaumnogn meumam .mmgau lam» cmmggmu pmos mo minmmoa cow» -mgoccm .Aaomv anucm gmaoga saw: mqwcum cw anwmmoa mcwzww> omcmum .mmmcmgu Easwcws mcw>Fo>cw mmcaumuoca covuusuogamg meomam .xgamgmoposa mag wcauwupm zap >gm> Low mmmucm>u< .aoum «\P cacao macaw Fpam :oXVz .muoaamvm .omm oom\p appmcmcmm .umm ooo~\F on a: .umm oom\F .mumacmvm ucoumm \mEmgm F mcowamuwpnam pmgmcww .ccoumm Log mmsmgm wnm appmzm: .umm: mm Ep_m cmmgm mcwm mm mpnwmmoa game -mmcmpcm Auwpmzc coca ”Louumnoga mcwuumpwmg ncm Louumwogg mcwumgu .cmummc mcuWwocowEv gouumnogg mcwmgmpcm Eogm opnwmmoa mcwnams ucm acmzmw> .owpmsmpnoca cowamu loccm Amgos Lo aomv ampccm manna -mum saw: mpnwmmoa mcwzmw> omgwum .mpnmpwm>m apwmmm cowumgaw_mu mgngma< cmmgm gmupacm mswu mcwpuxumg owgamgaopocg mwm>_mcm mmmEH cowuuavocamg mmmEH as on 55 mm mczummu Awanv mamngW mLmEmu as on can 55 mm mo umscmucou 1 cm MAm916 ABOVE GROUND LEVEL (AGL H-h FIELD OF VIEW a NOTE: VARIATION IN TERRAIN ELEVATION WILL CAUSE SCALE VARIATION. HEIGHT ABOVE S L MEAN SEA LEVEL (MSL) FIGURE 14. The relationships between camera distances and flying height for vertical and oblique aerial photographs. 157 image opening in the fecal plane of the camera) would theoretically increase the ground coverage as well. Since this is standard for a specific camera, it is only a factor in comparing the utility of camera systems and in determining the actual cost of data collection per unit area. The §g21e_of a photographic image is determined by the ratio fly- ing height (Above Ground Level)/focal length. The determination of the AGL height can be problematic since terrain elevation varies. The plane altimeter, adjusted for the current barometric pressure, indicates the altitude above Mean Sea Level or MSL (see Fig. 14). By subtracting the average terrain elevation, the absolute altitude (AGL) can be cal- culated and used in combination with the focal length to determine the scale. Ground resolution is a measure of the ability Of an imaging system to separate adjacent objects. This ability can be measured in lines per mm, or the least separation in mm. being the width of a just-resolved line pair at the ground scale. The resolution varies with film and lens characteristics and can be affected by image motion at the time of exposure. Effort has to be made to optimize the ground resolution using fine grain films, high quality lenses and adequate short exposure time to prevent any image blur caused by image motion (usually 1/500 sec. is adequate). The actual linear resolution of an image can be determined by means of bar targets with varying line widths. The geometry of a photo image is perspective; a two-dimensional reproduction of a three-dimensional image. This also is the case for a vertical photography where varying terrain heights lead to radial 158 displacements. This characteristic is of great advantage since two overlapping images can be fused to a stereo model allowing three- dimensional viewing, height measurements, contour-mapping, etc. The typical vertical image perspective can be distorted by movements of the camera platform along the three axis of the aircraft resulting in pitch, roll and yaw movements of the camera focal plane. More precise photographic applications, like topographic mapping, therefore require gyro stabilization of the camera mount. Films and Filters The selection of the appropriate film-filter combination is an essential element in remote sensing information gathering, since it determines which portion of the electromagnetic spectrum is used to highlight selected features of interest. Panchromatic (PAN) or black and white film is sensitive to the visible portion of the EMS (400-700 nm.) and records reflected energy in various gray_tones. The use of a light yellow filter (Wratten No. 15) is recommended to reduce the effects of haze at lower flying levels ( 5000 feet). Higher altitudes require an increased capability to reduce the blue (haze) wavelength and in those cases a red filter (Wratten No. 25) is recommended. In all cases careful consideration should be given to the filter factor in determining the required lens aperture and exposure time. A filter factor of 8,(23) requires that the normal f-stop setting be opened up three stops. The general advantage of the use of PAN-film is the relatively low cost of film processing, printing and reproduction, and the pos- sibility of correcting improper film exposures during the printing process. 159 Black—and-white infrared film (IR) records reflected energy in the 700-900 nm. band of the EMS. Its principal advantages over panchro- matic film are the increased capability to provide contrast and excel- lent haze penetration, revealing detail not visible to the eye under limited visibility conditions. A typical use of infrared (IR) film is derived from its capability to register a wide range Of infrared reflectance of chlorophyll concentrations in vegetation cover, allowing for species differentiation and mapping. Typical filters used to eliminate the blue and ultraviolet radiation are Wratten 25, 29 or 70. B/W Infrared film is available in 20 exposure rolls for 35 mm. cameras. An important aspect limiting this film's use is the fact that it must be loaded in cameras in total darkness. In contrast to panchromatic film, correct exposure of infrared film is critical. Also, exposure meters, capable of measuring infrared radiation, have only been developed on an experimental basis and are definitely not available in the local photo shop. Therefore, an ex- posure test flight is recommended to determine different haze conditions at different altitudes in order to develop an exposure calibration chart for regular applications using certain film/filter combinations. Color film offers a greater advantage in aerial applications than B/W film, since it has a tonal range of natural color, a distinct benefit in photo interpretation. Color film consisting of three emul- sion layers sensitive to blue, green and red light, is sensitive to the complete visible light range (400-700 nm.) of the EMS. But variation in atmospheric conditions (haze and pollution) does affect the quality of the photo image, to a large degree, limiting the period of data 160 collection to good or excellent flying weather. Only for low altitude applications ( 2000 feet or 600 meters) can reasonable results be expected if the visibility is in the 7-15 miles range. The sensitivity to haze conditions of color film requires the use of an ultraviolet (UV) filter. For visibilities better than seven miles, a Wratten 1A is recommended. Flying altitudes over 5000 feet with visibilities less than seven miles require a Wratten 28 or an HF-3 aerial filter. It is important to distinguish between negative and positive (or reversal) color film. The color balance and the density variation of negative color film can be controlled during the printing process, which is a great advantage. Color negative film has a greater exposure latitude and versatility than reversal film. Negative color film can be used to produce color and black-and-white prints, slide or large transparencies. Positive color film produces transparencies or slides and allows for contrast enhancement by means of making an internegative. This requires special processing facilities but can "save" a mission flown under bad (haze) conditions. Film availability generally does not pose any problems, especially for 35 mm cameras. Color film for 70 mm cameras (120) has to be special ordered for use in lS-foot camera cassettes (Ektochrome MS 5256). Color infrared film (CIR) combines some of the advantages of infrared and color film. CIR consists, like color film, of three layers sensitive to green-red light and to reflected infrared radiation (700-900 nm.). The resulting images display therefore false natural £91955, recording healthy vegetation (large chlorophyll concentrations) as red, red as green and green as blue. The combined advantages Of 161 infrared and color film have made CIR film an invaluable data recording tool in vegetation studies, crop stress monitoring, pollution detection and land cover/use mapping efforts at a regional scale. CIR-films have to be exposed with a yellow (minus-blue) filter to eliminate the blue light scattered by the atmosphere, especially during hazy condi- tions. A Wratten 12 filter is recommended. Exposure charts should be developed for different flying conditions. It is important to consider the variation in reflected radiation of various surface features and ground cover, especially for flying altitudes below 2000 feet. The exposure latitude of CIR film is f 1/2 f-stop. Light Aircraft and Camera Mounts Light (single engine) aircraft have become increasingly valuable in aerial surveys and environmental monitoring practices. The choice of aircraft for large format photography is affected by operation cost/hour, the total flying time required, the complexity of the navigation equipment needed, the type of photography to be carried out and the maneuvering required. Most single engine aircraft have an operation cost (plane and pilot) within the $30-$60/hour range. The time saved by higher Operat- ing speeds of high performance single engine aircraft generally offsets the higher operation cost/hour. Since most of the photography is carried out with good-excellent VFR (Visual Flight Rules) conditions at target areas within two hours flying time of the departure point, at IFR (Instrument Flight Rules) equipped aircraft is not required by certainly useful in order to optimize the overall utility. 162 High wing aircraft are preferable for oblique photography and, using exterior side mounts, for vertical photography. The use of helicopters as a camera carrying platform could be considered if special requirements have to be met. Its great advantage is the capability to fly at very low airspeeds and very low altitude, allowing fOr extra-large scale photography of spot locations. In mountainous terrain this can be very important where facilities are limited for fixed-wing aircraft and where up and_down drafts make flying at low altitudes hazardous. There are, however, a number of significant disadvantages associated with helicopter use: operation cost/hour is 4-6 times that of fixed-wing aircraft, aircraft vibrations require high shutter speeds for mounted cameras and the payload is generally rather limited. Light Aircraft Mounts Several applications of 35 mm. and 70 mm. photography, using light aircraft camera mounts for exterior use, have been carried out. The "Montana System," one of the more recent and successful systems, was developed by the Remote Sensing Laboratories of the University Of Minnesota. A standard motor driven camera in an exterior side mount is clamped on the door Of a high wing aircraft. This has the advantage of low fabrication cost, the ease of reloading and the fact that it does not require modification of the fuselage, door or window of com- patible highwing aircraft (e.g. Cessna). Experience during various applications by personnel of the MSU Center for Remote Sensing have made it possible to formulate general specifications for a camera side mount: 163 a. Optimize adaptability of the mount without aircraft modifi— cation. b. Provide for in-flight leveling. c. Keep procedure simple for long flight-time operations. d. Make camera platform retractable for camera reloading and calibration. e. Make monitoring configuration standard to allow for con- sistent near-vertical coverage with lenses of various focal lengths. f. Provide for a crab-correction capability (camera rotation). 9. Provide for multicamera applications. h. Keep construction cost low. The School of Forestry of Austin State University, Texas developed a mount with capabilities to meet some of these specifications (Mason, 1978). A light weight versatile camera mount was designed for external attachment without aircraft modification. The basic mounting frame consists of aluminum tubing supported at the corners and stress points by sheet aluminum, with overall dimensions of 18" x 14" x 6". The mount may be attached to the door of a high-wing light aircraft (Cessna 172), by securing it through the open window by means of two swivel turn clamps and additional support of a number of heavy duty suction cups (Fig. 15). This basic design has been modified by the author (See Fig. 8). This new design makes single or multiple camera opera- tions possible with remote control from the airplane (intervalometer or manual). The camera platform pallet is attached to the mounting frame by means of three spring/bolts assemblies which allow for in- flight level adjustment. Rotation of the camera fixture in relation to the lower pallet makes crab correction during various wind condi- tions possible in order to improve coverage during flight-line photography. 164 .Egompmpa Lozop .Emumxm newsmo EEON\mm com cowpmgzmwmcoo uomwogn mcwmcmm macaw; :mz .ummgugwm “sump Low uczoe newsmu .m_ mmstd 14 \I. D.C. 11 e." W 1.11 :r: I flu Hmmluaaguu 9 . E a. _ . Soc :22: \\ \N. w :03 P JITH is, It \ APPENDIX 2 PHOTO INTERPRETATION KEY FDR SELECTED CROPS IN MOCA REGION In the Moca region, none of the cultivated crops to be identified are grown in fields having sufficiently unique shapes, sizes, sites, or associations to allow discrimination between them. These particular at- tributes will be generalized tO apply to all of the crops. The dis- tinguishable characteristics of pattern, shadow, texture, and tone will then be detailed for each crop. The interpretation key is for the Moca region only. The same crops in other regions may have different char- acteristics. Shape of Field: Variable. Usually regular to irregularly shaped polygons. Some fields, however, may have curvilinear boundaries that conform to the topography. Size of Field: Variable. Plantains sometimes are grown in larger fields than the other crops but there are many smaller fields of this plant, too. Site: All of these crops are preferentially planted on level to gently undulating topography. Association: None of the identified crops are associated with any particular location a production method. Cacao, how- ever, is often grown under large trees which provide it with shade. 165 ?’ DOW F. V. 166 Classifications Plantains - Platano* (Musa sp.) Yucca (Cassava) - Yuca* (Manihot sp.) Tobacco - Tobaco* (N, tubacum) Sweet Potatoes - Batatas* (Iopmoea batatas) Other Crops Cultivated Land* Natural Vegetation ** In the Moca region most of the natural tree cover is Guama (local name). Plantains (Platano) Pattern: Plantains are usually planted with a regular geometric arrangement and in the early stages of growth this is the crop's most distinguishing characteristic. A square pattern is most common but diamond, rectangular, or triangular designs exist. When the crop is mature, the foliage from adjacent trees tends to merge and obscure the pattern somewhat. Fields which have been cut and allowed to resprout exhibit an even more indistinct arrangement. *Covers fields being prepared for crops at the time of the photo- graphy. No identifiable crop is visible. **Includes: forested land herbaceous and shrub rangeland abandoned fields pasture Shadow: Texture: Tone: 167 On large scale photographs the distinctive shadow cast by the plant is observable within most fields of intermediate growth and along the edges of fields containing mature crops. In the intermediate and mature stages of growth a coarse texture predominates. In the early stages of growth the tone is dominated by the characteristics of the topsoil. Black and White: Throughout its growth the crop is a mixture of Color: light and dark tones, the dark predominating, until the crop is cut down and allowed to regenerate. Then the light-toned portions (litter on the surface) predominate. The crop has a green to dark green color. Second growth fields have distinct white portions representing the litter. ColorInfrared: Varies from pink to red. White portions exist in second growth fields. Sweet Potatoes (Batatas) Pattern: Shadow: Texture: Tone: No particular pattern is evident in this crop because the plant covers the whole field uniformly. However, on very large scale photography very recently planted crops will have a row pattern. No shadow effect. Sweet potatoes have a fine or velvety texture very soon after planting and throughout the other growth stages. In the ygry early stages of growth, tone is dominated by characteristics of the topsoil. 168 Black and White: This crop is light—toned at all stages of growth. Color: Green. Color Infrared: Bright pinkish-red. Tobacco (Tobaco) Pattern: Like Yucca, a row pattern is evident in the young and intermediate stages of growth. Unlike Yucca, a faint row pattern Often persists even in mature fields. Shadow: Shadow is not a viable component in the identification Of tobacco. Texture: A coarse texture is apparent in both intermediate and mature stages of growth. However, the texture at mature stage can be described as stippled. Tone: Tone is dominated by characteristics of the topsoil in the early stages of growth. Black and White: The crop is imaged in medium to light tones. Color: Medium green. Color Infrared: Red. “Yugga (Yuca) Pattern: For young and intermediate stages of growth a row pattern is evident. When the crop matures, however, the inter- mingling branches and leaves of adjacent plants eradicate any pattern. Shadow: Mature plants may cast a small shadow at the edge of a field but it is apparent only on quite large scale Photo- graphy. 169 Texture: Intermediate stages of growth evidence a texture similar to worn corduroy. Mature fields have a medium or wool-like texture. Tone: In the early stages Of growth tone is dominated by characteristics Of the topsoil. Black and White: This crop is imaged in light tones. Color: Light green. Color Infrared: Bright pink to medium red. BIBLIOGRAPHY BIBLIOGRAPHY Aldred, A.H. and J.J. Lowe, 1978. Application of Large-Scale Photos to a Forest Inventory in Alberta. Forest Management Institute, Canadian Forestry Service. Ottawa, Ontario. Aldrich, Robert C., et a1., 1977. Inventory of Forest Resources (including water) by Multi-level Sampling. Rocky Mountain For- est&Range Experiment Station, Forest Service/USDA, Fort Collins, Colorado. Aldrich, R.C., N.X. Norich, and W.J. Greentree, 1978. Forest Inven- tory: Land-use Classification and Forest Disturbance Monitor- ing. From: Evaluation of ERTS-l Data for Forest and Range- land Survey, USDA Forest Service Research Paper PSW-llZ, 67 p. Pacific Southwest Forest and Range Experiment Station, Berkel- ey, California. Anderson, James R., et a1., Land Use Classification System for Use with Remote Sensor Data, Geological Survey Circular 671. Arens, P.L. et a1., 1976. "Republica Dominicana: Diversificacion y Aumento de la Produccion Agricola en el Valle del Cibao. CNIECA, San Cristobal, RD. Barrett, J.P. and Philbook, J.S., 1970. Dot Grid AreaEstimates: Pre- cision by Repeated Trials. Journal of Forestry 68 (3),149-151. Bonnor, G.M. 1977. An Evaluation of Systematic Sampling in Malay- sia Forest Inventories. The Malaysian Forester. Vol. 40, No. 4. Bonnor, G.M. l975. Cluster Sampling with Large-Scale Aerial Photo- graphy in Forest Inventories. Canadian Forest Service, Forest Management Institute. Ottawa, Ontario. Inf. Rep. FMR-X-BO. Bonnor, G.M., 1975. The Error of Area Estimates from Dot Grids. Canadian Journal of Forestry Research, 5, 10, 1975. Brandenberger, A.J., 1976. "World Cartography," United Nations. Broek, J.M.M. Vander, 1969. Integrated Surveys for River Basin De- velopment. Proceedings of the Fourth International ITC/UNESCO Seminar. ITC, Enschede,The Netherlands. Cardenas, Manuel, et a1., 1978. On the Development of Small Area Estimators Using Landsat Data as Auxiliary Information. USDA/ ESCS Washington, D.C. 170 171 Chappelle, Daniel E., 1976. How Much Is Information Worth? Proceed- ings: Resource Data Management Symposium. Purdue University, West Lafayette, Indiana.. Claire, M. Hay, 1974. Agricultural Inventory Techniques with Orbital and High-Altitude Imagery. ASP Journal, pp. 1283-1293. Clement, J., 1973. Utilisation des photographies aeriennes au l/5000 en couleur pour la detection de l' okoume dans la foret dense du Gabon. Proceedings IUFRO, Frerburg, W. Germany. Cochran, W., 1977. Sampling Techniques. Wiley and Sons, New York. Consejo del Estado Azucar (Secretariat of Agriculture). Craig, M., et a1., 1978. Area Estimates by Landsat: Kansas 1976 Winter Wheat. ESCS, USDA. CRIES, 1977. Land Resource Base Report 77-1, USDA, A10 and Michigan State University (draft). "Crop Climate Taxonomy, A Second Approximation," Science and Education Administration, USDA, Unpublished Staff Report, 1978. Draeger, William C., 1974. Test Procedures for Remote Sensing Date. Journal of the American Association of Photogrammetry. pp. 175- 181. Fowler, Gary W. and Carl F. Davis, 1979. Sampling Natural Resources Populations: Mutually Exclusive Fixed-Area Sampling Units. Resource Inventory Notes. BLM-23. USDI. Bureau of Land Manage- ment. Denver, Colorado. Garcia, C., 1979. Personal correspondence. Director of Planning of the State Sugar Council, SEA. "Generacion de Empleo Productivo y Crecimiento Economico," ILO, Geneva, 1976. General Secretariat of the Organization of American States, 1969. Survey of the Natural Resources of the Dominican Republic. OAS, Natural Resources Unit, Washington, D.C. Hanuschak, George A., 1977. Pilot Study of the Potential Contributions of Landsat Data in the Construction of Area Sampling Frames, USDA/SCS, Washington, D.C. Hanuschak, George, et a1., 1979. Obtaining Timely Crop Area Estimates Using Ground-Gathered and Landsat Data. USDA/ESCS Technical Bulletin #1609, Washington, D.C. Hay, Alan M., 1979. Sampling Designs to Test Land-Use Map Accuracy. Photogrammetric Engineering and Remote Sensing, Vol. 45, No. 4, April 1979, pp. 529-533. 172 Heinsdijk, D., 1953. Begroeiing en luchfotografie in Suriname. ITC. Enschede, The Netherlands. Publ. 12. Hoffer, Roger M. and Michael D. Fleming, 1974. Use of Computer-aid- ed Analysis Techniques for Cover Type Mapping in Areas of Moun- tainous Terrain. Paper presented at the 14th International Congress of Surveyors. Washington, D.C. Holdridge, L.R., 1967. Life Zone Ecology, Revised Edition. Tropical Service Center, San Jose, Costa Rica. Holmes, Quentin A. and Robert Horvath, 1980. Procedure M: An Advance Procedure for Stratified Area Estimation Using Landsat. Paper presented at the 14th International Symposium on Remote Sensing of the Environment, San Jose, Costa Rica. Hooley, Richard, et a1., 1977. Estimating Agricultural Production by the Use of Satellite Information: An Experiment With Laotian Data. American Journal of Agricultural Economics, November 1977, pp. 722-727. Hord, Michael R. and William Brooner, 1976. Land-Use Map Accuracy Criteria. Photogrammetric Engineering and Remote Sensing, Vol. 42, No. 5, May 1976, pp. 671-677. Houseman, Earl E., 1975. Area Frame Sampling in Agriculture. Statis- tical Reporting Service, SRS #20, USDA, Washington, D.C. Huddleston, Harold F., 1976. A Training Course in Sampling Concepts for Agricultural Surveys. USDA/ESCS, SRS #21, Washington, D.C. Huddleston, Harold F., et a1., 1979. Use of Landsat Classified Pixels for Estimating Annual Livestock and Crop Inventories. Paper presented at the Third Conference on the Economics of Remote Sensing. Incline Village, Nevada. INPE, 1978. Use of Landsat Data to Identify and Evaluate Areas of Suger Cane. Instituto de Pesquisas Espacias, Sao Jose, Brazil. Instituto de Pesquisas Espacias, 1978. INPE Crop Survey Program Using Combined Landsat and Aricraft Data. INPE, Sao Jose, Brazil. International Bank for Reconstruction and Development, The/The World Bank, Dominican Republic, 1978. Latin America and the Caribbean Regional Office, Washington, D.C., U.S.A. International Bank for Reconstruction and Development/The World Bank, 1978. Dominican Republic. Its Main Economic Development Prob- lems. Latin American and Caribbean Regional Office, The World Bank, Washington, D.C. Itek, 1978. Large Format Camera, Description and Applications, Itek Optical Systems, Lexington, Mass. 173 Jensen, Mark S. and Merle P. Meyer, 1976. A Remote Sensing Applica- tions Program and Operational Handbook for the Minnesota Depart- ment Of Natural Resources and Other State Agencies. Remote Sensing Laboratory, University of Minnesota, St. Paul. Jessen, Raymond J., 1942. Statistical Investigation Of a Sample Survey for Obtaining Farm Facts. Iowa State University, Re- search Bulletin 304, Ames, Iowa. Jessen, R.J., 1978. Statistical Survey Techniqies. Wiley and Sons, New York. Kalensky, Z. and L.R. Scherk, 1975. Accuracy of Forest Mapping from Landsat Computer Compatible Tapes. Proceedings of the 10th International Symposium on Remote Sensing of the Environment, Ann Arbor, Michigan. King, A.J., and R.J. Jessen, 1945. Master Sample of Agriculture. Journal of the American Statistical Association,, Vol. 40, pp. 38-46. Kirby, C.L. and P.I. VanEck, 1977. A Basis for Multi-stage Forest Inventory in the Boreal Forest Region. Paper presented at the Fourth Canadian Symposium on Remote Sensing, Quebec. Lafortune, Robert, et a1., 1978. Landsat Applications to Land Use Mapping of the Cul de Sac Plain of Haiti, Proceedings of the 12th ERIM Symposium on Remote Sensing of the Environment, Ann Arbor, Michigan. Langley, Philip G., 1969. New Multi-Stage Sampling Techniques Using Space and Aircraft Imagery for Forest Inventory. Proceedings of the Sixth Symposium on Remote Sensing of the Environment, ERIM, Ann Arbor, Michigan. Langley, Philip G., 1976. Sampling Methods Useful to Forest Inven- tory When Using Data From Remote Sensors. Paper presented at the SVI IUFRO World Congress, Oslo, Norway. Lund, H. Gyde. Uniformly Distributing Sample Within a Type Island. USDI Bureau of Land Management. Denver Colorado. Michigan Land Cover/Use Classification System, 1975. Howard Tanner, Director of Land Resource Programs, Michigan Department of Na- tural Resources. Mukai, Yukio and Shoji Takeuchi, 1979. Estimation of Primary Pro- duction of Vegetation in Agricultural and Forested Areas Using Landsat Data. Remote Sensing Technology Center of Japan, Tokyo. NASA, 1976. Data Users Handbook, Goddard Space Flight Center, Green- belt, Maryland. National Academy of Sciences, 1977. Resource Sensing from Space; Prospects for Developing Countries, Washington, D.C. 174 Neil, Thomas E., et a1., 1973. Area Handbook for the Dominican Republic. Nielsen, U., et a1., 1979. A Forest Inventory in the Yukon Using Large-Scale Photo Sampling Techniqies. Forest Management Institute, Canadian Forestry Service, Ottawa, Ontario. Nossin, J.J., 1971. Promotion and Training Aspect of Integrated Surveys. Proceedings of the Fifth International ITC/UNESCO Seminar. ITC, Enschede, The Netherlands. Ondrejka, R., 1979. Personal correspondence. System Development Manager, Itek Corporation, Lexington, Mass. Perring, W. Edwards., 1960. Sample Design in Business Research. Wiley and Sons, New York. Sader, Steven A. and Robert W. Campbell, 1979. Cost and Accura- cies of Tropical Land Cover Mapping Using Landsat and Medium Scale CIR Aerial Photography: A Costa Rican Example. Pro- ceedings of the Third Conference on the Economics of Remote Sensing, Incline Village, Nevada. Sayn-Wittgenstein, L. and A.H. Aldred, 1969. A Forest Inventory by Large-Scale Aerial Photography. Pulp Paper MagaZTne. Canada, September 5. , Sayn-Wittgenstein, L. et 61.. 1978. Identification of Tropical Trees on Aerial Photographs. Canadian Forest Service, For- est Management Institute. Ottawa, Ontario. Inf. Report FMR-X-ll3. Schultink, G. and Karteris, Michael A., 1980. An Evaluation of Photographic Imagery Parameters Relating to Agricultural In- ventories in the Dominican Republic. Draft report MSU/USDA. Schultink, G., M. A. Karteris and R. Hill-Rowley, l979. Cane Rust Damage Assessment in the Dominican Republic. The Use of Light Aircraft-Small Format Photography in Samll Area Agri- cultural Inventories. Paper presented at the Third Symposium on the Economics of Remote Sensing, Incline Village, Nevada. Secretaria de Estado de Agricultura, 1975. "Economica Agropecuria," Vol. I, No. 1, Departamento de Economia Agropecuria, SEA, Santo Domingo, Republica Dominicana. Secretaria de Estado de Agricultura, 1976. Estudio del Efecto del Cambio en al Epoca de Siembra Sobre los Rendimientos e Ingre- sos Bruto en la Produccion de Algunos Cultivos en la Republi- ca Dominicana. Publicaciones Administracion Rural No. 6. Santo Domingo. 175 Secretaria General de la Organizacion de los Estados Americanos, 1977. Republica Dominicana: Plan de Accion para el Desarrollo Region- al de la Linea Noroeste. OAS, Unidad Technica del Proyecto DELNO, Washington, D.C. Secretariat of Agriculture, Diagnostico y Estrategia del Desarrollo Agropecuario, 1976-1986. Soil Taxonomy, A Basic System for Soil Classification for Making and Interpreting Soil Surveys. Soil Conservation Service, USDA, Agriculture Handbook No. 436, December, 1978. Stellingwerf, D.A., 1966. Interpretation of Tree Species and Mix- tures on Aerial Photographs. International Institute for Aerial Survey and Earth Sciences, ITC, Enschede, The Netherlands. TA/AGR/ESP, 1976. A Proposal for a Comprehensive Land and Water In- ventory and Evaluation System for Agricultural Planning. Taaffe, Kenneth E., 1979. Computing Optimum Plot Size for Wildland Inventories. Resource Inventory Notes, BLM—23 USDI, Bureau of Land Management, Denver, Colorado. Taranik, J.V., 1978. Characteristics of the Landsat Multispectral Data System, U.S. Department of the Interior, USGS, Sioux Falls, South Dakota. Tomlinson. Geographical Data Handling. US/AID/DR, 1971. Probability Area-Frame Sampling for Data Collection. Internal Document US/AID Mission to the Dominican Republic, Agricultural Development Division, Santo Domingo. USDA/MSU, 1980. CRIES Report. Special report #1 (draft). U.S. Geological Survey. 1980. Landsat Data Users Notes. USGS, EROS Data Center, Sioux Falls, South Dakota, 57198. Vink, A.P.A., 1975. Land Use in Advancing Agriculture. Springer- Verlag, New York and Berlin. White, Michael E., 1977. Surface Area Measurements and Seasonal Variation of Selected New Mexico Lakes. Technology Applica- tions Center, University of New Mexico.