nls m mommies l Mill Hill MIIGCH l lllllll\lllll‘llllllllml 31293 l l This is to certify that the thesis entitled Nitrate Leaching Potential as Affected by the Spatial Variability of Bt Horizon Morphology presented by Bruce Karl Johnson has been accepted towards fulfillment of the requirements for Master of Science degree in Crop and Soil Sciences i231,“ / Cm. / Major professor Date August 17, 1990 0-7639 MS U is an Affirmative Action/Equal Opportunity Institution LIBRARY ‘ Mlchlgan State L_ University PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. DATE DUE DATE DUE DATE DUE - 5:. 377-: JUN ‘43: 4‘ ( ¥ MSU Is An Affirmative Action/Equal Opportunity Inditmion {RUNWANHI NITRATE LEACHING POTENTIAL AS AFFECTED BY THE SPATIAL VARIABILITY OF Bt HORIZON MORPHOLOGY BY Bruce Karl Johnson A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Crop and Soil Sciences 1990 ABSTRACT NITRATE LEACHING POTENTIAL AS AFFECTED BY THE SPATIAL VARIABILITY OF Bt HORIZON MORPHOLOGY BY Bruce Karl Johnson The main objective of this study was to relate estimated nitrate leaching to field-scale spatial variability of Bt-horizon morphology. The six-hectare site contained coarse-loamy and fine-loamy Typic Hapludalfs. The site was cropped to corn and alfalfa. and irrigated with water and dairy lagoon waste. 220 soil profiles were grid-sampled and described. The Bt-morphology data were analyzed using geostatistical procedures. Semivariograms for Btl clay content. Btl thickness. and 28t2 thickness displayed strong spatial dependence over ranges of 10-30 meters. Control-section clay content varied from 7-28 percent across the site. Soil-water nitrate concentrations at a one-meter depth were sampled weekly using suction lysimeters. The suction- lysimeter data could not be directly correlated with Bt- herizon morphology. However. the CERES Maize computer model estimated nitrate fluxes under corn for the range of control-section clay contents. The model predicted consistently lower nitrate leaching with increasing clay content. and accurately predicted corn-grain yield and soil- water nitrate concentrations. ACKNOWLEDGEMENTS I thank the members of my guidance committee for their time, efforts, and many helpful suggestions. I am especially grateful to my major professor, Dr. Jim Crum. for the advice and friendship he offered throughout my knowledge. James Hart. Joe Master's program. I thank Drs. Boyd Ellis, Ritchie, and Phil Robertson for sharing their particular areas of research expertise and for their many useful editorial suggestions. It was a pleasure to work with such dedicated teachers. I thank the Crop and Soil Science Department and Dr. Jim Crum for providing financial assistance during my graduate study at MSU. Special thanks to my fellow graduate student Marty Rosek, who endured hours of pedon samplings and descriptions, and who shared his impressive knowledge of ‘ soil description and mapping. Thanks also to Brad Joern for many discussions about soil-science research and nitrate research in particular. To my dear friend Shawel Haile-Mariam. many thanks for the countless coffee breaks. gentle Ethiopian counseling, and irrepressible sense of humor. As always, thanks to Mom and Dad. Without them the thesis, and its author. would not be possible. Finally, thanks to my best friend and newlywed wife Sue, for her love, support. and humor throughout. iii TABLE OF CONTENTS List of Tables........ ....... . ......... . ...... . ........... 1v List of Figures.... ................... . ...... . .......... ..vi INTRODUCTION............ ...... . ........ ... ............ .....1 Hypothesis.... ...... ... ........ ....... ................ 3 Objectives.... ...... ..... ................ .............3 LITERATURE REVIEW.... ................................... ...4 I. THE NITRATE "PROBLEM".......... ....... ...............4 MANAGEMENT FACTORS AFFECTING FERTILIZER NITRATE......8 II. MOVEMENT TO GROUNDWATER Nitrogen Fertilizer Rates......... ....... ............9 Fertilizer Timing.................. .......... . ..... .10 Water Management................. ................. ..11 Tillage and Cropping Effects........................12 III. SOIL MORPHOLOGY FACTORS RELATED TO NITRATE.. ........ 14 CONTAMINATION OF GROUNDWATER Effects of Soil Morphology.............. .......... ..14 Variability of Nitrate Leaching Processes.. ..... ....16 IV. SOIL SPATIAL VARIABILITY..... ..... . ...... .. ....... ..18 Classical Statistics................................20 Systematic Versus "Random" Variation................21 Nested Soil Variation and Observation Scale.. ....... 22 v0 GEOSTATISTICSeeeeeeeeeoeeeoeeeeeeeeeeaeee ssssssss 00.2‘ The Theory of Regionalized Variables..... ..... ......25 Stationarity........................................27 The Semivariance and Semivariogram..................30 Semivariogram Models................................34 KriginOOOOOOOOOOOOOO000.0000...0.00.00.00.000000000036 Punctual Kriging Concepts...........................39 Block Kriging............. ...... ....................43 Geostatistical Applications to Leaching Studies.....45 iv METHODS AND MATERIALS Study Site Location.................................47 Soil-water Nitrate Sampling.........................51 Soil Variability Sampling and Analysis..............55 Computer Modelling of Nitrate Leaching..............57 RESULTS AND DISCUSSION......................... ...... .....61 I. SOIL VARIABILITY.......... ...... ....................61 Notes on Soil Eorizons.............. ............. ...63 Pedon Classification............. ............. ......68 Semivariance Statistics......... ........ ............71 Kriging Results........... ............. .............78 II. LYSIMBTBR RESULTS....o.......................o....o.89 Overview..................................... ..... ..89 Experimental Control..................... ......... ..90 Soil Water Balance..................... ..... ........93 Comparative Land-Use Effects................. ..... ..96 Corn................................................97 Alfalfa............................................102 Forest.............................................105 Nested Analysis and Required Sample Numbers........106 Nitrogen Analysis of Deep Soil Cores...............113 Soil Variability and Nitrate Measurements..........123 III. CERES Maize Modelling Results......... ......... ....127 Estimated Soil Hydraulic Characteristics...........127 Irrigated Nitrate Leaching.........................130 Rainfed Nitrate Leaching.................... ..... ..136 Estimates of Soil-water Nitrate Concentrations.....142 and Fluxes CONCLUSIONS........ ...... .......... ..... ...... ..... ......148 LITERATURE CITED..... ....... .............................149 APPENDICES Appendix I. Horizon Depth and Thickness Data ..... ..158 Appendix II. Data Input for Semivariogram..........169 Calculations Appendix III. Graphs of Lysimeter Nitrate..........176 Concentrations. 1987-89 Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table 10. 11. 12. 13. 14. 15. 16. LIST OF TABLES Horizon-thickness statistics .................. 62 Thickness statistics for E horizon and ........ 62 inclusions. Bt-horizon statistics..... .................... 64 Simple correlation matrix for St .............. 64 properties. Selected sub-sole characteristics.... ......... 68 (depth in meters). Semivariance couples. Btl and 28t2 ............ 73 thickness. Semivariogram statistics ...................... 73 Estimation variance versus sample ............. 87 variance. Mean Bt properties by control-section ......... 88 class. Nitrogen analysis of applied lagoon ...... .....91 waste. Soil-water balance. 1987-89 ................... 94 (Thornthwaite method). Maximum and minimum nitrate.. ................. 98 concentrations under corn and alfalfa. 1987-89. Nested analysis of lysimeter data. by ........ 108 sample date. Lysimeter samples required for given.. ....... lll precision. by number of sample dates. Soil-nitrogen values for deep soil cores.....119 Control-section clay contents for ....... ......124 study-site lysimeter clusters. vi Table 17. Table 18. Table 19. CERES soil hydraulic characteristics ........ .128 from soil morphology data. Selected model outputs for irrigated ...... ...137 simulations. Soil input for 23% control-section clay. Selected model outputs for rainfed...... ..... 137 simulations. Soil input for 23* control-section clay. vii Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure 2a. 2b 10. 11. 12. 13. 14. 15. LIST OF FIGURES Relationship between covariance and ........... 31 semivariance. Portion of hypothetical grid. illustratinG....33 semivariance computation for given lags. Hypothetical semivariogram resulting from ....33 computations in Figure 2a. Labels indicate semivariograms features. numbers refer to lags. Example of sample-point weights generated.....42 by punctual kriging. Illustration of sample-point weight.... ....... 42 calculations for block kriging. Location map of Kalamazoo County and.... ...... 49 Kellogg Biological Station (KBS). Schematic diagram of the study site ..... ......50 within the K85 center-pivot field. "Typical" Kalamazoo Loam pedon as. ........ ....52 observed at the study site. Flow diagram for CERES Maize modelling ....... .60 of Bt-horizon spatial variability. Pedon frequency distribution as a ......... ....70 percent of total pedons. Semivariogram for Btl clay content ............ 74 Semivariogram for 23t2 clay content ........... 7S Semivariogram for Btl thickness... ............ 76 Semivariogram for 23t2 thickness....... ..... ..77 Block-kriged map of control-section ........... 79 clay content. Area distribution of control-section..........80 classes. block-kriged map. viii Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. Topography of study site......................81 Inverse-distance map of control-..............83 section clay content. Area distribution of control-section..........85 classes. inverse distance map. Estimation variance for block-kriged..........86 Bt1 clay content. Soil-water balance for Kalamazoo Loam.........95 1987-89. Mean lysimeter nitrate concentrations. ...... ..99 (corn). 1987-89. Mean lysimeter nitrate concentrations ........ 103 (alfalfa). 1987-89. Nitrogen depth-profile for lysimeter.........115 cluster one. December 1988. Nitrogen depth-profile for lysimeter.........116 cluster two. December 1988. Lysimeter means for cluster one and..........117 two. 1987-89. Nitrogen depth-profile for lysimeter... ...... 118 cluster three. December 1988. Nitrogen loss under irrigated................131 conditions. 1987-88. Nitrogen loss under irrigated...... ...... ....132 conditions. 1988-89. Nitrogen loss under irrigated ............... .133 conditions. 1989-90. Nitrogen loss under rainfed conditions.......138 1987-88. Nitrogen loss under rainfed conditions.......139 1988-89. Nitrogen loss under rainfed conditions. ...... 140 1989-90. Measured versus modelled nitrate...... ..... ..143 concentration. 23% control-section clay. ix 1' ——IIIlIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII-Illlll-l----———— I Figure 34. Predicted cumulative nitrate leaching.. ...... 145 1987-88. Figure 35. Predicted cumulative nitrate leaching. ....... 146 1988-89. ‘ INTRODUCTION Nitrate contamination of groundwater is an environmental concern. and agricultural activities are a major source of nitrates. At the Kellogg Biological Station (KBS) in southwestern Michigan. a center-pivot irrigation system disposes dairy lagoon waste onto corn and alfalfa fields. Research scientists. KBS farm managers. and local residents were interested in estimating the degree of nitrate leaching from these fields. The purpose of this study was to estimate the nitrate flux under the KBS center- pivot system. to predict the effects of soil spatial variability on the nitrate flux. and to compare soil-water nitrate concentrations under corn. alfalfa. and hardwood forest. The goal was to provide information for farm- management decisions and to determine if the soil variability dictated special management practices for reducing field-wide nitrate leaching. Many factors influence the extent of agricultural nitrate leaching to groundwater. Key management factors are N-fertilizer rates and timing. water management. cropping systems and tillage, all of which are essentially controlled by the farmer. Soil morphology can also be an important determinant of the rates and degrees of nitrate leaching. Any management strategies which address nitrate leaching must consider the inherent soil properties that interact with management variables.v 2 Soil profiles with fine-textured horizons and/or textural discontinuities restrict water movement through the nt uptake or 1976). thus increasing the potential for pla ratt. et al.. 1972; Devitt, et al.. 8011 r denitrification (P Control-section texture has been significantly correlated with average soil nitrate concentrations below the root zone (Lund. et al.. 1974). Control-section clay content alone accounted for 68 percent of the subsoil nitrate variation. Profile drainage. and hence nitrate leaching. is ned by the least-permeable horizon i 973: Jones and Kiniry. 1986). For probably gover n the profile (Nielsen. et al.. 1 the Typic Hapludalfs of the KBS center-pivot field, the least-permeable horizons are the Bt horizons. A 1987 study spatial dependence of Bt-horizon at KBS demonstrated the Thus. it is (J.R. Crum. 1989. unpublished data). morphology -horizon morphology which the spatial variability of the Bt may control the degree of nitrate leaching under given management practices at KBS. By documenting the Bt-horizon ct on nitrate spatial variability and modelling its effe leaching, it can be determined whether soil variations are influential enough to warrant special management practices for areas more susceptible to nitrate leaching. Meals Field-scale spatial variability of Bt-horizon morphology significantly affects nitrate leaching in the Typic Hapludalfs at KBS. Soil profiles which contain thinner and/or coarser-textured Bt horizons may contribute disproportionately to field-wide nitrate leaching. Objectives 1) To characterize the spatial variability of Bt- horizon morphology for a portion of the KBS center- pivot field. To examine the effects of soil variability and land- 2) uses on soil-water nitrate concentrations. 3) To use actual soil-water nitrate concentrations and computer-modelled profile drainage to estimate the annual nitrate flux under corn. for a range of Bt- horizon variations. ’ 4) To compare computer-modelled nitrate leaching estimates for rainfed and irrigated simulations. and to validate model outputs where possible. LITERATURE REVIEW I. THE NITRATE "PROBLEM" Groundwater quality is a major environmental concern. st ubiquitous contaminants in groundwater is One of the mo In Michigan. nitrate (NOB-). an inorganic form of nitrogen. the incidence and severity of groundwater nitrate tamination has risen sharply in the last two decades. a con and internationally mirrored nationally trend which is rchild. 1987: Kittleson and Kruska. (D'Itri et al.. 1985: Fai 1987). Many uncertainties exist about the significance. future trends. and possible solutions mechanisms. effects. associated with this problem. Many natural and anthropogenic sources of nitrates environment. In nature. principal sources of mineralization (decomposition) of occur in the inorganic nitrogen are organic matter and inorganic nitrogen found within sources of nitrates minerals/geologic deposits. Man-made include industrial and automotive emissions. industrial discharge. urban sewage and runoff. agricultural N- fertilizers. livestock wastes. and rural septic systems (National Research Council. 1978). In rural areas. excessive roundwaters are strongly tied to nitrates in surface and 0 activities and fertilizer use (NRC. 1978: agricultural 1980: Hubbard et al.. 1984: Loehr. Spalding and Exner. account for three-fourths 1984). Production agriculture may 5 of the nitrogen in US streams (Keeney. 1982). Nitrates. due to their high solubility and negative charge. migrate via nt in most temperate soils. ates affect both surface and soil-water moveme Excess agricultural nitr ground-water systems. Small (<10 ppm) nitrate-level ave resulted in "moderate alter stream/lake ecology (NRC. increases h " increases in biotic productivity which can 1978). Nitrate is a limiting nutrient only in highly gen-to-phosphorous eutrophic lakes and streams with low nitro ratios (NRC. 1978). In most inland aquatic ecosystems. phosphorous is the limiting nutrient. and the eutrophication increases are relatively small (NRC. effects of nitrate 1978). Nitrate increases in groundwater generate concern primarily due to the degradation of drinking-water supplies. The impact of nitrates on human health is not clearly understood. Nitrates are directly toxic only in massive doses. but nitrates can be converted in human digestive (N02-) and possibly N-nitroso compounds. systems to nitrite In infants less than two years old. a potentially fatal condition known as methemoglobemia (cyanosis) can result from excessive nitrate ingestion and subsequent nitrite formation (NRC. 1978). Infants less than three months old y susceptible to the condition. due to the are particularl presence of gastric bacteria which can readily oxidize nitrates to nitrites. In 1962 the EPA established an upper limit for drinking water of 10 ppm nitrogen in the nitrate form (10 ppm No3--N). The incidence of methemoglobemia is .I 6 rare when drinking water contains less than 10ppm N03--N. but it measurably increases at higher levels (NRC. 1978). "is rarely fatal. is readily diagnosed. and Methemoglobemia (NRC. p.6). is rapidly reversible with clinical treatment" Milk and bottled water are safe alternatives for infant & diets if tap-water quality is suspect? No case of methemoglobemia has been documented in an adult human (NRC. 1978). A more serious health threat is represented by N- which are potential by-products of nitroso compounds. nitrate/nitrite ingestion. Nitrosamine derivatives are of particular concern. In laboratory animals. nitrosamines are nic for all vital tissue types and c Nitrosamine risk carcinoge an induce tumors from a single dose given at infancy. urately estimated; risk Ende (NRC I factors for humans cannot be acc estimates vary by one or two orders of magni 1978). Nitrosamines potentially metabolized from drinking water are minute compared to such sources as cured meats and 1978: Food Safety and Quality Service, cigarettes (NRC. health risks posed by typical nitrosamine l. but enough ambiguity exists for 1978). The real exposure are believed smal limiting exposure to these compounds and their precursors (NRC. 1978). Limiting the ingestion of inorganic nitrogen depends on knowledge of the relative contribution from various sources. An "average" American ingests 808 of dietary nitrates from per day of water at the EPA limit vegetables. and two liters A I 7 17 percent to total daily intake (data would contribute only from White. 1975: re-calculated by author for 10ppm NO3--N water). Because ingested nitrates are significant due to also consider their conversion to nitrite (NO2-). we must direct nitrite ingestion. The largest source of ingested but the slow rate of formation and nitrite is saliva. ingestion probably produces minimal effect (NRC. 1978). Otherwise. cured meats represent the largest single-dose Cured meats are also a direct source of and Quality Service. nitrite source. nitrosamine compounds (Food Safety 1978). Health risks associated with nitrates. nitrites. and nitrosamines are controversial (NRC. 1978). but drinking major source of these compounds. Because water is not a typical nitrate ingestion from water is relatively minor. infant health and public policy enf EPA standard (NRC. 1978). orcements are the prime reasons for maintaining the Trends in groundwater data suggest that nitrate regardless of current increases may persist for some time. activities. With the rates of nitrate accumulation in groundwater largely unknown. and health effects still uncertain. a conservative approach toward nitrate-leaching management seems wise. Furthermore. nitrogen losses continue sub-optimal yields for farmers. ing of nitrogen wastes. to represent economic input losses. and inefficient recycl on nitrates may seem questionable The research emphasis Given the apparent seriousness of such problems as pesticide contamination of groundwater. But nitrates do provide an .4 l opportunity to study how a rather ubiquitous. mobile contaminant leeches through soils and enters groundwater systems. Nitrate-leaching research provides important clues for soil-to-aquifer transport processes and contaminant travel times. This knowledge is applicable to a wide range of groundwater contamination problems. II. MANAGEMENT FACTORS AFFECTING FERTILIZER NITRATE MOVEMENT TO GROUNDWATER Two conditions are essential for nitrate leaching to groundwater: 1) soil nitrates must be available for leaching. and 2) water must transport nitrates below the root zone (Smika. et al.. 1977). These two conditions highlight the central roles played by nitrogen application and water management as primary management variables for limiting nitrate losses. However. crop uptake of N is dependent upon so many interrelated factors that the entire farming system must be geared toward maximizing N-use efficiency. As Keeney (1982. p.626) states. "The greatest need is to predict accurately the N dose-response relationship for a given crop on a given farm.” There is no \/\;uk advantage in applying excess N if other factors are limitingB t~ production. Nitrogen Fertiliser Rates The environmental effects of nitrogen fertilizer rates are not due to the rates per se. but to the degree to which the crop can use the applied N (NRC. 1978). This fact has frustrated attempts to recommend simple. environmentally sound. fertilizer-rate guidelines. The difference between applied N and crop uptake represents a potentially leachable fraction. depending largely upon the degree of denitrification (Stanford. 1973). Given the variability in quantifying directly-measured N-leaching. the nitrogen mass balance remains a good overall indicator for estimating leaching potential at field scales (Pratt et al.. 1972). , While total N uptake may increase with N-application rates. the percent recovery of N decreases (Gerwing. et al.. 1979: Olson & Kurtz. 1982: Motavalli et al.. 1989). Nightingale (1971) found a positive correlation between N- fertilizer rates. soil NO3--N concentrations below the root zone. and groundwater concentrations of NO3--N. The study n—-'_“_ included several crops and soil types. and emphasized that N-use efficiency was more deterministic of leaching than N-fertilizer rates themselves. Residual N03--N after harvest is a major factor in leaching. and excess N may not appear in water wells for many years (Pratt et al.. 1972). Smika et a1. (1977) found a high negative correlation (r--0.99) between nitrate leaching at 1.5 meters and total dry matter DIOdllCtiOh for corn. SUQQGSCIDO I strong inverse A I 10 relationship between N-leaching and N-use efficiency. Total residual-N in soil profiles is related to fertilization rates (Olsen et al.. 1970: Chichester & Smith. 1978). and long-term overapplication can result in greater losses to subsurface water (Chichester & Smith. 1978). Fertilizerlfiming If excess soil nitrate is limited throughout the growing season. the probability of a precipitation or irrigation event transporting nitrate is reduced. The theory behind timed N-applications is that fertilizer applications should be synchronized with crop demand. thereby placing nitrogen in the environment only when there is good probability of crop uptake (Stanford. 1973: Olson a Kurtz. 1982). In Minnesota. single N-applications increased aquifer nitrate levels by 7 to 10 ppm under irrigated corn. but no nitrate increase resulted from the same rate split over four applications (Gerwing et al.. 1979). Timmons and Dylla (1981) reported 12% greater NO3--N losses for one-time broadcast versus split fertilizer applications. but only under higher (Sen/application) irrigation rates. The Timmons study lacks treatment-specific yield data. however. leaving crop uptake differences as an uncontrolled variable. On a sandy Wisconsin soil. Saffigna et al. (1977) reduced seasonal nitrate leaching under potatoes from 200 to 120 kg 11 N/ha by decreasing N-fertilizer rates and increasing the frequency of split applications. Potato yields were the same for both the conventional and improved treatments. Crop simulation models provide an interesting theoretical glimpse at benefits derived from timed applications. Alocilja and Ritchie (1989) used the CERES Maize model to schedule nitrogen applications. optimizing economic returns versus nitrate leaching. If the model is validated for a given geographic area and proper irrigation equipment is available. this approach may represent a state- of-the-art. dynamic management strategy for optimizing nitrogen use and environmental quality. The practice of splitting applications is a farm management decision. Split applications for most farmers are limited to an initial broadcast and single sidedress. due to traffic considerations and operational costs (Keeney. 1982). Farmers cannot synchronize these early applications with the peak nitrogen demand by corn and other non-leguminous crops. For operations with irrigation systems. multiple split applications can maximize N-use efficiency and reduce nitrate losses (Keeney. 1982). Water Management Nitrogen application rates and methods are intimately related to water management in minimizing nitrate losses to groundwater. Water movement plays a critical role in nitrate 1“ ml.- 12 leaching because: 1) nitrate is extremely water-soluble. and nitrates are not adsorbed by soils with net CEC: 2) insufficient moisture may result in poor N-uptake. leaving large amounts of leachable residual N. 3) intense precipitation events may drive nitrates below the root zone. and 4) hydraulic discontinuities or saturation may accelerate denitrification. Owens (1960) directly related leaching losses to water movement in soils. Pratt et a1. (1972) found excellent correlation between observed soil NO3--N amounts to 30 meters and flux estimates calculated by multiplying excess N/year estimates with water transit times. Limiting percolation below the root zone by varying sprinkler irrigation reduced nitrate losses in two studies (Saffigna. et al.. 1977: Smika et al.. 1977). Hergert (1986) demonstrated that even incrementally applied N can be leached if sandy soils are over-irrigated. In an extreme demonstration. Endelman et a1. (1974) noted nitrate movement of 15-20 cm per day on a loamy sand under 2.5 centimeters of applied water per day. This finding underscores the high potential for nitrate leaching through coarse-textured soils. Several factors influence the effects of tillage practices on nitrate leaching. Tyler and Thomas (1977) found 13 higher leaching rates for N03--N and a chloride tracer under no-till versus conventional tillage. This was consistent with the generally-accepted view that. on average. no-till produces greater water infiltration. Kanwar et a1. (1985) demonstrated an opposite effect. with far less nitrate leaching to 1.5 meters under no-till. Gilliam and Hoyt (1987) attributed the discrepancy to differences in nitrogen distribution within the soil matrix. Macropore flow accounts for proportionately greater solute transport when N is relatively unincorporated: displacement flow transports the N in the soil matrix (Tyler and Thomas. 1981). Most current theories regarding nitrogen dynamics and water infiltration/movement suggest a probable increase in N03- leaching under no-till (Gilliam and Hoyt. 1987). Cropping systems can affect seasonal nitrate leaching. but long-term effects are unclear. Crops which require high N inputs obviously engender some increased risk of N-loss. Olsen et al. (1970) related higher nitrate leaching levels to the frequency of corn in a corn-fallow rotation. Alfalfa can reduce soil profile N03--N in rotations with non- leguminous crops (Stewart et al.. 1967: Schertz 5 Miller. 1972). and the residual NO3--N can be removed down to several meters (Mathers. et al.. 1975). The literature regarding magnitudes and rates of nitrogen release after legume plow-down is scarce and conflicting. Legume residues can contribute to higher leached-N levels than residues from N-fertilized non-legumes (Adams and Pattinson. 1985: 14 Groffman et al.. 1987). Further research is needed regarding long-term nitrogen balances and redistribution of soil-N under crop rotations. III. SOIL MORPHOLOGY FACTORS RELATED TO NITRATE CONTAMINATION OF GROUNDWATER Eff. tests. -. of ....Srailflnornhelosx Soil morphology can strongly affect rates and degrees of nitrate leaching. Such factors as soil texture. structure. horizonization. and microtopography determine soil hydraulic behavior. which subsequently affects leaching rates and denitrification potentials (Nielsen. et al.. 1973: Van De Pol et al.. 1977: Cameron et al.. 1979: Wagenet. 1984). Soil morphology and hydraulic characteristics are spatially variable and are. therefore. difficult to relate statistically to nitrate leaching (Nielsen et al.. 1982) Soil texture is a major determinant of nitrate leaching potential. Lund et al. (1974) related control-section texture to deep (1.8-8m) nitrate concentrations on Alfisol and Entisol soils: the regression explained 86% of the nitrate variability. Well-drained. coarse-textured profiles typically exhibit low denitrification potentials and high hydraulic conductivities (Devitt et al.. 1976: Saffigna et al.. 1977: NRC. 1978). These conditions favor nitrate persistence and transport below the root zone. Coarse- 15 textured soils have low water-holding capacities and require frequent irrigation. thus increasing the potential for nitrate leaching (Smika et al.. 1977). Many studies document the nitrate-leaching problems associated with coarse- textured soils (e.g. Devitt et al.. 1976; Saffigna and Keeney. 1977: Hughes. 1983: Hergert. 1986). Morphological properties which tend to restrict water movement reduce the probability of nitrate leaching. The combined effects of soil texture. structure. horizonization. and pore continuity strongly affect leaching processes (Bouma. 1983). Finer-textured layers may decrease percolation rates. increase probability of plant uptake. and promote denitrification (Pratt et al.. 1972: Nielsen et al.. 1973). However. well-structured fine layers can be rapidly permeable. Textural discontinuities between layers can suspend water and create saturated zones favorable for denitrification (Lund et al.. 1974). The hydraulic characteristics of the least-permeable soil layer probably govern profile drainage and hence nitrate leaching (Nielsen. et al.. 1973; Jones and Kiniry. 1986). - Solute transport is often modeled using displacement theory. which apparently fails to describe real flow in structured soils (Tyler and Thomas. 1981). McMahon and Thomas (1974) demonstrated faster solute movement in undisturbed soil cores versus disturbed soil cores. Many studies (Wild and Babiker. 1972: Quisenberry and Phillips. 1976: Tyler and Thomas. 1981; Richter and Jury. 1986: Priebe 16 and Blackmer. 1989) implicate preferential flow via macropore channels as a major avenue for solute movement. Visual dye tracings in two studies confirmed water movement in continuous soil channels (Wild and Babiker. 1972: Tyler and Thomas. 1981). Macropore flow may occur more often where field microtopography produces pended conditions (Cameron. et al.. 1979). The extent of macropore flow depends upon soil moisture conditions and precipitation intensity. Many macropore-flow studies employ water applications at or near soil saturation. and saturation is not representative of normal field conditions (Cameron et al.. 1979). Quisenberry and Phillips (1976) found that applied water is less likely to flow through channels when the initial soil moisture is well below field capacity. At water inputs greater than one pore volume. structured soils can actually leach less solute than unstructured soils. due to non-mixing of percolating water with the soil matrix (Tyler and Thomas. 1981: Kanwar et al.. 1985). Soil-structure effects depend upon such factors as the distribution of solute in the soil matrix. water infiltration rate. and initial soil-water content (Quisenberry and Phillips. 1976: Gilliam and Hoyt. 1987). Several reviews document the spatial variability of soil morphology (Beckett and Webster. 1971: Webster. 1977: 17 Wilding. 1984). The morphological factors which determine soil hydraulic properties often interact independently over variable scales (Trangmar. 1984). Therefore. it is not ies typically display surprising that soil hydraulic propert r variation than soil morphology properties. The much greate spatial variability of hydraulic characteristics is further complicated by the high degree of temporal variability in soil-water content and distribution (Wagenet. 1984). Accurate estimation of field-scale nitrate leaching ires data for both nitrate concentrations and profile I These data sets often requ drainage at specific points in time. exhibit large spatial and temporal variability. and thus ntensive sampling schemes for reliable suction or block lysimeters require i characterization. Typically. collect samples for nitrate concentration measurements. Several methods. including lysimeters and water-balance calculations. estimate soil-water drainage volume. Bigger and Nielsen (1976) indicate that estimating solute flux as a product of average solute concentrations and average water I drainage is theoretically unsound. A logical approach is to calculate a solute flux for each sampling date. using a mean field-wide solute concentration and estimates of water drainage below the root zone (B.G. Ellis. 1989. personal Seasonal flux totals can then be calculated communication). from the incremental flux calculations. In summary. intrinsic soil properties are major determinants of nitrate leaching potential. These properties .....-___4 | 18 also affect leaching by influencing management requirements and practices (e.g. irrigation of sandy soils). Soil morphological properties affect nitrate leaching (Lund et al.. 1974: Devitt et al.. 1976) and also display spatial variability (Wilding and Drees. 1983). Soil surveys account for some morphology variation. but morphological variation within map units and fields can create differential leaching (Richter and Jury. 1986). Estimation of soil spatial variability is necessary for identifying extreme soil conditions which may contribute disproportionately to excessive nitrate leaching (Wagenet. 1984). IV. SOIL SPATIAL VARIABILITY Soil variability is a traditional problem in the agricultural sciences. As early as 1915. Harris remarked that soil variability could "profoundly" affect agronomic experiment results (Campbell. 1979). Classical "aggie statistics" were devoted to estimating means from crop-yield uch experimental- trials. which required the control of s In the 1920's error sources as soil changes (Gutjahr. 1984). R.A. Fisher developed randomization and blocking techniques to minimize the effects of soil variability on agronomic experiments. but without estimating its magnitude or structure (McBratney. 1984). Soils vary systematically across landscapes. which is an essential paradigm for pedologists (Wilding and Drees. " mad I 19 1983). Soil variance is partitioned geographically by mapping soils into relatively homogeneous units. and ic classes taxonomically by separating soils into diagnost (Webster. 1985). The basic objective is to enhance predictive capabilities of soil-property occurrence by minimizing within-unit variance and maximizing variance The intended soil use often dictates the between units. which is tied to activities scale and nature of observation. such as soil characterization. land-use planning. agronomic experiment design. and fertilization recommendations. not map significant soil variation Typical soil surveys do for many intensive uses. though there is little value in than the minimum management capability mapping soil at less (Beckett and Webster. 1971). s not until the 1970's that widesprea n of statistics to the degrees and nd Drees. 1983). The It wa d interest emerged in the applicatio patterns of soil variability (Wilding a emergence of geostatistical methods. or statistical analyses which consider the spatial orientation of observations. was perhaps the most significant development (Webster. 1985). This accompanied more intensive land-uses and an increasing t. Systematic level of sophistication in soil managemen pedogenic processes and landscape position determine soil occurrence. therefore random statistics are not easily applied to spatial variation. The systematic spatial variation of soil properties. if geostatistically 20 provide greater predictive capabilities than quantified. can conventional statistics (McBratney and Webster. 1983). ClassicaLssstiatipm “..- Statistical procedures first require definition of a sample population. In soil science. the samples are usually drawn from a geographic volume (e.g. horizon. pedon. field) or from defined taxonomic units (e.g. series. map unit. sub- group) (Webster. 1985). A basic statistical characterization istribution. requires knowledge of the sample probability d mean. and standard deviation (Warrick and Nielsen. 1980). Most soil properties are either normally or log-normally distributed (Wilding and Drees. 1983). A normal distribution is required for conventional statistics. so log-normal or complex distributions require transformation prior to statistical analysis. The relationship between the sample mean and standard deviation is an indicator of sample variability. A commonly used statistic is the coefficient of variation (CV). which is the standard deviation as a percent of the sample mean (Wilding and Drees. 1978: Warrick and Nielsen. 1980). The CV is unitless and therefore allows comparisons of variability It is a valid statistic only when there is among data sets. a normal distribution. non-zero mean. and no covariance between the mean and standard deviation (Wilding and Drees. 1978). The mean and standard deviation can also indicate the 21 number of samples required to estimate a soil property to a given level of precision: that is. place confidence limits ndomly-drawn observation for a specified probability. on a re the number of samples required As soil variation increases. can increase drastically (Wilding and Drees. 1983). Soil properties exhibit some general trends in their degree of variation. Typically. soil morphological and physical properties are less variable than management- affected properties (Wilding and Drees. 1983). Cultivated fields tend to display greater nutrient spatial variability ltivated fields (Beckett and Webster. 1971). Soil than uncu re among the most—variable properties. hydraulic properties a which have great implications for soil pedogenic processes and soil behavior (Bouma. 1983). Total sample variance but contributions from increases with size of sample area. various observation scales "follow no consistent pattern" (Wilding and Drees. 1983). Beckett and Webster (1971) indicated that any square meter of soil can account for up to half of the total within-field variance for many soil properties. Systemic Versus "Randomkaariation Soil variability results from interactions of soil- forming factors. processes and soil management: long-range phenomena produce long-range variations. and short-range uce changes over small distances (Beckett and phenomena prod 22 Webster. 1971: Trangmar. 1984). Soil-forming factors and processes are themselves spatially-dependent. which results in spatially-dependent soil property variation (Burrough. 1983). However. these pedogenic and management factors act stochastically (probabalistically) over many inter scales. which produces both systematic variation and apparent randomness in soil-property observations. Systematic variation (i.e. "spatial correlation") has been observed for soil properties at virtually all scales of observation (Burrough. 1983). A given soil property. displays structure measured at different sample intervals. s different ranges of spatial et al.. 1984). Soil in the variance but show dependence for different scales (Uehara properties can be considered as fractal quantities in which variation patterns are a function of observation scale. In fact. systematic versus random variation is entirely scale- dependent and increasingly-finer scales reveal structure to apparently random variations (Burrough. 1983). The quantification of soil variability is also dependent upon the soil property and sampling methodology. a crucial fact 1984: Wagenet. often ignored in such studies (Trangmar. 1984). Scale dependence reflects the "nested” nature of soil variation: that is. small-distance variations occur within 23 the context of larger variations. Variances of soil properties do not increase constantly ever increasing distances. but step-wise across new scales of variation. These abrupt changes in variance reflect the predominance of a new controlling factor or process (Webster. 1977). When nested variation occurs. it is advantageous to identify abrupt changes in the mean for a property (Burrough (1983b). Nested sampling and analysis partitions variance between hierarchical sub-divisions of a population. which can be divided geographically or taxonomically (Youden and Mehlich. 1937). The variance contribution and scale differ with the sample population and property at a given observation scale. The total variance. of course. increases with increased sample area (Webster. 1985). Nested analysis assumes that the variation has independent components of variation at each level (Webster. 1985). The complex interaction of soil-forming phenomena and the nested character of genetic factors make this a tenuous proposition. Geostatistical techniques do not require these assumptions. but only characterize the continuous nature of spatial variation. In doing so. they can often identify nested variation over several scales (Trangmar. et al.. 1985). 24 V. GEOSTATISTICS "Geostatistics" refers collectively to the procedures for sampling and estimating spatially-dependent variables (Trangmar. 1984). The techniques originated primarily within the South African gold-mining industry. where statistician D.G. Krige sought an empirical method for predicting gold ore placement. Georges Matheron generalized these empirical techniques into a rigorous mathematical theory during the 1950's and '60's (see Matheron. 1971). The foundation of Matheron's spatial statistics is the theory of regionalized variables. It not only accommodates the statistical analysis of spatially-related data. but provides theories for sampling variability and sample size. including a complete theory of estimation error. A major application is the optimal. unbiased interpolation of spatial data points. with an associated variance estimate (i.e. confidence) for each point (Trangmar. et al.. 1985). Significantly. it allows an evaluation of sampling-scheme variance before sampling. provided a basic idea of the spatial variability is known (McBratney and Webster. 1983). Several review papers summarize the development of geostatistics and its application to soil science. Among them are Burgess and Webster (1980a.b.&c). Trangmar et al.. (1985). and Webster (1985). Simplified derivations of the underlying mathematical theories are given by Olea (1975). along with applications for exploration geology. The most 25 complete and rigorous discussions are presented by Matheron (1971) and Journel and Huijbregts (1978). but they are difficult for most readers. This review begins with a discussion of regionalized variables and stationarity. concepts that provide theoretical justification for semivariance and kriging calculations. The semivariance is the major statistic for indicating spatial dependence. and kriging (after D.G. Krige) is the subsequent interpolation procedure. The. Theory of Regionalized Variables Consider a data set of soil pH values collected from a farmer's field. Each value is a random variable which is part of an infinite set of sample pH values for that field. When a particular random variable is associated with the coordinate where it was sampled. the variable becomes a regionalized variable. That is. both the pH value and its position in space are relevant to the statistical analysis. If the infinite set of pH values were associated with their respective infinite sample points. it would generate a probability density function. This function is called the random function. The concepts of random variables. regionalized variables. and random functions constitute the core of regionalized-variable theory. The concept of a "random function" may seem paradoxical in describing spatially-dependent phenomena. In fact. a 26 random function may describe a highly-structured. spatially dependent set of data. or the converse. The random function does not imply necessary randomness. but simply indicates that any element within the probability distribution can theoretically associate with any given geographic point. This satisfies the requirements of statistical randomness and allows the application of some conventional statistical concepts to geostatistics (Olea. 1975). Gutjahr (1984) describes the random function as a "spatial stochastic process". a phrase that well-describes many soil property occurrences. Regionalized data must exhibit a normal distribution or be transformable to a normal distribution (Trangmar. 1984). The regionalized variables possess several characteristics not usually shared by conventional data. The geometric support describes the sample size. shape. and orientation (Olea. 1975: Webster. 1985). It can be a critical consideration. as many measured soil properties (e.g. hydraulic conductivity) are highly dependent upon the sample's characteristics (Wagenet. 1984). The larger volume from which the samples are drawn is termed the geometric field. Spatial data may exhibit anisotropy. or differential variance according to sampling direction. Regionalized variables generally display continuity at most scales of observation (Olea. 1975). The exact nature and determinants of the probability density (i.e. "random") function are usually unknown. as in 27 conventional statistics. The essential assumptions required for geostatistics involve the concept of stationarity. which is analogous to the independence of observations and errors in classical statistics (Olea. 1975). The concepts of random functions and stationarity serve as the basis for statistical inferences regarding expected values and variances within a region. Stationarityl The random function. herein designated Z(x). is defined as the set of infinite random variables (of one property) which are associated with any location "x" in a specified region. Stationarity (i.e. statistical independence) requires that the random function be identical for all sample locations. Expressed in statistical terms. the expected value of a randomly-drawn sample is the mean of the random function: E[Z(x)] = u = mean It follows that two random samples separated by a vector "h" (termed the "lag") have the same expected value u. and therefore the expected difference is zero: “..-—....“ ..- ~—...: 1 For simplicity. statistical notations used herein are consistent with Trangmar (1985). 28 ElZ(x)-Z(x+h)] ' 0 If the random function satisfies these two requirements. it exhibits first-order stationarity. It is "first order" in the sense that the mean estimate has a power of one. Variance statistics are squared terms (02. 82) and are therefore "second-order" statistics. It is important to note that "h" is a vector quantity which contains both distance and directional components. Whereas first-order stationarity implies regional stability of the distribution mean. second-order stationarity indicates constancy of the spatial covariance C(h): C(h) = Ethx)-u][Z(x+h)-ul If the spatial covariance is constant for each pair of observations separated by lag "h". regardless of pair location in the region. then there is second-order stationarity. The existence of second-order stationarity indicates that the sample variance s2 is finite and constant throughout the region. Certain natural phenomena exhibit unlimited dispersion. and cannot be described correctly using a finite variance (Olea. 1975). Thus there is no strict second-order stationarity. In such cases. a weaker assumption of variance stability is used. which requires only a finite variance 29 between observation pairs separated by lag "h". Once again. the statistic must be independent of location: VAR[2(x)-Z(x+h)l 8 E[Z(x)-Z(x+h)]2 = 21(h) This describes the variance of the difference between pairs of observations. which must be divided by two to yield a per-observation variance. This is why the resulting statistic t(h) is known as the semivariance. For soil data. stationarity via the intrinsic hypothesis is usually realized for local neighborhoods within a region. This is sufficient for spatial analysis where the variance is relatively stable within some maximum lag radius. but may break down if strong local trends are present (Trangmar. et al.. 1985: Webster. 1985). The semivariance statistic possesses several advantages over similar techniques such as autocorrelation. Autocorrelation must have second-order stationarity. a condition frequently lacking in soil data. Soil change is systematic over landscapes. and soil properties do not typically exhibit spatial covariances which are independent of location (Yost et al.. 1982). The semivariance reveals the nature of the property variation. and can also account for local trends (drift) in the data. Perhaps most importantly. the semivariance provides statistics for kriging techniques. which are used for unbiased. optimal 30 interpolation between known data points and the efficient design of sampling schemes (Burgess and Webster. 1980a). T s-._S.emiv.ar_ian_cs amusemivariogran For spatially-related data. the semivariance statistic confirms what we know intuitively: that points closer together are generally more alike than those separated by greater distances. The semivariance is a measure of the average similarity between points a given vector apart (Burgess and Webster, 1980a). The spatial relationships among data points is represented by plotting the semivariance versus the lag distance "h". and the graph is known as the semivariogram or variogram. The semivariogram is the basic tool for understanding and modelling spatial variation. If second-order stationarity applies. the semivariance can be defined by the total sample variance and the covariance for lag "h" (see Figure 1): t(h) = 52 - C(h) The intrinsic hypothesis is usually assumed instead. and the semivariance is estimated by the following equation: t(h) 8 MNh2[Z(x)-Z(x+h)]2 31 701) = em) = s2 cm) ‘l (h) C(h) C( )=0 Lag "h" Figure 1. Relationship between covariance and semivariance. 32 where Nh is the number of sample observations (not pairs) separated by lag "h". This equation derives directly from the definition of the intrinsic hypothesis for the random function. The concept is quite analogous to the sum of squares for estimation variance (32) in conventional statistics. A schematic construction of an idealized semivariogram is depicted by Figures 2a and 2b. The sample points represent a portion of a square grid. and assume a general increase in property variance with increasing distance. The semivariogram has three basic components: the sill. range. and nugget variance. The sill is a region of relatively constant semivariance. and approximates the total sample variance (82). It represents a lack of spatial dependence over the corresponding lag distances. Semivariograms which increase continuously do not define a sill or range: this indicates non-stationarity and the presence of trends. requiring some form of de-trending (Burgess and Webster. 1980c). The range is defined by the lag value at which the curve reaches the sill: it is the geographic range over which the property exhibits spatial dependence. The nugget variance (or simply "nugget") is the y-intercept value of the semivariance. Theoretically. the semivariance should be zero at zero lag. but usually it is not. The nugget represents unexplained or "random" variance which cannot be characterized by the sampling scale or methodology 33 I I I I I I I I f I I Lag 1 .§o Lag 1.4 .IIIIIIII.IIIIIII- I Lag 2.0 Figure 2a. Portion of hypothetical grid. illustrating semivariance computation for given lags. -§'!'.:=..S.2. ...... a- * F— 2.0** * * 1.4 A/ 3 £335" $1.0 l- “Nugget" [— Variance“ l- , , . . . . - 0 2 4 6 8 Lag FiCure 2b. Hypothetical semivariogram resulting from computations in Figure 2a. Labels indicate semivariogram features. numbers refer to lags. 34 (Trangmar. et al.. 1985). The difference in value between the sill and the nugget variance is the "structural variance" due to both systematic and random variation with increasing lag (Wilding and Drees. 1983). The lack of any recognizable pattern to the semivariogram indicates that there is either no spatial dependence for the measured property. or that the study methodology and sampling scheme were inappropriate for its characterization (Trangmar. et al.. 1985). The observation scale and sampling interval are key considerations for detecting a given degree of spatial variation. The required precision of spatial characterization is a function of the investigation objectives. Determining an appropriate sampling scheme is often an iterative process of using preliminary transects to gauge variability. implementing a resulting sampling scheme. and sampling additional points where variance is large or where short-interval information is required (Burgess and Webster. 1980b). All of the considerations of nested soil variation. observation scale. etc.. are essential to a well-designed study. These factors are tempered by practical constraints. such as the investigator's time and resources. Semivariogram,MOdels Semivariogram construction requires fitting a mathematical model (i.e. a linear or non-linear function) to 35 the plotted semivariance points. Soil variation is continuous at most observation scales. and semivariograms are likewise continuous functions (Webster. 1985). No general mathematical formula exists for fitting the variety of semivariogram patterns (Burgess and Webster. 1980a). Though models only approximate the true soil variation pattern. the model choice is critical. It is the model which ultimately generates statistics for kriging procedures. and model selection may be the largest source of ambiguity in kriging (Vieira et al.. 1981). Models are typically fitted to sample variograms by a least-squares approximation. Variogram points are weighted for fitting according to the number of sample pairs used in their calculation (Vieira et al.. 1981: Trangmar. et al.. 1985). The least-squares method is a reasonable compromise between model fit and the computational time required by more elaborate procedures (McBratney and Webster. 1986). Not just any model which appears to fit the semivariogram is valid (McBratney and Webster. 1986): models listed in Journel and Huijbregts (1978) are safe choices. Prediction equations include the linear. spherical. double spherical. and exponential models. The most commonly used models in soil science for stationary semivariograms are the exponential and spherical models (McBratney and Webster. 1986). There is no definitive procedure for choosing an appropriate semivariogram model (Webster. 1985). Usually 36 some combination of statistics. general semivariogram appearance. study objectives. and knowledge of soil variation is used for model selection (Vieira et al.. 1981: Webster. 1985). The general statistical criteria which favor a particular model are a high r2 value. small nugget variance (relative to sill). and a large range. The statistical considerations are balanced against subjective factors such as model-fit within the range of spatial dependence. which is critical for kriging procedures. One cautionary note involves the use of "linear-with- sill" models. The model fits some data sets well. and it is attractive due to its simplicity. However. McBratney and Webster (1986) indicate this model is theoretically sound only for one-dimensional semivariograms. It should not be used for two-dimensional semivariograms! Raising Kriging (pronounced "KREEG-ing") is simply a set of techniques for interpolating values between known data points. Perhaps due to its relatively complex mathematics. some mystery surrounds the process. This is unfortunate because the basic concepts of kriging are simple to understand. All interpolation procedures use some method for weighting the surrounding data (sample) points. but most methods are empirical. In kriging. the weights are mathematically chosen to simultaneously provide a 37 statistically unbiased estimate and a minimum local variance (Webster. 1985). Matheron (1971) performed the difficult theoretical derivations for kriging equations. and computers do the actual calculations. The only difference between kriging and other interpolation procedures is calculation of sample-point weights. but this has great implications for the quality of interpolation and its applications (Trangmar. et al.. 1985). Kriging is a weighted local averaging of the sample values in a neighborhood (Trangmar. et al.. 1985: Webster. 1985). It provides a statistically unbiased estimate of the interpolated (or "kriged") value. This means that the expected value of the estimate equals the estimate itself. and the expected difference between the kriged point and its observed value is zero: E[2'(xo)] = z(xo) and Elz'(xo)-z(xo)] = 0 where z'(xo) = kriged estimate. and 2(xo) = observed value. The result is that kriged values equal the original data values at sample locations. a condition frequently lacking with other procedures (Olea. 1975: Trangmar. et al.. 1985). Kriging also provides an estimation variance for each predicted value. Hence the reliability is known for each interpolated value and statistical confidence limits can be set. Where spatial dependence is present. kriging provides 38 an optimal. unbiased interpolated value of known variance. No other interpolation techniques can provide this. All kriging techniques assume an underlying normal distribution and stationarity of sample data via the intrinsic hypothesis (Trangmar. et al.. 1985). Kriging generally provides more reliable interpolation estimates because it considers local sample points and the variance only within that neighborhood. In contrast. classical statistics makes predictions based largely upon the total regional variance. which is usually larger than the local variance (Webster. 1985). The common endproduct of kriging. and other interpolation methods. is an isarithmic map of the observed property. An isarithm connects points of equal inferred (predicted) value for a property. The kriged map has the important advantage of known reliability. because estimation variances are provided for each interpolated value. Punctual kriging interpolates point values which theoretically have a size, shape. and orientation identical to the sample. In many situations. estimations are desired over small areas surrounding the sample points. Block kriging is a procedure for making value and error estimates averaged over a defined area. Punctual and block kriging differ only by the computation of sample-point weights. The following discussion begins with punctual kriging concepts. which are later related to block kriging techniques. 39 Punctual. .Kr i.s.ino.-..99.nsents The calculation of sample-point weights is the principal computation for kriging procedures. The semivariogram model is the basis for determining the weights for each sample point. using the lag vector between the sample and kriged point. The semivariogram range defines the neighborhood radius within which sample values are considered: kriging is not useful beyond the range of spatial dependence (Webster. 1985). Essentially. a distinct weight is calculated for each sample point based upon the semivariance between itself and the interpolated point. The sample-point weights must satisfy simultaneously the conditions that their sum equals one and the estimation variance be minimized (Burgess and Webster. 1980a). The actual weight calculations involve matrices. partial derivatives and Lagrangian multipliers and are beyond this review. Readers are referred to Journel and Huijbregts (1978) and Webster (1985) for further explanation. Once a unique weighting term is derived for a sample point. it is used to calculate both the interpolated value and its associated estimation variance. The basic equation for interpolating points via kriging is simply the sum of each sample value multiplied by its corresponding weight. summed for the ”n" sample points used: z'(xo) - zLi 2(xi) 40 where z'(xo)= kriged estimate z(xi) 8 sample value Li - weight applied to sample value z(xi) n = number of sample points used for interpolation Trangmar. et al.. (1985) state. "The estimation variance is minimized by finding the unique combination of weights which minimize the sum of semivariances between the interpolated point and sample locations." The corresponding equation is: 02 = £Lit(xi.xo) + u where t(xi.xo) = semivariance between sample and kriged point (from semivariogram model) u a Lagrangian multiplier associated with function minimization Li- weight applied to semivariance t(xi.xo) n 8 number of sample points used In both equations. the sample value and its associated variance use the same weight for a given sample location (Burgess and Webster. 1980a). Figure 3 shows actual weights calculated for a kriged point "P". The closest sample points are weighted heavily. which conforms to our intuitive notions for interpolation (Webster. 1985). Due to the large weights placed on the 41 nearest points. semivariograms should ideally be well- estimated at the shortest lags within the neighborhood radius. In Figure 3 the kriged estimate for point "P" would be calculated by multiplying each sample value by its associated weight. and summing up the results for all sample locations. The estimation variance is similarly calculated. except the weight at each location is multiplied by the semivariance between the sample point and point "P". and the Lagrangian multiplier is added for minimization. In this figure. the weighting differs by both distance and direction. reflecting semivariance anisotropy. The weights calculated for individual sample points depend upon several factors. Distance between the kriged point and sample point is often most determinant of weight. This is mediated by the sample-point geometry around the estimated value. Lone points tend to receive more weight than clustered points. and nearby points can "screen" more distant points lying in the same general direction. Thus the use of regular grids is efficient. and for irregularly- spaced sampling the addition of points in sparsely-sampled areas can greatly reduce local estimation variance (Webster. 1985). Estimation variances are always higher along the borders of sample areas. due to reduced number of neighboring data points (Trangmar. et al.. 1985). 42 0.0:: 0.006 ~0DOI 0.095 0.100 0.070 0.019 0.0 )9 0.00: 0.2“ 0.107 -0.002. 0.006 . 0.006 0.079 0 Figure 3. Example of sample-point weights generated by punctual kriging. Note the effects of distance. screening. and anisotopy on sample weights (from Webster. 1985). Figure 4. Illustration of sample-point weight calculation for block kriging. In this example. point P1 is weighted via an average of 16 point-to-block semivariances. 43 Block. Kriging Soil scientists are generally interested in average soil properties within localized areas. Even with punctually-kriged data. point estimates are usually considered to represent an area. Punctual kriging results in detailed maps. but generates isarithms having local discontinuities and rough patterns. These discontinuities can obscure longer-range trends which are more important to the study objective or soil uses (Trangmar. et al.. 1985). Furthermore. isarithm discontinuities can shift if the map origin or orientation is shifted (Burgess and Webster. 1980b). Punctual kriging is therefore commonly used for locating additional sampling points which will significantly reduce large local variances. Block-kriging. with its lower estimation variances and wider applicability. is generally preferred for characterization and mapping of soil properties. Block kriging differs from punctual kriging only in the determination of the weights used for interpolation and estimation variance. The determination of weights for sample points is accomplished by calculating an average semivariance between a sample point and several points within the interpolated block (Fig. 4). The individual semivariances are calculated using the semivariogram and the lag between the sample point and the within-block point. As with punctual kriging. the combination of sample-point 44 weights are chosen so the estimation variance is minimized (Trangmar. et al.. 1985). In block kriging. the variance estimate is partitioned into a within-block and between-block variance (Trangmar. et al.. 1985). The additional variance term in block kriging is the within-block variance of classical statistics (Webster. 1985). Large nugget variances contribute excessively to the total estimation variance. and if partitioned out by blocking can greatly increase interpolation precision. The general concept for block kriging variance is: Estimation variance = Total local variance - within-block variance Since the estimation variance alone affects the interpolation precision of the block. interpolation variances are always smaller when a given data set is block kriged rather than punctually kriged (Burgess and Webster. 1980b). For example. the kriging computer program BLOCK (Robertson. 1987) generates 16 points within each kriged block to calculate an average point-to-block semivariance. This average semivariance is then used to calculate the sample-point weight required for interpolation. In Figure 4. the sample-weight calculation for point "P1" is illustrated. Similar calculations would be performed for point "P2" and other neighbors. such that the sample weights sum to one. The within-block variance is also calculated from the 16 45 within-block points. again using the semivariogram model. The estimation variance for the block is the weight-averaged difference between the sum of all point-to-block variances and the within-block variance. In practice. kriging is often performed using an isotropic variogram model. and hence the weights depend only upon their distance from the kriged point. If significant anisotropy exists. then an anisotropic semivariogram model should be used for determination of weights. This results in more accurate interpolation (Trangmar. et al.. 1985). The number of sample points required to reasonably krige an estimate depends upon point geometry and the degree of anisotropy. Reported values range from seven (Vauclin et al.. 1983) to 25 (Webster and Burgess. 1980a). Geostatistical techniques can be applied to any soil property which may affect leaching. This application. however. is subject to the many sampling and methodological considerations referred to in Section IV (see also Wagenet. 1984). Recognition of spatial dependence relies largely upon the design of appropriate sampling schemes and experimental methodologies. Otherwise. spatial relationships may be obscured by experimental errors. Soil-water properties and solute transport are particularly variable and difficult to characterize (Nielsen 46 et al.. 1973: Bouma. 1983: Wagenet. 1984). Many studies indicate large spatial variability in these properties. with short-ranges of spatial dependence. Kriging techniques have been used to estimate the efficiency of sampling schemes for water infiltration (Vieira et al.. 1981). and to characterize variability of soil-water tension (Yeh et al.. 1986). Co-kriging. which utilizes spatial correlation between a sampled property and a less-sampled covariate. has been used to estimate available water content based upon soil texture (Vauclin et al.. 1983). One study (Flaig et al.. 1986) used semivariograms and kriging to directly estimate nitrate leaching under high- intensity irrigation. Nitrate movement was monitored using suction lysimeters arranged in two 35-meter transects. and one 10m by 10m grid. Sample spacing ranged from one meter to 0.25 meters. The results indicated spatial dependence of nitrate pulse velocity and total nitrate loss at lag distances of less than five meters. The nugget variance accounted for 50-70% of the total variance. Semivariogram generation and kriging estimates were constrained by low sample numbers. but some improvement was gained in areal leaching estimations. As geostatistical applications to soil studies become more sophisticated. the potential increases for predicting leaching variability. The short-range spatial dependence 0f many soil-water properties may practically limit characterization of large areas. owing to large sample 47 numbers. Co-kriging techniques may be useful where spatial variation of correlated properties is already known. For intensive land-uses (e.g. animal feedlots). a thorough spatial analysis may be feasible and reduce the impact of point-source contamination. METHODS AND MATERIALS Study.site Location The study site is located at the W.K. Kellogg Biological Station (KBS). in the northeast corner of Kalamazoo County. Michigan (NEW. NEW. section 5. T.1 S.. R.9 W: see Figure 5). The site is situated on a pitted outwash plain approximately one mile southwest of a recessional moraine. Elevation is 290 meters above sea level. The major soils at KBS are the Kalamazoo Loam (Fine-loamy. mixed. mesic. Typic Hapludalf) and Oshtemo Sandy Loam (Coarse- loamy. mixed. mesic. Typic Hapludalf). Small areas of Cohoctah (Coarse-loamy. mixed. mesic. Fluvaquentic Haplaquoll). Pella (Fine-silty. mixed. mesic. Typic Haplaquoll). Plainfield (mixed. mesic. Typic Udipsamment). and Spinks (Sandy. mixed. mesic Psammentic Hapludalf) soil series occur within the KBS boundaries (Whiteside. 1982). The work reported here was done on a 6-hectare study site. located within a 60-hectare field (Figure 6). The eastern two-thirds of the field was cropped to continuous corn from 1984 to 1989. The western third was cropped to alfalfa from 1984 to May 1989. after which it was planted to no-till corn. The field has been under a center-pivot irrigation system since 1984. which is used to apply dairy lagoon waste. fertilizer. and irrigation water. In 1989 the dairy maintained 285 dairy cattle. 150 of which were adult 48 49 / KBS Kalamazoo County Figure 5. Location map of Kalamazoo County and Kellogg Biological Station (KBS). 50 Z Baseline Road \\\\\\\\\\\\\\\\\\\\ I I I\I\I‘I\I\I\I\l‘I\I\/\I~l\l\l‘l\l I I I I I I \ ‘IWIIIIIII'II‘\IIIIIII \ . IIIIIIIIIIII \ tu it ssxsssx \ “fill/I’ll!!! \ \\\ I I I \ I‘ll! \ I I I I II II II I I I I III \\\\\ \ III'I'XIII’I'IWII[It’ll/«III! “'- ‘ifi \I\IIII (11” as I’ll! \\ x \ II II I I I’ll! I I I’ll ’II’I II I’ll I I I 1 I I I l I I l I l d n \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ ‘6 \ '5 \ I I I I I I I I I I s \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ I I I I I I I I I I I I I I I I I I I I I I I I I I I I s \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ I I I I I I I I I I I I I I I I I I I I I I I I I I l I e \ \ \ \ \ K \ \ \ \ S \ \ \ \ \ \ \ \ \ \ \ \ \ ‘\ \ I I I I I I ’ ’ ' ’ ’ ’ ’ l l ’ I I I I I I I I I I I s \ \ \ ‘ \ \ \ \ \ \ \ I I I I I I I I I I I \ II II I .\\\\ "C -P t\\\\\\\\\\\ III] III enter IVO III/I’ll!!! \\\ \\\\\\\\\\ Ill I’ll/IIIIIIIIIII’ll/I’ll! \\\\\\\\ xx 0 \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ I I i I I I I I I I I I I I I I I I I I I I I I I I I I s \ \ \ ‘5 \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ l I I I I I I I I I I I I I I I I I I I I I I I I I I I I ~ \ \ \ '6 \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ I I I I I I I I I‘ I I I I I I I I l I I I I I I I I I I I s \ \ \’ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ St a \ \ \ \ I I I I I I I I‘ I‘ I I I I I I I I I I I I I a \ \ \ \ \ \ \ \ \ \ S \ \ \ \ \ \ \ \ \ I I I I I I I I I I I I I I I I I / I I I t \ \ \ ’1‘ \l S’ S ’1‘ \ \ \ \ \ \ \ \ \ \ \ \ I I I I‘ I I I I I I I I I I I I I I I I I I I I I I I I I I I l‘ I I I I I I I I I I I I I I I I I I s \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ I’ll/IIIIIIIIIIIIIIIIIIllII/I .\\\\\\\\\\\\\\\\\\\\\\\\\\\\ I I I I I I I I I I I I I I I I I I I I I I I I a ‘I \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ I I I I I I I I I I I I I I I I I I I I I I \0 I: \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ III/IIIIII/IIIIIIIIII .\\\ \\\\\\\\\\\\\\\\\\\\\\\\ \\\\\\\\\\\\\\\\\\\\\\ III/IIIIIIIII’(III/IIIIII/ Lagoonmflflb Dairy Buildings OsD Kg]. 8 m Duck Lake VVVVVVVV Figure 6. Schematic diagram of the study site within the KBS center-pivot field. Numbers refer to lysimeter clusters. Soil map unit boundaries from Austin (1979). Symbol Ken is Kalamazoo Loam. 0-2k slope: OsD is Oshtemo Sandy loam. 12-18% slope. Map scale 51 cows. Manure is washed into tanks and separated into liquid and solid fractions. The liquid fraction is temporarily stored in lagoons, while the solids are composted. re-used for bedding. and subsequently spread on fields. Sqilflyater-Ni§rateh§smpling Suction lysimeters were installed in June 1987 to monitor soil-water nitrate levels underneath the center- pivot system. The lysimeter cups were installed immediately below the contact between the sandy-loam Bt horizon ("2Bt2") and the underlying banded sand/loamy-sand outwash material ("3C" horizon) (Figure 7). Depth of cup placement ranged from 0.8-1.6 meters. Horizon depths and thicknesses were recorded for each lysimeter installation. Seven clusters of four to six lysimeters were installed: four clusters in the corn. two in the alfalfa. and one in the adjacent hardwood forest (Figure 5). The lysimeters were arranged in a rough square shape approximately five meters to a side. and numbered as one (north). two (east). three (south) and four (west). Clusters one and two contained two additional shallow lysimeters (1.5 and 1.6: 2.2 and 2.5). The lysimeters were vacuum-pumped to a 0.7 bar tension and usually sampled on a weekly basis from June 1987 through June 1989. Samples were cooled to four degrees centigrade and analyzed at the earliest opportunity with a LACHAT flow- injection autoanalyzer (FIA) (cadmium-reduction method). 52 , Depth All colors are most colors (cm) Ap. O to 25 cm; dark grayish brown 0 (1 OYR 3lZ) loam; weak medium granular structure; friable; abrupt smooth boundary. E- 25-30 cm; brown (7.5m 5:4) 25 loam; weak medium sub- angular blocky structure; friable; gradual wavy boundary. 5 uction L sirneter Bt1 - 30-70 cm; brownldark brown y (7.5YFl 4l4) clay loam; moderate medium sub-a ngular blocky structure; firm; clear wavy . boundary. ZBtZ- 70-120 cm; brown (7.5YR Sl4) sandy loam; weak medium sub-a ngular blocky structure; friable; gradual wavy boundary. 3::- 120-150 cm. brown (1 OYR SIB) 1 20 "-"T"-'-"-' sand; single-grained; loose; dark brown (7.5YFl 4l4) loamy sand to sandy loam bands; massive; very friable; bands 1-7cm thick and occur at _ 2-15 cm spacings. 1 50 Lower boundary estimated at 200-300 cm depth. Cup 300 - 500 cm. lnterbedded calcareous sands and grave ls. Figure 7. "Typical" Kalamazoo Loam pedon as observed at the study site. Lysimeter cups were placed at the 28t2-3C horizon boundary. 53 A deep soil core was taken at the center of each lysimeter cluster in December 1988. A 7.6cm-diameter bucket auger was used to take soil samples by horizon in the upper solum. and at 15-centimeter increments below the sandy-loam 28t2 lower boundary. Maximum core depth was dictated by the presence of gravel layers at 2.5-4.5 meters. Soil samples were cooled to four degrees centigrade. extracted with 1N KCl. and analyzed on the LACHAT. An attempt was made to quantify the spatial variability of nitrogen applications via the center-pivot irrigation system. Collection bottles were placed on the same grid points used for soil profile sampling (222 bottles) on June 29. 1988. and liquid manure was applied. Due to a mechanical problem, only the southeast corner of the study site was irrigated. providing 40 samples. Insufficient points were available for geostatistical analysis. but limited data were provided on the variability of the liquid manure nitrogen contents during a field application. The effort was not repeated due to a lack of lagoon waste and time. Soil Variability Sampling and Analysis Sample points were located along a rectangular grid at 20-meter intervals. The interval was chosen based upon a soil variability study performed one mile from the current study site (J.R. Crum. 1989. unpublished data). The grid was situated to take advantage of three existing lysimeter 54 clusters (two in corn. one in alfalfa). At each cluster. two right-angle transects were sampled for short-range soil variation. using intervals of 0.5. 1. 2. 4, 8. and 16 meters. The entire grid measured 440 by 140 meters. with the long axis oriented east-west. The grid enclosed an area of 6 hectares (15 acres) and contained 222 sample points (see Figure 5). The grid dimensions were ultimately chosen because they encompassed the three lysimeter clusters. did not exceed the resources of the study. and provided ample points for geostatistical analysis. Ultimately. 219 out of the possible 222 pedons were sampled: the three unsampled points were too close to existing lysimeters. Grid points were located and flagged using a WILD Distomat and Theodolite (Total Station). Points were located to within a five-centimeter error. and elevations were recorded for topographic map production. The raw survey data were converted to coordinate data using the WILDSOFT computer program (Wild Heerbrug Instruments. Inc.. 1987). Soil sampling at each point was accomplished using a Giddings hydraulic probe mounted on a pickup truck. A 6.3cm- diameter sampling tube was used to take soil cores to a minimum depth of 1.5 meters. Horizon depths and thicknesses were measured and recorded in inches. Horizon descriptions were made and samples taken for all designated horizons greater than three inches (7.6 cm) thick. The main criterion for splitting sub-surface horizons was soil texture: other morphological criteria (e.g. color) were used for splitting 55 only if judged relevant to water movement/leaching conditions. The goal was to minimize horizon numbers for each pedon. thus aiding statistical analysis and modelling. without compromising on relevant morphological characteristics. Roughly 300 grams of soil were taken for each horizon. The soil samples were air-dried. crushed. and passed through a two-millimeter sieve. Percent coarse fragments were recorded for gross characterization. Particle-size analyses were performed on the fine-loamy and coarse-loamy Bt horizons using a modified hydrometer method (Grigal. 1973). Dispersed soil samples were first washed through a 53-micron (#270) sieve and the sands oven dried. Sands were sieved into the five USDA sand fractions and the weights for each class recorded. Silt was determined by difference. GeostatisticalwErpcedures Semivariograms were calculated for argillic horizon thickness and clay content using the SEMIVAR program (Robertson. 1987). The semivariograms were first generated using the maximum available lag distance of 462 meters. The lag distance at which a clear sill became apparent was the maximum lag distance used for subsequent variogram generation and model fitting. Step sizes were chosen by observing the frequency distribution of couples and the visual smoothness of the curve. The selected semivariograms 56 were run through SAS (SAS, 1988) and fitted with linear. spherical, exponential and gaussian models. Models were chosen based upon a combination of highest r-squared values. lowest nugget variances. and logical agreement with field observations of soil properties. The Akaike Information Criterion (AIC) was also used for model selection. The AIC ranks each model based upon model fit versus the number of model parameters (Webster and McBratney. 1989). Each property was block-kriged with the statistics generated from the chosen semivariogram model by using the BLOCK computer program (Robertson. 1987). A width of four meters was chosen for the block size. which defined 3850 blocks within the study area. A maximum search radius of 462 meters was used. and the 32 nearest data points ("neighbors") were used for kriging. Control-section2 clay content was calculated for each block using the kriged values for Btl clay content. Btl thickness. and 28t2 thickness. The semivariogram for 28t2 clay content indicated virtually no spatial dependence. so the horizon sample-mean clay percentage was used in all calculations. The generated control-section data was gridded in SURFER (Golden Software. 1987) using the minimum- curvature grid method. The gridded estimates were used to produce an isarithmic map of control-section clay content. ea..- 2 For the Typic Hapludalfs of this-study. the control section for particle-size class is the upper 50cm of the argillic horizon(s). or the entire argillic horizon(s) if less than 50cm thick. 57 using 18 control-section classes (e.g. 14% clay. 158 clay. etc.) The isarithmic map. morphological data. and series definitions were used to map the soil series within the site. The land area occupied by each control-section class was also estimated by a SURFER area-of—surface routine. Computer Modelling_of Nitrate Leaching The effects of soil variability on nitrate leaching were modelled using the CERES Maize computer program (Ritchie. et al.. 1989: version 2.10). To model these effects. argillic-horizon variations were considered as treatment variables. while management inputs were held constant. The standard soil input file. SPROFILE.MZ2. contained 22 soil types. Each "soil" in the file represented a typical pedon for each 1% class of argillic-horizon clay content. To generate the necessary silt and sand data for the kriged blocks. the silt sample data were block-kriged into 4m x 4m blocks using SURFER. which assumed a linear semivariogram. The silt data were used because the semivariogram displayed more predictable spatial variation than the sand data. The clay and silt percentages were summed for each block. and the sand percentage was determined by difference. An "average" pedon was then constructed to represent each 1% class of control-section clay content. To do this. 58 all blocks within a given 1% class were used to calculate a mean texture and thickness for the Btl and 23t2 horizons. For all 22 pedons. constant values for hp and 3C texture and thickness were used. Mean thicknesses for the Ap and 3C horizons were calculated from morphological data, and horizon texture was estimated by hand-texturing. The result was 22 distinct pedons with the same Ap and 3C horizons. but with varying argillic-horizon textures and thicknesses. The pedon data were input into the SOILW program. which estimated soil hydraulic properties for the CERES model from the morphological data (Ritchie and Crum. 1989). SOILW required data for horizon bulk densities and organic-matter content. and these were approximated using data from a nearby study site (D. Reinert and J.R. Crum. personal communication. 1989). The output from SOILW for each pedon was incorporated into the SPROFILE.M22 file. which was then in the proper format for running the CERES model. The pedons were all limited to a one-meter depth to allow leaching comparisons with the suction-lysimeter data. All management factors were input according to available KBS farm records (Harold Webster. 1989. personal communication). Since no precise irrigation records exist. an available-water threshold of 55% was used for model irrigation scheduling. Where dairy lagoon waste was used as fertilizer. the model input choice was ammonium nitrate. This decision was based on a preliminary study which indicated rapid mineralization rates for the dairy waste at 59 25 degrees centigrade (B.G. Ellis and J.R. Crum. personal communication. 1989). The corn variety used in the model was Pioneer 3780. which had similar genetic characteristics to the variety planted at the study site (Great Lakes 582). The model source code was modified to run on a planting-date to planting—date basis. This was done so that 1) the simulation would run beyond the harvest date. and 2) the modelled nitrogen leaching could be related to a complete cropping cycle. The weather file contained actual KBS weather data for the period January 1. 1987-November 30. 1989. The weather data were collected at the block-lysimeter weather station at KBS. supplemented when necessary with data from the KBS pond and/or Gull Lake weather stations. To complete the modelling of the 1989-90 year. data from December 1987-April 1988 were used. These data were chosen because they were close to the historical precipitation means for each month. The model was run to provide leaching estimates for each control-section class. under both irrigated and rainfed conditions. The nitrogen-balance output file (OUT4) was modified to provide incremental and cumulative nitrate leaching for the April-to-April year. Output files were also generated for water balance. plant phenological development. and grain yield. Figure 8 summarizes the methodology for creating the soil-profile input data which were used in the CERES Maize modelling. 60 I Original particle-size data I Kriged clay z Kriged silt x -~‘“‘*=g~”"’fl- r3 Sand by difference I l Complete particle-size data for all kriged blocks l Blocks grouped according to control -section clay content (1le class intervals) i For each 12 class, mean texture and thickness is calculated for the Btl and 28t2 horizons J. Using constant Ap and 3C characteristics, an average profile is created for each class, varying only the Btl and ZBt2 characteristics l SOILW program takes pedon morphological data and estimates soil hydraulic properties for the CERES Maize model W File SPROFILEMZZ created I or CERES Maize input, containing pedons with 7-2 62 control-section clgy Actual KBS weather and management / inputs CERES Maize V Leaching output for each 1% variation in control -section clay content, under irrigated and rainf ed conditions Figure 8. Flow diagram for CERES Maize modelling of Bt-horizon spatial variability. RESULTS AND DISCUSSION I. SOIL VARIABILITY The basic soil morphology data is presented in Appendix I. which contains horizon thickness. control-section clay. soil-series classification. and grid coordinate data. The six horizon designations were sufficient to accurately describe all pedons in the field for the purposes of the study. In general. the sequence of horizons and their gross morphology (e.g. texture. structure. color) were remarkably similar across the field. This is reflected in the series classification of the 219 pedons: 148 (68%) classified as Kalamazoo. and 65 pedons (30*) as Oshtemo. The remaining six pedons were comprised of three Spinks (18) and three Miami taxadjunct (18) pedons. The major differences between pedons were related to the presence or absence of a fine-loamy Bt horizon over the underlying coarse-loamy and/or sandy materials. Where the fine-loamy 8t horizon was thicker and finer-textured. Kalamazoo or Miami pedons were found. Where the fine-loamy St was thinner. coarser-textured. or absent. the pedons classified as Oshtemo or Spinks. All pedons were categorized as well-drained to moderately well-drained. The only indications of reducing conditions within the profiles were high-chroma mottles found in the lower St horizons of Miami profiles (2 pedons), and mottles or gleying found in fine-loamy inclusions of 12 61 62 other pedons. No significant mottling was observed above any of the fine-loamy inclusions. Statistics for horizon-thickness data are presented in Tables 1 and 2. Conventional statistics were applied to Ap. Btl, and 28t2 thicknesses and Btl and 28t2 clay contents, as these frequency distributions all approximated a normal distribution (Charles Cress, personal communication. 1989). The thickness data for E horizons and fine-loamy inclusions followed non-normal distributions, which were not amenable to simple transformations. Therefore, median, mode, and range were used to describe these distributions. Tgble 1. Horizon-Thickness Statistics Mean Std Dev CV Max Min (cng (cm) (8) ion) (cm) AD Thickness 23.0 4.1 17.8 38 15 Btl Thickness 34.1 18.3 53.7 89 O 28t2 Thickness 43.1 28.1 65.9 142 0 Depth to 3C or inclusion 105.6 29.7 28.1 216 46 Table 2. Thicknegs Statistics for §_Horizon and Inclusions Median Mode Max Min (£51, Lgm) (cm) (cm) E Horizon 13 10 51 7 Pine—loamy Inclusions 50 40 134 15 63 Statistics for the Bt-horizon characteristics are presented in Table 3. The coefficients of variation (CV) indicate that Bt-horizon thicknesses were much more variable than clay contents for the sample population. The low CV for Btl clay content is noteworthy because the Bt1 clay content strongly influences the control-section clay content. Plots of Bt1 clay contents and thicknesses versus 23t2 characteristics revealed no relationships. Likewise, there was no relationship between Btl clay content versus Btl thickness. or 28t2 clay content versus 28t2 thickness. The matrix shown in table 4 demonstrates the absence of linear correlation for these properties. Motes on Soil Horizons The A horizons varied in texture from loam to sandy loam, with field-textured clay contents estimated between 10 and 18 percent. Estimates of sand content varied between 40 and 70 percent. Color was most commonly a 10YR 3/2 (moist) and 10YR 5/2 (dry). The major variation in the A horizon was thickness, which was a function of whether the sample point was located in a tillage row or between the row. The lower horizon boundary was abrupt and smooth. An B horizon greater than or equal to seven centimeters was described in 87 of the 219 pedons (40‘). In pedons with fine-loamy Btl horizons, the B horizon was loam textured, usually less than 15 centimeters thick. and had a color of 64 Table 3. gtrhorizon statistics. Btl Thickness (cm) 2Bt2 Thickness (cm) Btl Clay (8) 2Bt2 Clay (k) Control-section Clay (8) Table 4. Simple correlgtiongmatrix;for Bt properties Btl Clay 28t2 Clay Btl Clay 1 23t2 Clay 0.07 Btl Thick 0.19 28t2 Thick 0.18 All values are "r” 0.06 0.13 0.14 (simple correlation) Btl Thick 23:2 Thick Mean Std Dev CV Max Min (cm) (on) (t). (cm) (cm) 34.1 18.3 53.7 89 0 43.1 28.1 65.9 142 0 25.6 4.5 17.6 37 11 10.0 3.0 30.0 23 5 20.4 6.0 29.7 35 5 65 7.5YR 5/4 to 10YR 5/3. In many pedons. tongues of B horizon material penetrated into the Bt horizon. Where the B horizon occurred above coarse-loamy or sandy at horizons. the texture was sandy loan to sand. thicknesses were generally greater (15-51cm). and typical soil color was 10YR 5/2 or 5/3. The B-horizon lower boundary was gradual and wavy to the underlying 8t horizon. The major difference between pedons. and their subsequent classification. was the presence. texture. and thickness of a fine-loamy textured Btl horizon. This Bt horizon displayed a characteristic moderate. medium sub- angular blocky structure and firm moist consistence. Clay skins were well-developed on ped faces. root channels. and coarse fragments. Color was quite constant across the field. usually a 7.5YR 4/4. with clay skins often a value darker. Mean coarse-fragment content was 4.58 by weight. and in many pedons the coarse fragments appeared to be concentrated near the lower horizon boundary. The lower boundary was clear and wavy to the next horizon. The Bt horizon in the underlying material was most often designated 2Bt2. where it occurred below a fine-loamy Btl. If no fine-loamy Btl was present. it was designated as a Btl horizon in the pedon description. but considered with the ”2Bt2" horizons for statistical purposes. The texture was consistently sandy loam. and soil color ranged from 7.5YR 4/4 to 7.5YR 5/6. Structure was characterized as weak medium sub-angular blocky structure. Clay bridging was 66 observed between sand grains. and occasional clay patches were noted. Mean coarse-fragment content was 10.9% by weight. The lower boundary to the 3C horizon was gradual and wavy. Where a fine-loamy inclusion was present. the lower boundary was clear and wavy. A fine-loamy "inclusion" of till-like material occurred in 27 out of the 219 pedons. most often "within" the 28t2 or 3C horizons (labelled "BC” in Appendix I). The field texture was loam or clay loam and the structure was massive. The only pedogenic development appeared to be leaching of carbonates and development of color. Occasional clay patches were observed in the upper portion of some inclusions. High- and/or low- chroma mottles were observed in nine inclusions. while three exhibited dominantly gleyed colors. An increase in apparent soil moisture was often observed at the lower boundary. sometimes approaching saturation. Based upon mottled/gleyed soil colors and soil-moisture observations, saturated conditions could exist for denitrification. However. organic carbon at that depth is likely a limiting factor for denitrifiaction. A spatial analysis revealed no observable pattern to their distribution within the study site. The lower boundary was usually clear and wavy to the 3C horizon. In eight of the 219 pedons. the inclusion extended to the deepest sampling depth. In four of these pedons. it occurred too deeply to affect the classification. Two of the inclusions occurred beneath a 3C horizon. 67 The 3C horizon was the lowermost horizon described in all but ten pedons (eight pedons ended in a fine-loamy inclusion and two in 2Bt2). The 3C was easily recognized in the field by two distinct features. lamellae and an overall change in sand-size distribution. The lamellae were approximately one to seven centimeters thick. and occurred at 2-15 centimeter spacings. The coarser-textured bands were sand to loamy sand. 10YR 5/3 color. and the finer-textured lamellae were loamy sand to sandy loam. 7.5YR 4/4 in color. The contact between bands was most often abrupt. but the apparent eluviation/illuviation of clay sometimes produced clear sub-horizon boundaries. The sand fraction of the 3C contained a noticeably higher proportion of fine and very fine sand than the "23t2" horizon. but was still within a ”medium” USDA sand textural class. This sand-fraction change was uniform across the site, and with the lamellae was considered diagnostic for the horizon designation. Two major associated observations were made regarding the 3C horizon. Whether the 3C was overlain by fine-loamy or coarse-loamy material. increases in soil moisture were occasionally observed just above the 3C upper horizon boundary. This apparent hydraulic discontinuity justified the placement of the lysimeter cups. where regular soil water samples could be drawn reliably below the root zone of corn. Secondly. although the 23t2/3C boundary was only one meter deep on average. no corn roots were ever observed in a 3C horizon on the site. Minirhizotron images from another 68 field at KBS reveal little root biomass in these banded sands (A.J.M. Smucker. J.T. Ritchie. personal communication. 1989). The complete lack of roots also justified the assumption that nitrates reaching the 3C layer are essentially unavailable for uptake by annual crops. and are likely to be leached. Although the 219 pedons were sampled to a minimum depth of 1.5 meters. not one of these soil cores extended into unaltered outwash material. In December 1988. deep cores were taken for soil nitrogen analyses at the center of each lysimeter cluster. The cores were sampled with a bucket auger and depth of sampling was limited by layers of impassable gravel. Table 5 lists the main features encountered in the sub-sole. Cluster three. in the alfalfa field. displayed a much shallower solum and depth to gravel than clusters one and two. Table 5. Selected Sub-sole Characteristicggjdepth in‘metergl Lysimeter Lower Boundary Depth to Depth to Cluster of 3C horizon Caggpnates Impagggble Grgyel -------------- meters----------—--- 1 3.0 3.0 5.0 2 2.9 3.5 5.2 3 1.9 2.0 2.9 ggdon Classification The data presented in Appendices I and II support the field observation that the study site displayed the range of 69 an Kalamazoo-Oshtemo series continuum. Figure 9 shows the frequency distribution of pedons as a function of control- section clay content. The criterion separating the two series is the 188 control-section clay content. which separates the coarse-loamy and fine-loamy particle-size classes. This 188 break is difficult to define in the field. especially where two Bt horizons define the control section. Based upon the pedon data. 25 percent of the study site was occupied by coarse-loamy soils and 75 percent by fine-loamy soils (Figure 9). Pedons which contained only sandy loam Bt horizons represented the Oshtemo endmember. and pedons with fine-loamy Bt horizons greater than 50 centimeters thick represented the Kalamazoo endmember. The pedons classified as Spinks were distinguished by their sandy profiles and the presence of the 3C lamellae as a discontinuous argillic horizon. The lamellae did meet the Psammentic Hapludalf requirements of summing to six inches (15 cm). and the Spinks series criteria of occurring at less than 36 inches (91 cm) deep. Although the illuvial formation of the finer-textured lamellae is debatable. the soil is morphologically and interpretatively a Spinks. The two pedons classified as Miami taxadjuncts were fine-loamy throughout and were moderately well-drained. The pedons do not meet all the Miami series requirements. but the taxadjunct classification is interpretatively accurate. 7O 5 II 8 ' / ’/// ’//////l ///////4 '/////l Ill/Ill l/l/l/I/l/I/l/l '////////1 ’/////////////////////. ////////////////////// '////////////////////// ///////////////////////////. ///////////I////////////////////////////// '//////////////////////////////I '///////////////////////. l/l/I/l/l/I/I/I/l/l/l/ 'l/I/l/I/I/l Kalamazoo 1.4; /// ///////‘////////////////. '////////////. //////////// ////////////a ///////. ZZZZEb '//////// /////. ///////// a) 113151719212325272931 121416182022242628 '1 6810 [x EZZZQJD Oshtemo q-m—“—‘— CD 00 JD ‘Q’ C“ (D P (suoped mo; ;o as as) Aouenbea 30323436 Control-section Clay (weight %) Pedon frequency distribution as percent of total pedons. 9. Figure 71 Semivariance Statistics The argillic-horizon thickness and clay content data (Appendix II) provided ample couples for semivariance analysis. Semivariograms were originally generated using the maximum lag of 462 meters. and step sizes (lag class intervals) between 5 and 10 meters. After viewing the resulting couple distributions and the visual smoothness of the semivariogram output. a step size of 8 meters and a maximum lag of 90 meters was chosen for all final semivariograms. Table 6 shows the couple distribution for Btl and 2Bt2 thickness using these inputs. The semivariance calculations for clay content had fewer couples (approximately 4800). due to ”missing" clay values when horizon thicknesses were zero. The rectangular shape of the sampling grid did not allow enough couples for anisotropic semivariogram analysis. Isarithmic maps of the raw data did reveal some potential anisotropy. but it occurred over distances much greater than the isotropic semivariogram ranges. Therefore. isotropic models were fitted to the semivariograms. and were considered sufficient for kriging at less than 50-meter ranges. The choice of a semivariogram model was relatively straightforward. For Btl clay content and thickness. and 2Bt2 thickness. the r2 values were comparable for 72 exponential. spherical. and gaussian models. However. the exponential model consistently provided a much lower nugget variance. which increased the precision of kriging interpolations. The exponential model also produced the lowest index for the Aikake Information Criterion (AIC). which essentially indicates it was the most parsimonious model tested (i.e. used a minimum of terms to fit a curve well). Figures 10-13 display the semivariograms for each Bt property. The semivariograms all display linear or convex- upward shapes. indicating that drift was not serious and the intrinsic hypothesis was valid (Webster. 1985). Semivariograms for Btl clay content. Btl thickness. and 2Bt2 thickness all display a high degree of spatial dependence. The semivariogram for 28t2 clay content indicates virtually no spatial dependence at this scale of observation. As indicated by the low nugget variances for the three spatially-dependent properties, methodological errors are not a major factor in these semivariogram analyses. Table 7 lists the principal semivariance statistics for the Bt properties. For the three Bt properties exhibiting spatial dependence. over 808 of the total variation was correlated with distance (i.e. ”explained” ; the nugget variance accounted for the "unexplained" variation. The sill values for the three properties closely approximated the sample variance of the conventional statistics. The reported range values in table 7 were calculated as 95 percent of the 73 sill value (Journel and Huijbregts. 1978). The range values were sufficiently large to insure effective block kriging within the 20-meter-square grid cells. stlgwéimfiemizaria c ,Quples.18t1mand-2893 thickness -.fl. -v _,. Laqmjm)_-mmflumbermpf-gguple§ 1.6 75 7.4 102 16 116 21 436 29 439 40 427 45 694 56 405 62 919 72 619 81 852 88 724 Total 5808 Table lavgemiyarigsram Ste :61 6* *‘m—— ,‘__ Nugget 8 Variance EEQperty :2 VarianceM_§ill_misxnlained11.83003.Jm) Btl Clay (8) 0.82 3.7 19.0 80.5 22.9 ZBt2 Clay (8) 0.04 - - - - Btl Thick (cm) 0.73 0.0 315.9 100.0 11.7 2Bt2 Thick (cm) 0.90 125.6 837.2 85.0 27.5 ‘ Exponential model used to fit all semivariograms. 74 8e om om ox om om ow on on .ucoucou heao Hum now Eenoofiuebfisom .oa eusowm A9055 .... mm; P b P pl p 0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000 00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000 396; 3 a. Eacoo 30 :m 75 .uceucoo asap «you uOu Eeuoofiue>fieom .HH ousowm @395 L om; 09.0.0 om on om om We on ON or o ID 991-3131131198 396; 3 as 2661:8166 mime 76 8.80m on omomow omom .nnesxoacu «on you Eeuuo«ne>maom .mu encode @295 z 03 - 0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000 000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000 as s 363 86565. :m swam/muss 77 .nnecxounu numm nou EeuOOaue>wEem .na ousofih @295 r. 03 38 E EmB mmmconk mmmm mNWeS 78 Kriging Results The block-kriging results are presented as an isarithmic map of control-section clay content. produced from estimates of Bt horizon properties (Figure 14). The kriged output resulted in a minimum control-section clay content of nine percent and a maximum of 26 percent. The map revealed an extensive. continuous area of coarse-loamy textured soil across the center of the study site. Figure 15 depicts the distribution of control-section classes as a percent of total land area. Oshtemo soils occupied 47 percent of the study area and Kalamazoo soils 53 percent. Nearly half of the area was occupied by soils with 16-20 percent control-section clay. Soils with less than 12 percent clay accounted for only three percent of the area. as did soils with greater than 24 percent clay. A plot of the study-site topography is presented in Figure 16. The southeast grid corner (440.0) was designated as "0" meters elevation. and all elevations were relative to that point. A comparison of the block-kriged isarithmic map and a topographic contour plot revealed no consistent relationship between topography and control-section clay. This result confirmed a similar finding from a nearby field at KBS (J.R. Crum. personal communication. 1989). Figure 17 displays a corresponding map of the morphological sample data. produced via inverse-distance interpolation (Golden Software. 1987). The inverseédistance (LU).— 91011ng009 A 80 -a... nnn-neeuv-p-u...~n as... e..-- u “ ~— oea us. .. n fi‘ur 28 l : i //, g i : ///////, i 1 ; l 3 /////////AAA Kalamazoo ! g ///////////////////, /// ////7////////.4///////////////7//77// /////////////'/// / ////////////'/// ////////z , 04:A//////////yvgonaaz////////(/////// anooovyxnoc277/444077/4007/ '/////An "/////////////////////////////// / €077/A097/A6077A607//' AonV/Anoovxxooovxn< //////////?//////, l l . l i ///////////, g . , 4 I g . , ' I i D 0 4 l 5 2 i l Oshtemo l r l i l 1 CD 14 g . l i . 6 co «:1 N 12 104 ._.-....-........ --. 8er [310110 % 26 30 25 27. 29 31 24 Clay (96 by weight) 1113151719 2123 Control Section -////7772//7/7777777//’EU 00 r- (O P Z l 0 on (w) 40/3 0 '9 Topography of study site. Figure 16. 82 function weights neighboring sample points according to the general equation: 221(1/d11x z. 291/9119x I interpolated value 21 - value of ith neighbor (i I from 1 to "n" points) distance from neighbor to interpolated value exponential power assigned to function The exponent "x" is often chosen arbitrarily. and usually given a value of two or three. The function is not an exact interpolator. as the sum of the weights does not necessarily equal one. The procedure is empirical and the estimation variance of interpolated values is unknown. Visually, Figure 17 displays greater local detail, but at the expense of obscuring more general trends. A comparison of Figures 14 and 17 reveals more extensive areas of extreme values in the inverse-distance map. which created a disjointed map appearance. The block-kriged data essentially averaged these extreme values via the block weighting procedure, which had the effect of smoothing the isarithms. The block-kriged map corroborated field observations that extreme clay-content values were basically point variations that are not mappable. The inverse-distance procedure gave much greater weight to these extreme values, which complicated the map appearance. Based upon field observations, practical soil management objectives, and theoretical interpolation 83 .NfiNA M 30.1m0»A .N0.WNIQH fl OJHD 'NG.RHINH H 00L .*N.nv N COOLO "XOV. .uCOuCoo ono COauOUMIHOLucoo v0 00E oocuumHUIOmLOZLH .AH OLDOHm w‘ /\ . rm 3“: x o @ y. ; 3P ; gaggurligfla 84 considerations, the block-kriged map is the best representation of study-site soil variability. Whereas the block-kriged map estimated 47 percent Oshtemo and 53 percent Kalamazoo soil by area (Figure 15). the inverse distance map estimated 33 percent Oshtemo and 67 percent Kalamazoo (Figure 18). The inverse-distance method also produced a wider range of control-section values. A key advantage of the kriging output is the interpolated values were statistically unbiased and each had an associated estimation variance. This allows statistical confidence limits to be placed on estimates of Bt properties. The semivariogram models produced low nugget variances. relatively high r2 values. and sufficient ranges which increased estimation precision. For the kriged 8t properties. the pooled block variances were approximately one-third of the conventional sample variances (Table 8). The kriging estimation variances displayed a repetitive spatial pattern which is characteristic of grid sampling. Figure 19 depicts the estimation variances for Btl clay content. Lower estimation variances occurred near the lysimeter clusters due to increased sampling density. The pattern was similar for Btl and 28t2 thickness, but with different variance values. The block-kriged output was used to determine the mean argillic-horizon properties for each 13 control-section class interval. The results are presented in Table 9. For the mean Btl textures. sand contents consistently decreased 85 C) Q :3 m g , 8 To" ’/////////. )5 //////2//////,E$ , $3 I ' 8 en - v- , (O . . F_ ? ' ,//nooyynnov / g ' i i § § ' i i g a. i * w 9 1 9 i i i g !' i 2 . l ! T I l l j l V N O Q ‘0 V N O '- v- 1- wv mo; :0 % 23 25 27 29 31 Clay (% by weight) 7 9111315171921 Control Section ion of control-section classes, inverse distance map. Area distribut Figure 18. 86 Estimation variance for block-kriged Btl clay content. Figure 19. 87 with increasing clay content. Mean Btl silt contents were somewhat variable for the Oshtemo profiles. but generally increased with increasing clay contents in Kalamazoo profiles. The 28t2 clay contents varied randomly at this observation scale. and were not considered a factor in systematic control-section clay variation. The sample means for 28t2 texture were therefore used in the table. The mean 28t2 sand and silt contents varied randomly within small ranges. The 23t2 thicknesses actually decreased slightly with increasing control-section clay. The data in Table 9 were used for modelling each control-section class (i.e. "soil type") with the CERES Maize model. Table 8._§_timatifin_2aria c ve sus Sample variance -— .-.—po. —....~.-._—-—.-—- “0* ..— ..-—- — Ergnsr§2;_l_fl-u_-“Estimatigalxatia-gs_m. “gamnlsnysriance 3:1 Clay (8) 7.0 20.4 Btl Thickness (cm) 121.1 333.7 28t2 Thickness (cm) 255.5 792.3 now 3 BB 3 ed. 8 men. 3;... can an: on BB 8 BB 3 9o. 3 o: 0.9... can 9..... B 98 B «B 3 9o. 3. 2.. new. 0.8 2...... 8 1mm mm n8 ......B 2: we 3.. m8 o8 can: 3 .....m «m 08 E S: 8 B9 on... .8 BF: ..m «.8 8 BB E 2: B QB 03 SN 6.": 8 emu :. .8 3 od. 8 3m .3... BB «may... «a 3m :. BB ma 2: 8 new 0.8 new 8m": 8 m8 3. I» ma 2: mm «.6 new new 87.. om m9 8 98 .3 cc. 3 BB 3.». new Be": 2 m9 9 0.8 2. 99 8 «.8 3w. «.8 m3": 2 m5 9. BB we 2: mm BB... BB. Nam Eu: 2 me. me 18 0... ed. on one at SN 8?: 2 mm. 8 3B 2. 9o. 8 98 e: 0.2 can: .2 mi 8 B3 3 2: mm .3 B5 ..B. we": 3 ma. 8 0.8 9... od— 8 BB 2: we. 8.": 2 ca. 9. SB 3 2: mm “.8 2: 3. Bu: 8 ...: 3 m3 mm 2: 3 QB 98 Ba. 8n: : no. a. ode to ad. ..m 30 3. w: 8n: 2 .3 8 ..8 on 2: 8 Q: Q8 3 Eu: a .3 2. a...» ..m as. 9 3K to... we Bu: B as 3 QB ed 2: m. 0.3 ..2 3.. an: 5 cm. 3.3 so so >305 Eu. 3.. E g 9.83 ammo 2 so; 28 am so is: .58 ..B B6 3 a 8 830 :08 -.....mmzwaoa «Ba Ba: - twawmamma mamas... ---, 12:50 .nneno couuoouaaouucoo ha eoauuoooun on can x .O ”HQUH II. LYSIMBTBR RESULTS Overview The lysimeters were sampled on 72 dates during the two- year study period. The nitrate concentrations are summarized graphically in Appendix III for 27 lysimeters. Four shallow lysimeters (1.5. 1.6. 2.2. and 2.5) were not included because they were damaged by freeze-thaw during the first winter. The first reported data are for July 2. 1987 (day "0”). due to the anomalously high readings recorded during May and June 1987. These high readings were assumed to be the effects of lysimeter installation and equilibration, which was supported by the observation that such continuously high readings were not recorded again during the two-year study. Four lysimeters (1.4. 2.3. 4.3. and 5.4) I apparently took slightly longer to equilibrate. The initial 80-ppm readings from these lysimeters were not included when calculating means and standard errors. as 80 ppm NO3-N was also the upper detection limit for those particular analyses. Lysimeters were not sampled during three extended periods: days 140-201. 410-442. and 536-575 (see Appendix III). For these periods. nothing could be definitively inferred about the pattern of nitrate concentrations. Other ”missing" dates. especially for the alfalfa and forest lysimeters. indicate the lysimeter did not provide a sample 89 90 on that date. The lysimeters usually retained a vacuum even if no sample was obtained. so inadequate soil moisture near the cup is postulated to explain "dry" samplings. Individual lysimeters did occasionally fail due to leaks in the rubber tubes. but these were isolated. random events. Each land use was considered separately for statistical purposes. Statistical analyses of the lysimeter data were complicated by differences in the timing of nitrate concentration patterns. Maxima and minima between lysimeters often occurred several weeks apart. and particular time periods defined increasing trends for some lysimeters and decreasing trends for others. The statistical effect was to "dampen" the mean concentration patterns and cause fluctuations in the standard errors for those periods (Figures 21 and 22). Due to low sample numbers (typically 10-15 in corn. 3-8 in alfalfa). a normal distribution had to be assumed for each sample data. Mean nitrate concentrations and standard errors were calculated on a per-sample-date basis (Figures 21 and 22). The means were considered to be reasonable estimates of the mean nitrate concentration which reached the 3C horizon on a given date. for each specified land use. §§nerimentgl Control The lysimeter data were collected under actual farm management conditions. which reduced experimental control. A 91 major question was the variability of center-pivot nitrogen applications across the study site. In late June 1989. 222 bottles were placed along the soil-sampling grid to collect samples from a lagoon-waste application. Due to a mechanical failure of the center-pivot system. only 40 samples were obtained, all from the southeast corner of the grid. Table 10 lists the nitrogen analysis results for the applied lagoon waste. Table 10. Nitrogen Analysis of Applied Lagoon Haste. Inorganic N Organic N Total N (ppm as NHQ. N03) inpml. (DDQL Mean 135.3 150.6 285.9 Std. Dev. 31.7 33.0 38.5 CV (t) 23.4 21.9 13.5 The low coefficient of variation (CV) for total Kjeldahl nitrogen (TKN) suggested that a relatively uniform nitrogen concentration reached the soil surface. The higher CV's for the inorganic and organic nitrogen simply indicated differences in the proportion of these fractions. The inorganic fraction was over 95 percent ammonium. A recent study indicates that the organic fraction could be mineralized within two weeks at 25 degrees centigrade (B.G. Ellis and J.R. Crum. personal communication. 1989). Therefore. the TKN values can be practically considered as 92 the best indicator of nitrogen concentration variability which would affect the lysimeter measurements. Reliable measurements of application-volume variability could not be made. as the center-pivot movement was halted during application. Thus for even a small portion of the grid. the total N application could not be accurately quantified. This essentially left site-wide application variability as an uncontrolled factor. The inability to define this variability was unfortunate. but realistically it would have characterized only a single event. Factors such as lagoon contents. wind speed and direction. corn height. temperature. etc. may have affected each irrigation event differently and influenced the spatial variability of nitrogen applications. Thus. establishing control under actual management conditions was practically unfeasible over a growing season. It was not possible to reliably quantify the total nitrogen application over a growing season. When lagoon waste was applied as nitrogen fertilizer. nitrogen concentrations were assumed and liquid was applied to the nearest 2.5 millimeters. Applying a uniform. prescribed amount of nitrogen to the lysimeter clusters was unlikely. especially through a center-pivot system. Also. farm records were not detailed enough for precise quantification of nitrogen inputs. Therefore. no attempt was made to calculate a thorough nitrogen budget. A partial nitrogen budget was 93 calculated for corn. considering only the recorded N- applications and corn yields (see "Corn” below). Soil Water Balance A soil water balance was calculated for a Kalamazoo Loam soil using Thornthwaite's method and actual 1987-89 precipitation and temperature data (Table 11: Figure 20). The water balance did not consider irrigation inputs. but served to indicate when rainfed leaching below a one-meter depth could be expected. The soil waster storage capacity was calculated from soil survey data (Austin. 1979) and KBS bulk density measurements (D.J. Reinert. personal communication. 1989). The calculated water capacity was 15.1 centimeters to a one-meter depth. The calculated 1987-89 soil-water balances correlated with the relative 1987-89 corn-grain yields in Kalamazoo County. The 1987 and 1988 growing seasons were characterized by low soil-water storage values (Table 11). resulting in low corn-grain yields for both years (Rossman. et al.. 1989). In 1989. soil-water storage values were high for the entire growing season. placing little water stress on corn crops. For Kalamazoo County. 1989 corn-grain yields were approximately double the 1987 and 1988 yields (Rossman. et al.. 1990). Assuming similar management practices. the low yields in 1987 and 1988 would have left a larger amount of nitrogen in the soil profile than in 1989. 94 Table 11. Soil Water Balance. 1987-89 (Thornthwaite_mgthodl. Soil storage capacity - 15.10 cm of water in upper 1 meter. Mean Total Start Bvapo- Final Month Temp Precip Storage trans Storage Surplus A (°C) (cm) (cm) _*‘(cm) (GEL..___fll(CEl 1987 Jan -3.1 2.96 10.00 0.00 12.96 0.00 Feb -0.9 1.13 12.96 0.00 14.09 0.00 Mar 3.6 3.19 14.09 1.20 15.10 0.98 Apr 10.3 5.89 15.10 4.71 15.10 1.18 May 16.8 3.28 15.10 8.48 9.90 0.00 Jun 21.3 5.01 9.90 9.09 5.82 0.00 Jul 22.6 6.38 5.82 8.73 3.47 0.00 Aug 19.9 17.05 3.47 11.23 9.29 0.00 Sep 16.3 13.18 9.29 7.15 15.10 0.22 Oct 7.8 6.44 15.10 2.83 15.10 3.61 Nov 5.6 6.11 15.10 1.61 15.10 4.50 Dec 0.1 12:94 15.10 0.01 15.10 12.93 ANNUAL 10.0 83.56 55.04 23.42 1988 Jan -5.4 5.40 15.10 0.00 20.50 0.00 Feb -5.5 4.10 20.50 0.00 24.60 0.00 Mar 2.1 6.00 24.60 0.64 15.10 14.86 Apr 9.0 7.01 15.10 4.05 15.10 2.96 May 15.9 3.18 15.10 8.07 10.21 0.00 Jun 20.2 3.61 10.21 8.06 5.76 0.00 Jul 23.2 10.66 5.76 12.00 4.42 0.00 Aug 22.3 12.25 4.42 12.44 4.23 0.00 Sep 16.3 16.37 4.23 7.19 13.41 0.00 Oct 7.0 12.71 13.41 2.51 15.10 8.50 Nov 4.7 14.32 15.10 1.33 15.10 12.99 Dec _r2.4 __5.09 15410 0.00 20.19 9.90 ANNUAL 9.0 100.70 56.29 39.32 1989 Jan 0.1 3.60 20.19 0.02 15.10 0.00 Feb -6.2 3.48 15.10 0.00 18.58 0.00 Mar 2.4 6.82 18.58 0.89 15.10 0.98 Apr 6.6 5.02 15.10 3.07 15.10 1.18 May 12.7 18.80 15.10 7.30 15.10 0.00 Jun 19.4 13.66 15.10 11.93 15.10 0.00 Jul 22.3 9.96 15.10 14.14 11.45 0.00 Aug 20.2 11.45 11.45 11.69 11.27 0.00 Sep 15.3 14.30 11.27 6.92 15.10 0.22 Oct 10.5 2.63 15.10 4.38 13.45 3.61 Nov 2.2 17.39 13.45 0.63 15.10 4.50 Dec -2.4 5.09 15.10 0.00 20.19 12.93 ANNUAL 8.6 112.20 60.29 51.91 95 .mwlhflmfl .EGOJ OONBEUHDM MON OUCUHUQ hflufltlflflom .ON Ohflufih 8GP _ 8a.. _ nmmw fifi§<§ufiDZOw<fifi§<§ufiOZOw<fi B9 .m 22. u 0 BB 98 com com com coo com com cow 0 .inlarinlrlmewimumfillLllLllrliriirwim. meow.lL-lIrli. -0 m. .8. (LULU) 13d .10 uogzeudgoexd .. - ..ONr ._ . ...- 1.8—. 1...... ...... .... ............... mmsoohn—iuhulxhha -... 5.. .. .. ...: fiomP (I I...) .000. EE 5 292 25:02 mmfimmr .mocm_mm hmE>> :90. mm! 96 The surplus values (i.e. profile drainage) indicated drainage below one meter for the fall and spring months in each year. For years 1987 and 1988. 99-100 percent of soil- water drainage occurred between October 1 and April 30 of the following year. For 1989. 67 percent of the surplus soil water drained during the same months. Most of the predicted growing-season drainage in 1989 occurred during May. The soil-water balance for 1987-89 indicated that even with relatively wet growing seasons. the majority of soil-water drainage can be expected between October and April. Comparative Lgnd-Use Effects There were several differences in lysimeter results according to land use. The average number of dates in which a lysimeter provided a sample was 62 under corn. 41 under alfalfa. and 13 under hardwood forest. Large differences in sample numbers were noted between clusters one and two (corn) and cluster three (alfalfa) even though soil profile differences were minimal. Although water management cannot be ruled out as a variable. differential water uptake above the 28t2/3C boundary according to vegetation type may have affected sample numbers. Both alfalfa and forest vegetation could extract stored soil water during May and June. when corn was still immature. The number of samplings may indicate the relative frequency of water movement into the 97 3C horizon. but nothing definitive can be said about the total volumes of percolating water. The relative magnitude and temporal variability of nitrate concentrations also differed with land use. Seasonal peaks produced much higher maximum values under corn than alfalfa (Table 12). The minimum nitrate concentrations under corn were comparable to or higher than the maximum values under alfalfa. The maximum concentration peaks for the corn lysimeters appeared to be bi-modal. with the highest peaks occurring in July and slightly smaller peaks in November/December. The mean minimum concentrations under corn were typically 2-3 times higher than the mean minima under alfalfa. The maximum values under alfalfa were recorded during spring. when minimum values are recorded under corn. Nitrate concentrations under forest vegetation were consistently under two parts per million. with the exception of a single outlier in February 1988. Forest lysimeter samples were obtained principally between April and June 1988. November and December 1988. and April to June 1989. These periods were during or immediately following periods of predicted soil drainage via Thornthwaite's method (Table 11). Corn The cornfield lysimeters (clusters 1. 2. 5. and 6) displayed marked variation in the patterns and magnitudes of 98 .Nom. m .mmN. av .mm. an v.v .mhmv w .mva 6N <\z n.v .mbm. OH .NvN. om .mm. vH N.v amhmv n <\z .mw. wN n.v .Nmmv v Ath. NN .mwv an v.n .mhmv N Ath. ca .omv OH n.n .mhmv w .vhN. NH ..o. v N.n .Namv N .th. an Ash. m H.n «to. om.«.w.nm,-....-i-i:r253? ,. _ ...... 3.2.85.3»)- --Nw..isoe§mi we»??? spawns: essaxs: Essawez ‘hd‘hd‘ .Hmw. be .va. mb Anon. mw «nan. DH .ovu. hN v.w .Hmw. ON .Hmvv mm “anv on .mmN. m .Ovnv hN n.w <\z <\z cuoowudmhnwmw-¢nmtuwamfi 120 volumes, cluster one leached 2-4 times as much nitrogen into the 3C horizon during the fall months. Relatively high amounts of ammonium were found in the 28t2 and 3C horizons of the cluster one profile, while little ammonium was found in cluster two. The ammonium analyses for those horizons provided three consistent sub- sample analyses, so the value appeared reliable. The high ammonium values in cluster one may have been due to the May 1988 lagoon-waste application. The inoraanic fraction of the waste was over 90 percent ammonium, and some may have percolated before it could be nitrified. The nitrogen depth profile under alfalfa displayed high values only in the upper meter of soil (Figure 26). The Btl ammonium value in Figure 26 was adjusted due to an erroneously high-ammonium subsample. 0f the three Btl sub- samples extracted. two sub-samples contained only one ppm ammonium, while the third contained nearly five ppm. The five ppm value may have been due to alfalfa root (nodule) material in the third sub-sample. The profiles all displayed a general decrease in total nitrogen with depth. but with several periodic nitrogen peaks. Denitrification in the well—drained subsoil was unlikely. The soil-water balance (Table 11) predicted a surplus of 22 centimeters for October and November 1988. Given the 15 centimeter water-holding capacity of the upper meter of soil. and assuming the 3C banded sands had a water- holding capacity of 10 centimeters per meter of soil. 121 miscible displacement would have moved a solute front 1.7 meters downward. Thus nitrate at the soil surface would leach to 1.7 meters. a depth which coincided with the nitrogen peak in cluster one. However. the nitrate was probably concentrated below the soil surface and should have theoretically leached more deeply. Because of cation adsorption. nitrification. and root uptake effects. the leaching of ammonium to 1.7 meters by displacement was considered unlikely. This would suggest that displacement was not the only nitrogen transport mechanism. Wild (1972) found that nitrate leaching was slower than that predicted by displacement theory. and attributed the finding to water by-pass flow through large continuous pores. When soil is at high water contents. a large percentage of precipitation may percolate directly through the soil while displacing little of the existing soil water (Quisenberry and Phillips. 1976). This macropore flow may accelerate the transport of some solute. while retarding the transport of solute contained in smaller pores (Tyler and Thomas. 1981). This uneven transport can result.in a breakthrough of solute well ahead of the major solute front (Wild and Babiker. 1976: Priebe and Blackmer. 1989). Many studies confirm a high degree of spatial and temporal variability in measuring solute transport rates and depths of solute fronts (Biggar and Nielsen. 1976: Richter and Jury. 1982: Priebe and Blackmer. 1989). 122 The patterns and depths of nitrogen distribution in these profiles were inconsistent with what can be explained by displacement theory alone. It is hypothesized that some of the solute which occurred at depth was due to macropore flow. This may explain the high ammonium content in the 3C horizon of cluster one and the occasional ammonium peaks at depth in all of the profiles. If macropore water was percolating during or immediately after a lagoon-waste application. the ammonium could be transported before being adsorbed or nitrified. The occasional peaks in nitrogen contents at depth were attributed principally to soil hydraulic factors. Zones of high water content were occasionally observed during the profile sampling. where water perched above an apparent hydraulic discontinuity. Since most of the nitrogen was highly water-soluble nitrate. this would have effectively concentrated the nitrogen within those zones. Thus the nitrogen peaks really represented increased soil-water contents where the nitrates tended to reside. Table 15 presents the same data expressed according to actual horizon thicknesses and summed for one-meter increments. All values refer to elemental nitrogen. with the total sub-divided into nitrate and ammonium contributions. Large differences were apparent in the total amount. distribution. and form of the soil nitrogen. Total nitrogen was twice as great in cluster one as cluster two. mostly due to the near absence of ammonium in the cluster-two profile. 123 The soil profile in cluster three had the lowest amount of total nitrogen for each of the upper three meters. All profiles contained only small amounts of total nitrogen at the deepest sampling depth. Although the cores represented only three samples. it was clear that large amounts of nitrogen remained in the upper meter of soil. The fall lagoon-waste applications added large amounts of nitrogen. unfortunately confounding the attempt to monitor the movement of post-harvest residual soil nitrogen. The total nitrogen amounts suggested that as of mid-December 1988. not much nitrogen had moved below two- meter depth. It could be expected that internal soil drainage through winter-spring 1989 would further leach the nitrogen. to a point where subsequent corn crop recovery was unlikely. Soil Variability and Nitrate Measurements Several methodological problems confounded the research objective to relate soil variability and soil-nitrate measurements. The lysimeters were originally installed to monitor nitrate concentrations in "representative" field soils (Rice. et al.. 1986). The lysimeters were then used for the present study under the false assumption that an unknown range of soil variants were included. Particle-size analyses were not performed when the lysimeters were installed. so the assumption was not directly negated. Table 124 16 lists the control-section clay estimates calculated from the block-kriged Btl clay contents. the 2Bt2 sample mean of 10 percent clay. and actual Bt thickness measurements. The estimated control-section clay contents varied over a range of only six percent. from 19 to 25 percent. Although the estimates certainly contained some error. it was clear that only a small portion of the study-site soil variability was included in lysimeter clusters one. two. and three. Table 16. Control-section clay contents for study-site lysimeter,clusters. Control- Btl 28t2 Section Lysimeter Clay Thick. Clay Thick. Clay (8) h.-(cm) (8) (cm) _g (8) 1.1 25.8 40 10.0 28 22.6 1.2 25.7 47 10.0 15 24.8 1.3 26.0 32 10.0 14 20.2 1.4 26.9 38 10.0 64 22.8 2.1 26.9 35 10.0 21 21.8 2.3 25.1 36 10.0 23 20.9 2.4 25.4 30 10.0 10 19.2 3.1 26.0 28 10.0 42 19.0 3.2 25.9 41 10.0 42 23.0 3.3 26.8 32 10.0 25 20.8 3.4 27.3 32 10.0 26 21.1 Statistically relating soil variability to nitrate measurements was difficult. A major problem was to choose a nitrate measurement or statistic which was meaningful as an indicator of nitrate leaching. The lysimeter data reflected only nitrate concentrations. not fluxes. Thus. totals were meaningless. Seasonal nitrate concentration means were 125 suspect due to temporal variability in peak timing and soil moisture. Finally it was decided to use the nitrate concentration maxima and minima. as they could be most reliably identified on a seasonal basis. A series of multiple regression equations were performed to relate the maximum and minimum nitrate concentrations to soil factors at each lysimeter. The independent variables were Btl clay and thickness. 28t2 clay and thickness. combined Bt thickness. control-section clay content. and lysimeter cup depth. All F values were insignificant at the 0.1 confidence level except 23t2 thickness for the maximum fall 1988 values. In that case. the r-sguared value was 0.52 with a positive slope. Upon examination. a single statistical outlier was revealed to be the source of the "significance". When removed from the regression. the P value was insignificant and the r-sguared value was 0.02. It was concluded that maximum and minimum nitrate concentrations could not be related to Bt thicknesses or cup depth. Several problems became apparent during the regression analysis. Given the demonstrated variability of the lysimeter measurements. it was improbable that nitrate concentrations could be related to such a limited range of Bt-horizon characteristics. Because regressions were done separately for corn and alfalfa. the sample numbers for each regression were limited to seven in corn and four in alfalfa. Clay contents were not available for'clusters four. 126 five. and six. which would have doubled the regression sample numbers. The dependent variables (nitrate maxima and minima) were unlikely indicators of nitrate leaching potential. and the range of Bt characteristics was limited. Ideally. the range of soil variants should have been regressed against seasonal nitrate flux totals. This was eventually accomplished only through computer modelling. II. CERES Maize Modelling Results Estimated ,-§Oil-_!ixd;aulic Characteristics The morphological data for the 22 soil types (7-28% control-section clay) were input into the SOILW sub-routine. which estimated soil hydraulic characteristics (see Ritchie and Crum. 1989: also Methods). The Ap characteristics were held constant for all soil types. The argillic-horizon texture and thickness were input according to the results in Table 9. All 22 profiles were limited to a one-meter depth. so simulated leaching output could be compared against actual suction lysimeter data. The thickness of the 3C horizon was therefore dependent upon the thicknesses of the overlying horizons. so the one-meter limit could be maintained. Table 17 lists the estimated hydraulic characteristics for each soil type. Definitions of the CERES model terms are as follows: SWCON- soil-water drainage constant. the fraction of excess soil water drained per day (unitless). Excess water is the difference between the horizon saturated water content (SAT) and the drained upper limit (DUL). SAT— saturated water content for t e soil horizon. expressed as cm water per cm soil. DUL- drained upper limit: soil-water content comparable to "field capacity" for the soil horizon. Expressed as cm water per cm3 soil. In Table 17. the only DUL values are for the Btl horizon. because the Ap. 28t2. and 3C values were constant for all soil types. 127 128 lower limit of plant-extractable soil water for the horizon. in cm water per cm soil. Corresponds to a metric potential of approximately -2 MPa (Ritchie and Crum. 1989). As with DUL only the Btl LL values are reported in Table 17. LL- total plant-extractable soil water in the profile. PESW- comparable to water-holding capacity expressed in cm. Table 17. CERES Soil Hydraulic Characteristics from Soil ,rMorphpquthatal- a. um-..flvumq. ." - I”- NOTE: Values in bold type are adjusted values (see text). Control- Section ESTIMATED INPUT _Bt1 Horizon Profile Class SWCON SWCON DUL L PESW (8 Clay) - .9.. h_ -N .- é-(cm319m_):t. (cm) 7 .58 .68 .164 .044 13.5 8 .58 .62 .166 .045 13.3 9 .54 .57 .184 .064 14.0 10 .51 .53 .194 .075 14.3 11 .48 .49 .207 .088 14.3 12 .46 .46 .215 .097 14.4 13 .43 .43 .227 .109 14.4 14 .42 .42 .231 .114 14.5 15 .39 .39 .242 .125 14.5 16 .38 .38 .247 .130 14.5 17 .36 .36 .253 .136 14.5 18 .34 .34 .260 .143 14.6 19 .32 .32 .268 .150 14.6 20 .30 .30 .278 .160 14.7 21 .29 .29 .283 .165 14.7 22 .28 .28 .287 .169 14.7 23 .26 .26 .294 .176 14.7 24 .25 .25 .295 .177 14.8 25 .24 .24 .301 .181 15.0 26 .23 .23 .306 .186 15.1 27 .24 .24 .302 .182 15.3 28 .22 .22 .309 .187 15.6 In Table 17. the bold numbers indicate that the SWCON value was adjusted above the estimated value. This was done because the Ap horizons of the coarsest-textured profiles 129 were coarser than average. and also because the 3C banded sands were weakly developed. The SWCON value is a drainage coefficient for the entire profile. and is intended to reflect the drainage rate of the profile's slowest-draining layer. The adjusted SWCON values were thought to be better reflections of profile drainage characteristics for those coarse-textured soil types. With increasing clay content. the estimated SWCON values decreased. and LL and DUL increased. The saturated value (SAT) equalled 0.337 for all Btl horizons. because the estimate is based upon a constant horizon bulk-density value. Except for the 27 percent control-section class. all estimates followed a consistent trend. The estimated hydraulic variables for the 27-percent class were similar to those for 25 percent control-section clay. except for a slightly higher plant-extractable soil water (PESW) (Table 17). The particular combination of soil- layer textures and thicknesses for the 27-percent class apparently resulted in slightly different hydraulic estimates. The estimation of these values was critical for modelled leaching predictions. Not only do the values affect drainage conditions. but the soil-moisture values partly determine the modelled mineralization. nitrification. nitrogen uptake. and denitrification rates. Thus the hydraulic factors interact to affect both the simulated nitrogen availability and subsequent transport. 130 Irrigated Nitrate Leaching The CERES model output for years 1987-89 predicted approximately the same amount of nitrogen losses for each year. despite large differences in precipitation (Figures 27-29). Annual nitrogen losses. tabulated from planting date to planting date. were between 120 and 145 kilograms of nitrogen per hectare. The graphs revealed three consistent trends with increasing control-section clay content: 1) the total nitrate leaching gradually decreased. 2) the proportion of nitrate leached prior to January decreased. and 3) the amount of denitrification increased. These effects were all consistent with the literature regarding soil texture and nitrogen losses (e.g. Lund. 1974: Devitt. 1976). The generally slower drainage of finer-textured profiles decreases the soil drainage rates and increases the potential for reduction/denitrification. In dry years. the finer-textured profiles can increase water availability and nitrogen uptake. thereby reducing residual soil nitrogen. These effects were all reflected in the CERES output (Figures 27-29; Table 18). The effect of weather patterns. even under irrigation. was noticeable in the yield and nitrogen-loss data. Table 18 lists selected leaching-related outputs for the 18 percent control-section clay soil. which was representative of the majority of the study area. Although there was no major difference in the predicted water or nitrogen stress between 131 mmlhmma .nsoHuflpgoo pougoauuw hops: anH cwoouuwz 95925 .3 at >20 cozoom- mm mm vm mm ON we 9 v— .NN- mm m- mp — . .1 O F. (V O'll'tlll.’ - b P h — \. .-\ i . . was ' 7////////////////////////// ms: 7////////////////////////// m I/////////////////////////// WW .... O 6250 N.. or m mw : m p . . . _ p ...:‘I. -. via. .000.- ..t-. u'al.£-‘. .‘bwa .:o. o. .hn ousowm k m - o ow m / fl ow w M 8 y w on y y fl 00.. m . our -20: ...... some (Bu/5x) 630'! N 132 .mmlwme .ngowuapcoo pousoauuu hops: mnoH gooouufiz 1 £3.95 3 ob >30 cozoowlozcoo mm mm gm NN ON m... or mm mm mm 5 rm .1 '0) ,— \ . 16666666666466666666666666664666’ VZ666666666666666666666666666666606 .3 mr 3. N_. m.._ _ : or ,_ 96666666666666666666666666666666666/ m h~ 666666609999999999996666666666609979(D A W] litl .mm ousowm 69 09 9: 3.2550 m :36. .3 omzommm I Ema—smog oozommj m 00—. (en/6x) $301 N 133 .omnmomH .mgoHuapgoo poucoHuuH Lops: nmoH cwoouqu mm vuzowh 2:925 >2 as; >20 cozoomlozcoo mm mm vm mm om 9 m— 3 w. or m mm. mm .mm .5. 9 2 9 9 3 m a -nlll. / . ...p..... . .3510 w M / / / / fl ” om / / y y ,/ / ov / / / / ” fl / / ............... ,/ / ow / / ” fl / / -1. ” fl om / / / / 7 fl 4 GOP ; m ......... GNP . .i;§%a§i£- ---.) i- 3.-§:;fs:ovw (Itmm;owr (lull!) I. / 8:56me W =64 >9 omsomg - Emocmo B 8:83 7 (en/5)!) 3501 N 134 years. the predicted grain yield for 1988 was only 80 percent of the 1987 and 1989 yields. This was apparently due to the high temperatures in July and August 1988: the monthly mean for August 1988 was more than two degrees centigrade higher than in 1987 or 1989 (Table 11). The predicted result was a shortened grain-filling period and a kernel weight which was 80 percent of the 1987 and 1989 values. This kernel-weight variation accounted for the grain-yield differences. The predicted differences in mature biomass were also due to the abbreviated grain-filling stage. as evidenced by differential biomass additions during the modelled grain-filling growth stage. Since the CERES fertilizer inputs were identical to the 1989 farm management practices. the predicted and actual yields for 1989 were compared. The model predicted 178 bushels per acre. The actual grain yields were 158 bu Ac’l without starter fertilizer and 173 with starter. The results were sufficiently close to lend credibility to the model predictions. in terms of nitrogen uptake and residual soil nitrogen. It was unknown how well the modelled variety (Pioneer 3780) matched the phenological characteristics of the planted variety (Great Lakes 582). in terms of growth stages. required season length. etc. The predicted leaching losses were inversely related to the amount of nitrogen uptake (Table 18). For 1987 and 1988. less than 10 percent of the predicted total nitrate leaching occurred before September lst. In the relatively wet growing 135 season of 1989. 30 percent of the total leaching occurred during the same period. Thus. the majority of the predicted annual nitrate leaching was strongly related to post-harvest soil-nitrogen contents. The estimated denitrification losses were nearly identical for the three model years (Table 18). The CERES model predicts denitrification based upon soil nitrogen concentrations. available carbon. and an excess of soil water above the drained upper limit (DUL). Because the first two factors varied little between runs. it appeared that soil moisture was the likely determinant in this case. The soil-moisture argument was consistent with the heavy post-harvest precipitation in 1988. The leaching losses for the coarsest soil profiles (e.g. 7-10 percent clay) were expected to be relatively high due to the estimated hydraulic characteristics (Table 17). The model predicted slightly less total leaching for these profiles (compared to 11-13 percent clay) in 1987. and equivalent or slightly greater leaching in 1988 and 1989. The predicted nitrogen uptake was constant for all soil types within a given year. so the annual nitrogen inputs and outputs were the same. The leaching differences were not related to annual drainage volumes. nor to high-volume leaching events. The leaching output file (OUTLCH) revealed slight but consistent differences in the incremental nitrate leaching across soil types. Final residual nitrogen values in the soil profiles were unrelated to the total nitrate leaching. 136 A possible explanation for this was the nitrogen mineralization sub-routine was affected by the low soil- moisture in the coarse-textured profiles. The low DUL values and high SWCON values would drain a relatively large amount of excess water quite rapidly. The lack of soil moisture would slow mineralization and reduce the inputs to the soil inorganic nitrogen pool. This effect would not be observed in the standard CERES output files. but would explain slight differences that were calculated in the total nitrogen budgets for each soil type. Ba_inf.e_d.-_Ni crateleaching Table 19 and Figures 30—32 summarize the rainfed leaching output. For the rainfed runs. a single broadcast of 125 kg N ha"1 was used. compared to the split applications totalling 288 kg N ha'1 for the irrigated runs. Although the rainfed simulations input less than one-third of the irrigated nitrogen fertilization. the rainfed runs for 1987 and 1988 leached between 75-90 percent of the corresponding irrigated runs. This was apparently due to the moisture stress which resulted in poor nitrogen uptake and yields under rainfed conditions. The difference between applied nitrogen and nitrogen uptake was similar between the rainfed and irrigated runs for 1987 and 1988. Hence the nitrate leaching totals were comparable. For 1989. the difference for the rainfed simulation was only one-fifth 137 Table 18. Selected model outputs for irrigated simulations. 1_,_.U1_11§911_input_f9r lfimperge t 99.9W9}FS§C£109131§Y- .— EIQPQIEXW1111__“___M1__,-_m_12311_-.11”12§§fl.A- . 1989 Grain yield (bu Ac’l) 186 149 178 Mature biomass (kg ha'l) 20771 18567 19815 Nitrogen uptake " 246 215 234 Nitrate leaching " 121 130 125 Denitrification " 7 8 7 Table 19. Selected model outputs for rainfed simulations. ,H_umww_gMSoilminput for_18 percent controlesection clay. Propertywwuw,g_vm 71.”.,_,___19871.-,111”198§_11_.__"1989 Grain yield (bu Ac'l) 58 55 161 Mature biomass (kg ha'l) 7387 5711 15280 Nitrogen uptake ” 87 68 115 Nitrate leaching " 99 125 77 Denitrification " 4 6 5 138 ____ Denltrifled M”; ched by April g ‘ Leeched by January I Lea ! 1 1 1. ~ ' 1 1 1 m ‘ '//////////////_/////// 29, an I////////////////////. R an: . ’////////////////////// {8 g a“ ///////////////////////. a l ’. | l ¢ co N I 1 i i N 1 '/////////////////////////. a] 7////////////////////////. {5 17c '///////////////////////// ///////////////////////////. 0’ ’/////////////////////_//////. ' ’//////////////////////////// ’//////////////////////////// 20 17 '////////////////////////////z 13 15 12 14 16 181 Control-section clay (% by weight) 7////////////////////////////, I//////////////////////////// 3: ////////////////////////////// '//////////////////////////////, m ’///////////////////////////////// Q 97%6092Z60972Z60’/Zb00%%0’/ 1s 1 1 1 1 1 1O 1 1 1 1 i 1 s z ; 1 . ~ I 1 1 1 1 8 8 8 8 ° (Bu/511) 880") N 1mg ..."... ....- Nitrogen loss under rainfed conditions. 1987-88. Figure 30. 139 9:90? >b 5 >20 cozoomlotEoo ON 9 m— 3 NH op m pm OH up ML. .0... 2 m n 01003 I cottage - Eo<>o genomes - bmncmn >np : ,om ON 0v (BU/5’1) 5801 N OOH ONH 11111 .103 140 .omlmomn .uaofluwpcoo poucugh hops: onoH cooouuwz 9:903 >2. .5 >20 cozommiozcoo mm on em mm on E 0. E. .9 2 km. mm mm 5 ..9 mp m. or I. fill-lull b1 In] .P 1’! (P1 1b a. 0106...! .’ ///////////////////////// 7/////////////////////////// .01.,1; ..ll‘oon. .. III-‘1'! seq 3 .8583 I 8.350 m o a1111o 1;;t;ON .. 111ziov . . ..... o . ...-1 .o . ...: --..-.sioop 111111111111111m Bunsen .3 00:03... E .Nn oupuwh m ‘0 G) >03 (Bu/Bx) N 141 that of the irrigated run for most soil types. resulting in significantly lower predicted leaching. The effect of soil type on predicted nitrate leaching was more pronounced under rainfed conditions for all three years. The effect was most obvious for the relatively wet 1989 growing season. Whereas the irrigated runs showed approximately a 10 percent difference in leaching across the range of soil types (relative to maximum leaching). the 1987 and 1988 rainfed output showed a 20-25 percent difference. For 1989. the rainfed output indicated nearly a 50 percent reduction in predicted nitrate leaching. going from 7 percent to 28 percent control-section clay. The degree of rainfed leaching variation due to soil variation was again apparently due to predicted moisture stress and its effect upon nitrogen uptake. The range of rainfed nitrogen uptake across all soil types for 1987 was 78—98 kg N ha’l: for 1988. 52-71 kg N ha'l: and for 1989. 77-123 kg N ha'l. Both the range and magnitude of nitrogen uptake values increased with decreasing moisture stress. When moisture stress was severe. the relative effects of soil hydraulic properties were minimal. But in years with adequate precipitation. the higher water-holding capacity of the finer-textured soil types resulted in lower moisture stress. higher nitrogen uptake. higher total biomass and grain yields. and large decreases in predicted nitrate leaching. 142 Estimates of~Soil-water Nitrate Concentrations and Fluxes The output contained in the OUTLCH files provided values for both drainage volume and nitrate flux. Thus it was possible to calculate the predicted soil-water nitrate concentration for each weekly increment. Based upon the estimated control-section clay contents of lysimeter clusters one and two. the modelled soil profile with 23 percent control-section clay was used for comparison. The results are presented in Figure 33. along with the lysimeter measurements for that time period (note: the days are n3; Julian calendar day numbers). The model closely predicted peak nitrate concentrations and the temporal pattern of soil-water nitrate variations. The predicted peaks were unimodal and described a rather regular temporal pattern. The actual lysimeter measurements were bi-modal and also displayed a well-defined pattern. In both 1987 and 1988. the modelled nitrate peak occurred between the measured bi-modal peaks. The modelled nitrate concentrations decreased earlier- in time and to a greater degree than the field nitrate measurements. The missing data points for the modelled concentrations were due to two factors. One. the model re-initialized the soil-profile initial conditions for each run. which lowered the initial soil-moisture and drainage from the end of previous runs. Two. when soil profile drainage did not occur. there was no way to calculate an estimate. For the 143 .haao cowuuwmt Houusoo «nu poHHopoE ascu .:OfiUUHUflUOGOU GUDHUMG w> Uwhfiucflz Ill,"‘l‘l‘nll"l‘ I IWMMWW.._o1poE .u 002 «.2385. .1 _ 1"!"1’ -‘ll- lull-ll ' .l '7‘" to? .m 2.... .0. >09 880 0% 08 0%. one 08 Em ll—l 'Flp‘. {LII—..luhu‘rll—IIP‘LI omp om 1:1..1P111P1L L111~11p1..1_1. .F 00' -..! a... .40:- 00430 as ’II'IIIIIII'IIII" ' I; [Ilil'l‘I'l'l'I'l’lll-l ’0 l’l'l'lll!"il""l 5300 EH .>20 20200012500 0 0:00 22:2 00:80.). ...; 09:80.2 11b11..1.1>.1 .nn ousouh o ..-. o (wdd) “ouoo SON pamseew / peuepow 144 field-measured values. missing data points indicated that a sample was not obtained on that date. A "valley" occurred between the measured nitrate peaks around days 60 (9/1/87) and 420 (8/26/88). These dates correspond roughly to the end of the corn grain-filling stage. and the temporarily depressed measurements were attributed to nitrogen uptake and a subsequent decrease in soil-nitrogen levels. After grain-filling. it was hypothesized that continued mineralization. decreased nitrogen demand and decreased plant water uptake could provide a temporary pulse of nitrate through the soil profile. The minimum predicted values were approximately half of the measured minimum concentrations (days 300 and 660). This probably resulted from two fall lagoon-waste applications. which were not included in the modelling because they were unknown at the time. Also. the model does not account for a frozen soil surface. which restricts field infiltration and drainage during the winter months. However. the Thornthwaite water balance and the CERES drainage predict similar drainage totals between November and April of each run. With no incremental information for the Thornthwaite estimates. it is difficult to test the frozen-surface and drainage hypothesis. For the sake of comparison. the incremental drainage volumes predicted by the CERES model were multiplied by the corresponding measured nitrate values for dates where both 145 .owlbwmd ocwzosou obguuH: 0>HucHsE=u pouoHpoum .vn ousoHu I'll‘."! III" I!“ i'ullll'.laul:la - 0' -1a'-l‘» I. ’Il'.b moz o2_moo_2 a no: 00522.2 ‘1'"! 'Il!‘ III, ‘I I - ‘l" I ’10,- .... l . .+. \ l i"...'l'l..l.‘."" £02 .N 22. u .0. .89 £8 08 Ba 9a 0a 8. 0m: 0m. 8 111.111.1L..1.111111p"1111~.1|1lll.1h1 cm on o .111...) 7.1.1! 1 :..—1.1.1.1... c 1 1 . . gmgapmmm: ...sa 1.111).1..1|1.1.11lL.11_| 1.. 1: P "C I‘D. ‘79- II... I -5-:2§§2§§-§¥f¥23§ -i:i§§-?i¢¢ ¥ -3 1.1-11. ....... w. .0:: .......... QUE. .... .. . jaw :1. .KW.- . DUDE wmmmmxmmxwm U 8 wk —U x D HUD—U 08800000000 '0 '0. I." ‘10 ...-o It... '8- 9H/N'SON 5» 5111110261 8011 I. I [DVP «22.82 9. 026.3630 0218:. 92:80.. 29.2 “02.822 m> ...0050022. 146 Ill-l I'I’II'III'I'I'” in! [li!‘l’.‘"ll' 0' I II I o illil’lllli'l 88. .m 22. ... .9169 23 8m Em ea 05 8. of cm. 8 00 on .1111..1....P.|1..11._1111..1L1.1.111 .1111L1111111r11111 Old. .v'a al-lo alt-n ill .1 x“ .3. .m U a l . vill- -.'o .‘I ooll III.- csa: . *0 ‘ l... ...... 3.... o,- cloa .0. O . . .... ... Dan—DD Ill", 222-82 9. 026.3530 0.21.03. 9.383 22:2 .2382 m> ..oousmmmfi... 1 mwmmflmmmmfifilmwflms so can: ...-n .3. 3|. 0 .. ... . oo'. ...: .ol. ...-0 ...o. 3.: I... :... .0... I... 2... .11.. :10 III. to... .0:- : .0 ..l. .-. not. 6' I I ' ‘ B I m-m \ O -8 .- .8 ,8. BH/N'SON fix 5111110261 SON --:.:.t:sz§:«-§-.cww 1rog— 147 data existed. The sample mean for clusters one and two was used for the nitrate value. The cumulative leaching was plotted for the 1987-88 and 1988-89 (April to April) years. Figures 34 and 35 display curves for the CERES leaching prediction and also for the calculated estimate using the measured nitrate mean. Despite the several missing dates for measured nitrate values. and hence no incremental estimate. the cumulative leaching estimates were quite close. The model does not account for two fall applications of lagoon waste each year. which were learned about after the modelling was completed. This would certainly raise the CERES leaching estimate. However. this omission was somewhat offset by the missing field data which certainly would have increased the calculated nitrate flux. It appeared the estimates were reliable based upon checks such as Figure 33 and the comparison with the Thornthwaite predictions. Based upon the CERES modelling and the calculated leaching values. the best estimate for the nitrate flux at the RBS center-pivot field is: 1987-88 season : 95-115 kg N ha‘1 1988-89 season : 105-125 kg N ha-1 1) 2) 3) 4) 5) 6) 148 CONCLUSIONS Btl clay content. Btl thickness. and 2Bt2 thickness displayed a high degree of spatial dependence at the KBS study site. The range of spatial dependence varied between 10 and 30 meters. The ZBt2 clay content displayed no spatial dependence at the scale of observation. Block kriging produced an isarithmic map of control-section clay contents. which varied between 7 and 28 percent clay. Soil-water nitrate concentrations varied systematically over time and with land-use. Peak nitrate concentrations were higher under corn than under alfalfa. Under hardwood forest. the soil-water nitrate concentrations were consistently below 2 ppm nitrogen as nitrate. Using actual soil-water nitrate concentrations and computer-modelled profile drainage estimates. nitrate flux estimates were calculated for the range of control- section clay contents at the study site. Under irrigated corn. an estimated 95-125 kg N ha' was leached below a one-meter depth during the 1987 and 1988 cropping seasons. Nitrate leaching was computer-modelled for rainfed corn production. for the range of control-section clay contents at the study site. Predicted rginfed leaching losses were between 85 and 135 kg N ha' for the simulated 1987 and 1988 growing seasons. The CERES Maize model predicted decreased nitrate leaching and increased denitrification losses with increasing control-section clay content. for study-site soils. This effect was most dramatic for the 1989 rainfed simulation. where predicted losses under the finest- textured profile were only 55 percent of those under the coarsest-textured profile. The model accurately predicted corn-grain yields and peak soil-water nitrate concentrations. Close agreement was obtained between modelled nitrate leaching estimates and fluxes calculated from nitrate-concentration measurements and fluxes calculated from nitrate concentration measurements and predicted soil-water drainage. LI TERATURB CITED LITERATURE CITED Adams. J.A.. and J.M. Pattinson. 1985. Nitrate losses under a legume-based crop rotation in Central Canterbury. New Zealand. New Zealand J. Agric. Res. 28:101-107. Alocilja. E.C. and J.T. Ritchie. 1989. Coupling CERES models and dual-criteria optimization techniques for strategic planning. p. 11. In Agronomy Abstracts. ASA. Madison. WI. Austin. F.R. 1979. Soil Survey of Kalamazoo County. Michigan. USDA Soil Conservation Sevice. Beckett. P.H.T.. and R. Webster. 1971. Soil variability: a review. Soils and Fertilizers 34:1-15. Biggar. J.W.. and D.R. Nielsen. 1976. Spatial variability of the leaching characteristics of a field soil. Water Resources Research. 12:78-84. Bouma. J. 1983. Hydrology and soil genesis of soils with aguic moisture regimes. p 253-291. In L.P. Wilding. N.E. Smeck. and G.F. Hall (Ed.) Pedogenesis and Soil Taxonomy. 1. Concepts and Interactions. Elsevier. New York. Brockway. D.G.. and D.R. Urie. 1983. Determining sludge fertilization rates for forests from nitrate—N in leachate and groundwater. J. Environ. Qual. 12:487-492. Burgess. T.M.. and R. Webster. 1980a. Optimal interpolation isarithmic mapping of soil properties: I. The semivariogram and punctual kriging. J. Soil Sci. 31:315-331. Burgess. T.M.. and R. Webster. 1980b. Optimal interpolation isarithmic mapping of soil properties: II. Block kriging. J. Soil Science 31:333-341. Burgess. T.M.. and R. Webster. 1980c. Optimal interpolation isarithmic mapping of soil properties: III. Changing drift and universal kriging. J. Soil Sci. 31:505-524. 149 150 Burrough. P.A. 1983. Multiscale sources of spatial variation in soil. I. The application of fractal concepts to nested levels of soil variation. J. Soil Sci. 34:577- 597. Cameron. D.R.. C.G. Kowalenko. and C.A. Campbell. 1979. Factors affecting nitrate nitrogen and chloride leaching variability in a field plot. Soil Sci. Soc. Am. J. 43:455-460. Campbell. J.B. 1979. Spatial variability of soils. Annals of the Association of American Geographers. 69:544-556. Chichester. F.W.. and S.J. Smith. 1978. Disposition of 15N- labeled fertilizer nitrate applied during corn culture in field lysimeters. J. Environ. Qual. 7:227-233. Devitt. D.. J. Letey. L.J. Lund. and J.W. Blair. 1976. Nitrate-nitrogen movement through soil as affected by soil profile characteristics. J. Environ. Qual. 5:283- 288. D'Itri. F.M.. K.M. Kittleson. and R.L. Kruska. 1985. Spatial analysis of Michigan Department of Public Health nitrate data. Institute of Water Research. Michigan State University. Ellis. B.G. 1988. Nitrates in water supplies. p.75-84. In Water quality: a realistic perspective. University of Michigan. Ann Arbor. Michigan. Endelman. F.J.. D.R. Keeney. J.T. Gilmour. and P.F. Saffigna. 1974. Nitrate and chloride movement in the Plainfield Loamy Sand under intensive irrigation. J. Environ. Qual. 3:295-298. Fairchild. D.M.. 1987. A National assessment of ground water contamination from pesticides and fertilizers. p.273- 294. In D.M. Fairchild (Ed.) Groundwater quality and agricultural practices. Lewis Publishers. Chelsea, Michigan. Flaig. E.C.. A.B. Bottcher. and R.L. Campbell. 1986. Estimation of nitrate leaching using geostatistics. ASAE Technical Paper No. 86-2029. St. Joseph. MI. Food Safety and Quality Service. USDA. 1978. Final Report on Nitrates and Nitrosamines. US Government Printing Office. Washington. DC. 151 Gerwing. J.R.. A.C. Caldwell. and L.L. Goodroad. 1979. Fertilizer nitrogen distribution under irrigation between soil. plant. and aquifer. J. Environ. Qual. 8:281-284. Gilliam. J.W. and G.D. Hoyt. 1987. Effect of conservation tillage on fate and transport of nitrogen. p. 217-240. In Logan. T.J. (ed.). Effects of conservation tillage on groundwater quality. Lewis Publishers. Chelsea. Michigan. Golden Software. Inc.. 1989. SURFER Computer Program. version 4.0. Golden. Colorado. Grigal. D.F. 1985. Note on the hydrometer method of particle-size analysis. Minnesota Forestry Research Notes. Number 245. University of Minnesota. St. Paul. MN. Groffman. F.M.. P.F. Hendrix. and D.A. Crossky. Jr. 1987. Nitrogen dynamics in conventional and no-tillage agroecosystems with inorganic fertilizer or legume nitrogen inputs. Plant Soil 97:315-332. Gutjahr. A. 1985. Spatial variability: geostatistical methods. p. 9-34. In D.R. Nielsen and J. Bouma (ed.) Soil Spatial Variability. Proc. Workshop ISSS and SSSA. Las Vegas. NV. 30 Nov.-1 Dec. 1984. PUDOC. Wageningen. Netherlands. Hergert. G.W. 1986. Nitrate leaching through sandy soil as affected by sprinkler irrigation management. J. Environ. Qual. 15:272-278. Hubbard. R.K.. L.E. Asmussen. and H.D. Allison. 1984. Shallow groundwater quality beneath an intensive multiple-cropping system using center-pivot irrigation. J. Environ. Qual. 13:156-161. Hubbard. R.K. and J.M. Sheridan. 1989. Nitrate movement to groundwater in the southeastern coastal plain. J. Soil Water Conserv. 41:20-27. Hughes. Theresa. 1983. Nitrate contamination of the groundwater of Old Mission Peninsula: contributing effects of orchard fertilization practices. septic systems. and land reshaping operations. M.S. Thesis. Michigan State University. E. Lansing. Michigan. Jones. C.A. and J.R. Kiniry. 1986. CERES-Maize: a simulation model of maize growth and development. Texas A&M University Press. 152 Journel. A.G.. and C.H. Huijbregts. 1978. Mining Geostatistics. Academic Press. New York. Kanwar. R.S.. J.L. Baker. and J.M. Laflen. 1985. Nitrate movement through the soil profile in relation to tillage system and fertilizer application method. Trans. ASAE 28:1802-1807. Keeney. D.R. 1982. Nitrogen management for maximum efficiency and minimum pollution. p. 605-649. In Stevenson. F.J. (Ed.) Nitrogen in agricultural soils. ASA. Madison. WI. Kittleson. K.R. and R.L. Kruska. 1987. Spatial distribution and analysis of groundwater nitrate contamination in Kalamazoo County. Michigan. Research Report. Center for Remote Sensing. Michigan State University. Loehr. R.C. 1984. Pollution control for agriculture. Academic Press. New York. Lund. L.J.. D.C. Adriano. and P.F. Pratt. 1974. Nitrate concentrations in deep soil cores as related to soil profile characteristics. J. Environ. Oual. 3:78-82. MacGregor. J.M.. G.R. Blake. and 8.0. Evans. 1974. Mineral nitrogen movement into subsoils following continued annual fertilization for corn. Soil Sci. Soc. Am. J. 38:110-113. Matheron. G. 1971. The theory of regionalized variables and its applications. (In English.) Cashiers du Centre de Morplogie Mathematique. Fontainbleu. No. 5. Mathers. A.C.. Steward. B.A.. and B. Blair. 1975. Nitrate nitrogen removal from soil profiles by alfalfa. J. Environ. Qual. 4:403-405. McBratney. A.B. 1985. The role of geostatistics in the design and analysis of field experiments with reference to the effect of soil properties on crop yield. In D.R. Nielsen and J. Bouma (ed.) Soil Spatial Variability. Proc. Workshop 1885 and SSSA. Las Vegas. NV. 30 Nov.-1 Dec. 1984. PUDOC. Wageningen. Netherlands. McBratney. A.B.. and R. Webster. 1983. How many observations are needed for regional estimation of soil properties? Soil Sci. 135:177-183. McBratney. A.B..and R. Webster. 1986. Choosing functions for semivariograms of soil properties and fitting them to sampling estimates. J. Soil Sci. 37:617-639. 153 McMahon. M.A.. and G.W. Thomas. 1974. Chloride and tritiated water flow in disturbed and undisturbed soil cores. Soil Sci. Soc. Am. Proc. 38:727-732. Motavalli. P.P.. R.A. Kelling. and J.C. Converse. 1989. First-year nutrient availability from injected dairy manure. J. Environ. oual. 18:180-185. National Research Council (NRC). 1978. Nitrates: An environmental assessment. Natl. Acad. Sci.. Washington. D.C. Nielsen. D.R.. J.W. Biggar. and K.T. Ehr. 1973. Spatial variability of field-measured soil-water properties. Hilgardia 42:215-260. Nielsen. D.R.. J.W. Bigger. and P.J. Wierenga. 1982. Nitrogen transport processes in soil. p.423-448. In Stevenson. F.J. (Ed.) Nitrogen in agricultural soils. ASA. Madison. WI. Nightingale. H.I. 1972. Nitrates in soil and ground water beneath irrigated and fertilized crops. Soil Science 114:300-311. Olea. R.A. 1975. Optimum mapping techniques using regionalized theory. Series on Spatial Analysis No. 2. Kansas Geological Survey. Lawrence. KN. Olsen. R.J.. R.F. Hensler. 0.J. Attoe. S.A. Witzel. and L.A. Peterson. 1970. Fertlizer nitrogen and crop rotation in relation to movement of nitrate nitrogen through soil profiles. Soil Sci. Soc. Amer. Proc. 34:448-452. Olson. R.A. and L.T. Kurtz. 1982. Crop nitrogen requirements. utilization. and fertilization. p. 567- 604. In Stevenson. F.J. (Ed.) Nitrogen in agricultural soils. ASA. Madison. WI. Owens. L.D. 1960. Nitrogen movement and transformations in soils as evaluated by a lysimeter study using isotopic nitrogen. Soil Sci. Soc. Am. Proc. 24:372-376. Pratt. P.F.. W.W. Jones. and V.E. Hunsaker. 1972. Nitrate in deep soil profiles in relation to fertilizer rates and leaching volume. J. Environ. Qual. 1:97-102. Priebe. D.L.. and A.M. Blackmer. 1989. Preferential movement of oxygen-ls-labeled water and nitrogen-15-labeled urea through macropores in a Nicollet soil. J. Environ. Qual. 18:66-72. 154 Quisenberry. V.L.. and R.E. Phillips. 1976. Percolation of surface-applied water in the field. Soil Sci. Soc. Am. J. 40:484-489. Richter. G.. and W.A. Jury. 1986. A microlysimeter field study of solute transport through a structured sandy loam soil. Soil Sci. Soc. Am. J. 50:863-868. Ritchie. J.T.. and J.R. Crum. 1989. Converting soil survey characterization data into IBSNAT crop model input. p.155-167. In J. Bouma and A.K. Brget (ed.) Land Qualities in Space and Time. Wageningen. The Netherlands. Ritchie. J.T.. U. Singh. D. Godwin. and L. Hunt. 1989. A user's guide to CERES Maize.Internationa1 Fertilizer Development Center (IFDC). Muscle Shoals. AL. Robertson. G.P. 1987. Geostatistics in ecology: interpolating with known variance. Ecology 68:744-748. Rossman. E.C.. K. Dysinger. M. Chamberlain. and R.E. Leep. 1990. Michigan corn production: hybrids compared 1990. Extension Bulletin E-431. Cooperative Extension Service. Michigan State University. E. Lansing. MI. Rossman. E.C.. K. Dysinger. M. Chamberlain. and R.E. Leep. 1989. Michigan corn production: hybrids compared 1989. Extension Bulletin E-431. Cooperative Extension Service. Michigan State University. E. Lansing. MI. Saffigna. P.G.. and Keeney. D.R. 1977. Nitrate and chloride in groundwater under irrigated agriculture in central Wisconsin. Groundwater 15:170-177. Saffigna. P.G.. D.R. Keeney. and C.B. Tanner. 1977. Nitrogen. chloride. and water balance with irrigated Russet Burbank potatoes in central Wisconsin. Agron. J. 69:251-257. SAS Institute Inc. 1988. VMS SAS Production Release 5.18. SAS Institute Inc.. Cary. NC. Saxton. K.E.. G.E. Schuman. and R.E. Burwell. 1977. Modelling nitrate movement and dissipation in fertilized soils. Soil Sci. Soc. Am. J. 41:265-271. Schertz. D.L. and D.A. Miller. 1972. Nitrate-N accumulation in the soil profile under alfalfa. Agron. J. 64:660- 664. Smika. D.E.. D.F. Heerman. H.R. Duke. and A.R. Bathchelder. 1977. Nitrate-N percolation through irrigated sandy soil as affected by water management. Agronomy J. 69:623-626. 155 Smika. D.E. and D.G. Watts. 1978. Residual nitrate-N in fine sand as influenced by N fertilizer and water management practices. Soil Sci. Soc. Am. J. 42:923-926. Spalding. R.F.. and N.E. Exner. 1980. Areal. vertical. and temporal differences in ground water chemistry: I. Inorganic constituents. J. Environ. Qual. 9:466-478. Stanford. G. 1973. Rationale for optimum nitrogen fertilization for corn production. J. Environ. Qual. 2:159-166. Steel. R.G.D.. and J.H. Torrie. 1980. Principles and procedures of statistics: a biometrical approach. Second edition. McGraw-Hill. Inc.. New York. NY. Stewart. B.A.. F.G. Viets. Jr.. G.L. Hutchinson. and W.D. Kemper. 1967. Nitrate and other water pollutants under fields and feedlots. Environmental Science and Technology 1:736-739. Timmons. D.R.. And A.S. Dylla. 1981. Nitrogen leaching as influenced by nitrogen management and supplemental irrigation level. J. Environ. Qual. 10:421-426. Trangmar. B.B. 1984. Spatial variability of soil properties in Sitiung. West Sumatra. Indonesia. PhD Thesis. University of Hawaii. Honolulu. HI. Trangmar. B.B.. R.S. Yost. and G. Uehara. 1985. Application of geostatistics to spatial studies of soil properties. Adv. Agron. 38:45-94. Tyler. D.D.. and G.W. Thomas. 1977. Lysimeter measurements of nitrate and chloride losses from soil under conventional and no-tillage corn. J. Environ. Qual. 6:63-66. Tyler. D.D.. and G.W. Thomas. 1981. Chloride movement in undisturbed soil columns. Soil Sci. Soc. Am. J. 45:459- 461. Uehara. G.. B.B. Trangmar. and R.S. Yost. 1984. Spatial variability of soil properties. In D.R. Nielsen and J. Bouma (ed.) Soil Spatial Variability. Proc. Workshop ISSS and SSSA. Las Vegas. NV. 30 Nov.-1 Dec. 1984. PUDOC. Wageningen. Netherlands. Van De Pol. R.M.. P.J. Wierenga. and D.R. Nielsen. 1977. Solute movement in a field soil. Soil Sci. Am. J. 41:10-13. 156 Vauclin. M.. S.R. Viera. G. Vachaud. and D.R. Nielsen. 1983. The use of cokriging with limited field soil observations. Soil Sci. Soc. Am. J. 47:175-184. Viera. S.R.. D.R. Nielsen. and J.W. Bigger. 1981. Spatial variability of field-measured infiltration rate. Soil Sci. Soc. Am. J. 45:1040-1048. Wagenet. R.J. 1984. Measurement and interpretation of spatially variable leaching processes. In D.R. Nielsen and J. Bouma (ed.) Soil Spatial Variability. Proc. Workshop ISSS and SSSA. Las Vegas. NV. 30 Nov.-1 Dec. 1984. PUDOC. Wageningen. Netherlands. Warncke. D.D.. D.R. Christenson. and M.L. Vitosh. 1985. Fertilizer recommendations: vegetable and field crops in Michigan. Extension Bulletin E-550. Cooperative Extension Service. Michigan State University. E. Lansing. MI. Warrick. A.W.. and D.R. Nielsen. 1980. Spatial variability of soil physical properties in the field. p.319-344. In D. Hillel (ed.) Applications of Soil Physics. Academic Press. New York. Webster. R. 1977. Quantitative and numerical methods in soil classification and survey. Clarendon. Oxford. 269 pp. Webster. R. 1985. Quantitative spatial analysis of soil in the field. Advances in Soil Science. vol 3. p. 1-70. Webster. R.. and A.B. Mcbratney. 1989. On the Akaike Information Criterion for choosing models for semivariograms of soil properties. Soil Sci. 40:493- 496. White. J.W.. Jr. 1975. Relative significance of dietary sources of nitrate and nitrite. J. Agric. Food Chem. 23:886-891. (Corrections. J. Agric. Food Chem. 24:202. 1976.) Wild. A. 1972. Nitrate leaching under bare fallow at a site in northern Nigeria. J. Soil Sci. 23:315-324. Wild, A.. and I.A. Babiker. 1976. The asymmetric leaching pattern of nitrate and chloride in a loamy sand under field conditions. J. Soil Sci. 27:460-466. Wilding. L.P.. and L.R. Drees. 1978. Spatial variability: a pedologists viewpoint. 1n Diversity of soils in the tropics. Soil Sci. Soc. Am.. Spec. Pub.. 34:1-12. 157 Wilding. L.P.. and L.R. Drees. 1983. Spatial variability and pedology. In L.P. Wilding. N.E. Smeck, and G.E. Hall (Ed.) Pedogenesis and soil taxonomy. I. Concepts and interactions. Elsevier. New York. Wilding. L.P. 1984. Spatial variability: its documentation. accomodation and implication to soil surveys. 1h D.R. Nielsen and J. Bouma (ed.) Soil Spatial Variability. Workshop ISSS and SSSA. Las Vegas. NV. 30 Nov.-1 Proc. 1984. PUDOC. Wageningen. Netherlands. Dec. WILDsoft Surveying System Software. Version 1.3. Wild Heerbrug Instruments. Inc. Norcross. GA. 1987. Yeh. T.-C.J.. L.W. Gelhar. and P.J. Wierenga. 1986. Observations of spatial variability of soil-water pressure in a field soil. Soil Sci. 142:7-12. Yost. R.S.. G. Uehara. and R.L. Fox. 1982. Geostatistical analysis of soil chemical properties of large land areas. II. Kriging. Soil Sci. Soc. Am. J. 46:1033-1037. Youden. W.J.. and A. Mehlich. 1937. Selection of efficient Contr. Boyce thompson Inst. methods for soil sampling. Plant Res. 9:59-70. APPENDIX I Horizon Depth and Thickness Data 158 003529. ...-.00 8.028 0.0. oo~mE20x new 8.028 ...9 00~0E20x 0.0. 8050.00. 0. .0 080520: 0.0. 2.028 ..0. 8.0200 0.0. 00505200. v.9 OONQEM-flv- CNN 2.028 e0. 8028 0.: 8050.00. 0.00 2.250 o0. 050.500 0 .0 p OONGEU-GX 0. ..N 9.028 0.: OONGEQEX 0.00 2.2.00 0.: 2.2.00 0.0. 1... 00000 .00-0.0- .”an 00.000 3.00. . 02.0; 02.00 $700. «0700. A030 001.: 8.00 0000 02-00. 02-3. 02.2: «...-B. 02.2 02-00 00700— 02.02 007.0 00.0: 02-00. 00700. 003: 003.: 007.0 .Efag'rr c". . 00 00 .. III-ll!!! 5305 00.00 00—00 00— -0v .000 00.00 0040 009-; 3—0v _ 04.0.. 00.00 00700 0070 «00¢ 03-05 007—0 3 p-00 —0-00 «~00 I' I. 0" '1", l'ul? 0500 00.00 0000 00-0N 00.0w 3. — .00 0000 .000 3.00 00-0— 50. -0? 00-00 00-00 00.0w p000 0V0? 0500 .000 S .00 00-; 00-00 .1. ....In 0v0w 00-0w 09.0? m 9. flu-lb!“ “an rlpsrrrt!l tiprtt - ...r'rur :0..- 0.4 ll.n|llll'll.l)ollla I')’. 0.0.0.5000. 05000 20500: 'NE "unis”..fi'gaflu-u ”“U'ullillflttllfnfilu. — 0.00 0.0V w.-.>.x 15.5 mEzam-ooom 2.0028000 03:0; to. :8. 000 ”202 0.00 some-cogs: 800.. ._ 50:00.2 08050.0: 0.00 00. -.00 .000 00.00 00-0 0.. 0.00 0.00. 050.300 ..v. 007.0 .0-.v .700 00-0 .0. 0.0 0.00. 050200 0.... 05700. 00. ...0 .000 00-00 00-0 5.0 0.0... 0.00 050.000 0.0. 05.05 05.00 00-0. 0.-0 . 0.00. 0.00 00350.9. 5.00 0070. . 0. .-00 00.00 000 0.0 0.00. 0.00 00350.0: 5.0. ..0-00. 00700. 00700 00-00 00-0 0.0 0.00 0.00 000050.00. 0.00 007.0 .0-.0 .000 00-0 . 0.00 0.00 050.300 0.0. 0.005. 05700. 007... ...-00 00.0 0.. 0.0.. 0.00 050.000 0.0. 507.0 .00.. 0000 00-00 00-0 0.. 0.00 0.00 03050.9. 0.00 05750. 50. -00 00.00 00-0 0.. 0.0 0.00 050200 0.0. 00.00 05. -00. 00-... .700 00-0 5.0 0.00. 0.00 08050.5. 0.00 0570.. 0.7.0 .000 00-0 ... 0.00. 0.00 w 0502.00 0.0. 00750. 5070. . 0. .-00 00-00 00-0 0.0 0.00. 0.00 1 050.000 0.0. 007.0 .0-.v 3.00 00-0 0.0 0.00 0.00 005050.00. 0..0 00700. 007.0 .000 00-0 0.0 0.00 0.00 050.50 0.5. 00705 05.00 00.00 00-0 0.. 0.0.. 0.00 090050.00. 0.50 00.00 .000 00.00 000 0.. 0.00 0.00 050200 0.5. 00.05 0500 00.00 00.00 00.0 0.. 0.0 0.00 050.000 0.0. 00750 50.00 00-00 00.00 00.0 0.0 0.0... 0.0.. 00350.00. ..00 000.00. 007.5 .500 00-00 00.0 ... 0.00. 0.0.. cow-050.0: 0..0 007.5. 00.00 0000 00-0 . 0.00. 0.0.. .- 330 00-0.0. --..0m--...0m-....0.0m: -..:-- .. 0.... 50,..- .MUM-cxi 0..0-0.0.0....00 ...O0 lemmmwmw.m.zn.m-00.zm...mp- ...-05.523.12.008... :0... 0500020000 5005.. .0. .5. 000 ”0.02 0.00 058.083.. 8000 .. 02.00% 160 0.5.50 :— 05.50 «.0— 00553. 0.3 0055.3. 0.0— 05300 vs— 00555x 0.0m 0.5.50 0.0 0055.8. ..mm 0055.5. 0.00 0055.3. 5w 0055.5. Now 0.52.00 0.0— 05200 0...— 005550. 0.5 0055.00. 0.0— 005550. 0.3 08050.5. Yew 0.5.50 0.0 0.5.50. 0.0— 5.50 m...— 0.5...00 «.m— -... 8:8 .05.... 0..-555.....- 0—N-00— 007: :.mw 000 u .0 0.0N— 00¢— 0—N-ne— nv—é: v— — .mN 00.0 50 0.00— 0.0?— 00— .00 00-0w 0N-0— 0— -0 h .0 0.00 0.0..— 00N- K :00 00-00 00-0 — 0.00 0.0: 0k—-—0 —0.00 00.0w 00-0 —.— 0.00 0.0—.— 00— .00 00-00 00-00 000 N .— 0.0-w 0.00— 05700— 09.0... 00.00 00.00 000 N.— 0.0 0.03. 0239— m— —-—0 .000 00.0 N .0 0.03. 0.0m— 00—-h—— n— —-00 00-0w 00-0 5.0 0.0N— 0.0m— 00—-N0— N0—-00 00-00 00.0w 000 No 000— 0.00— 00—K— — 5.. Ten @500 00-0 50 0.00 0.0m— 0h—-00 00-00 00.8 00-0 — 0.00 0.00— 00—-—0 —0-0v 0v-0w 00.0 N.— 0.0V 0.0m— 00—00 00.00 v0-0a 00-0 0.— 0.0m 0.0m— N0—-NN— NN—-0v 0v-0— 0—0 0.— 0.0 0.0m— 05—00 004.5 VN-0N 00-0 0.0 0.0: 0.00— 00—KN— NN— .v0 v0-00 00.0w 00-0 0.0 0.00— 0.00— 0570.‘ 0v-00 00-0 0.0 0.00— 0.00— 00— -N0— «07—0 —0-00 00-0w 00-0 0.0 0.00 0.00— 00— -00 00-; —v-0N 000— 0—0 —.— 0.00 0.00— 0—N-00— 00—éN— VN—0v 0v.00 00-00 000 N.— 0.0.. 0.00— “EEO-[lourpkcrflrrrulflfivlh'tlilvl’hfvrErlilullrarilllrrllu '0' I”: fEfi'hrtl-r llEp-(IFE 0.. 8 9mm 5 m ..< N > x lif‘lllll’ol’lll 'lll’l'l'l'l‘lll'ellll'l‘lvl [lull-l Olllo‘ nv‘ll lull Olllllll lll'l-I'lrillllll 05.3...505155 2055: - .5. mtg-2.0500 55555555,”...555g.5"£'££-"7.f 055555555555. 0500.850 50.0.. .0. .5. 000 ”0.02 0.50 550-555: :05... .. 50503 Iii. r.‘M-I. .I. I. 161 0055.3. 0.0— 00.... .00.. $8 8.0 ...0 0.00 0.08 0000.00.00. 0.8 .80... 03-0 .00. 0..0 ..0 0.0.. 0.08 000050.00. 0.8 .800. 00. -.0. .0. .8 8-0 0.0 0.8 0.08 0805.00. ..8 00.00. 00.... .08 8-0 0.0 0.0 0.08 . 8005.5. 0.0 00.00 00.0 .08 8-0 ..0 0.0... 0.00. 0000.00.00. ..0. 00.00 00.0.. 00-8 8.0 ..0 0.8. 0.00. 0805.8. 0.0. 00.-..- ..00 00.8 8-0 ..0 0.00. 0.00. 8005.00. 0.8 00.00 00.8 8.0 0.0 0.00 0.00. 050.000 0.0. 0..-.0 00.0 .000 00-0 ...0 0.00 0.00. 9.0.00 0.0 00.8 8.8 8-0 0.0 0.0.. 0.00. 0.00.000 0.0. 00.00 00.00 00.8 8.0 ..0 0.8 0.00. 0.00.000 0.0. 08-08 08-00 00-00 00.8 8-0 0.0 0.0 0.00. 0.00.50 0.0. 0..-00. 00.-.. ..-.0 .000 00.0 0.0 0.0.... 0.00. 08050.00. .8 8.00. 00.0: 03.00 008 8-0 0.0 0.8. 0.00. 0805.00. 0..0 00.8. 87.0 .08 8-0 0.0 0.00. 0.00. 00850.00. ..8 08.00. 00.-.. :8 8-0 0.0 0.00 0.00. 0.00.000 0.0. . 08-00. 00.00 00.8 8-0 0.0 0.00 0.00. 0.00.50 0.0. 00.0.0 00-00 00.8 8-0 ...0 0.8 0.00. 0.00.000 ..0. 00.00 3.00 00.00 8.0 0.0 0.8 0.00. 0.00.000 0.0 0..-8. 8.00 0-8 8-0 0.0 0.0 0.00. 9.0.00 0.0 00.00 008 8-0 0.0 0.00. 0.0... 3.3 .00 8 .9--.00 ..--000...... .0..- - :02... ......- ..-w-....-..l ...:..-0mm ...-En... 00.000 .50... I Esoszcowrmwmmmmsz-Mmmullsgur .5. 022.0500 E'I 05.3.8000 ..0~..0.. .0. .5. 000 ”0.02 0.00 508-0800: 800.. .. 0.8008 162 0000.00.00. QON 00.00 00-00 00-8 8-0 0.0 0.0 0.08 0000.00.00. 0.8 ..000. 00.-.0 .08 8-0 0.0 0.00. 0.000 000000.00. 0.0. 00.-.0 .0-00 00-8 8-0 0.0 0.8.. 0.000 0000.00.00. 0.8 00.00. 00.00 00.00 00-8 8-0 0.0 0.00. 0.000 0000.00.00 0.00 .800 00-00 00-8 8-0 0.0 0.00 0.000 00350.5. mum cop-mo 8.00 00.94. mm... 0.0 0.9... 0.03 _ 000.00 0.0 08.00 0.000 8.0 0.0 0.0.. 0.000 0502.00 0.00 007.0 5.0.. 0..-co 86m omé ...o 0.8 0.0..“ 0000825. Qmw .0079 p N. ....o 3-; :.-mw 8.0 m6 0.0 0.00“ 0000.00.00. 0.8 00.00. 00.00 008 8.0 0.0. 0.00. 0.80 0000.00.00. 0.00 00.00 00.0. 0.-0 0.0 0.8. 0.80 0000.00.00. ...0 00.0.. 0..-.0. .08 8-0 0.0 0.00. 0.80 000000.00. ..8 00.00. 00.00 00-8 80 0.0 0.00 0.80 0.00.000 0.0. 00.0.. 0..-.0 .08 8-8 8-0 0.0 0.00 0.80 000.000 0.0 00.00. 00.8 8-0 0.0 0.0.. 0.80 0.00.000 0.0 00.00. 00.00 00.0. 0.-0 0.0 0.8 0.80 000000.00. 0.0. 00.0.. 00.00 00-8 8-0 0.0 0.0 0.80 0000.00.00. 0.8 00.00. 00.00 00-00 008 80 0.0 0.00. 0.08 000000.00. 0.0 007.0 .000 00-8 8-0 0.0 0.8. 0.08 0.00.000 0.0. 00.00 00.0.. 00.8 8.0. 0.-0 0.0 0.00. 0.08 00000000.. 0.8 00.00 0000 00.8 80 0.0 0.00 0.08 F 00000 00.0 0:0- .- 00 00 0.8 ..0 0 0K I0 > xi. <0<0 00.000 ...00 .22-000000. 00.000 20.0.00: 1 ..0. 022.0008! 0030:0000 00000.. .0. ...0. com .802 0.00 5000-00000... 00000 .. 0.00002 163 9828 oa— osozmo, no OONNEGEX Nap OONwEflmv— VON 2528 8. .58.. «.8 OONGEEMX 0.8 88.8.8 «.8 OONwEw.wv_ .30—- GONG-cw?! VON 9828 3 8.280 3. CONN—tag 0.0m CONN—.33 OS“ 888.3. «.2 8538 3 2828 8: 888.8. 8.2 8.280 m9 888.8. N8 _ 88.8.8 0.2 r 8.58 8.0 mo .53 8.88 EM. A9205. .h“ 873 3.8 8-8 8.8 8.8 8.8 878. 8.8 8-8 876 B8 8.8 873 8.8 8-8 878. 8.8 8-; :..8 8.8 8-8 8-8 878. 8.8 8-8 88: ~28 8.8 8.8. 878. 378 8-8 88: 8.8 8-8 878 5-8 8.8 878 8.8 8-8 88. 878. «2-8 8-8 878. 5-8 8.8 8.8 28.8.83: «:8 8.8 288. 8.8 88. 8.8.8.8. 873 5-8 8-8 8.8 8-8 8-8 8-8 878 8-5 6-8 8-8 83: 3.8 8-8 8.8 r 8 «.8 :m m 8. mxtmo ZONE-w: . I 5.0 0.00— 0.000 No 0.00 0.000 0.0 0.00 0.000 «.0 0.0V 0.000 0.0 0.0m 0.000 v.0 0.0 0.000 0.0 0.03. 0.00m m-0 0.0m- 0.00m No 0.00. 0.00m N0 0.00 0.00m No 0.00 0.00m #0 0.0V 0.00m 0.0 0.0m 0.00m m.0 0.0 0.00m 5.0 0.03 0.00m 0.0 0.0g 0.00“ 0.0 0.00.. 0.00N 0.0 0.00 0.00m 5.0 0.00 0.00m m.0 0.0V 0.00m v.0 0.0m 0.091.. N > x -13 8.82.880 . «coautomov 3.3-=9. .2 :8. 000 “£02 58 sag-gonzo... =88 .. 8:88 164 88.8.3. «.8 8.8. «...-8 8.8 8-.. 8... ...8 8.8 88.8.3. «.8 8.....- 8-8 8.8 8.... 8... ...o ...8 .....8 88.8.3. .8 8.8 8.8 8.8 8-.. o... ...o 8.8 88.8.3. 8. 8.8. 8.8 8-8 8.8 8.. 8... ...8. .....8 98.80 8... 8.8 8.8 8-8 8-.. 8... 0.8.. .....8 8980 «.... 8.8. 8.... 8-8 8-8 8-.. .... ...8. .....8 052:8 v.5 3.93— 579: N300 00.0.. 0700 00-0 0.0 0.00 0.9.0 8.2.8 .8 8«-..« .88. 8.8 8-8 8.. o... ...8 ...9... 88.8.3. 8.8 8.8. 8.8 8-8 8-.. «... ...... 8.8 GONE—33.x 0.0m 02.0: 03-: {-00 00-0w 0N.0 «.0 0.0m 0.9.0 888.3. 8.8 8.8... .....-8 8.8 8... .... o... ...9... 88.8.3. .8 8.8 8-8 8-8 8-.. .... ...8. .....8 88.83.. ...8 8.8 8-8 88 8-.. m... ...8. Q8» 8.2.8 8.... 8.8. 8.8. 8.8 8-8 8-.. 8... ...8. 08» 88.8.3. ...... ..«8. 8.8. «8.8 8-8 8-.. o... ...8 ...88 8.280 8.8. ..«8. 8.8. 8.8 8-8 8-.. m... ...8 8.8 88.3. 8.8 8.8. 8.8 8-8 8-.. .... ...8 ...88 88.8.3. ...... ...«8 8.8 8.8 8.. .... ...8 8.8 88.8.3. ...8 8.8 8.8 88 8-.. a... o... 8.8 88.8.3. «.8 8.8 8-8 8-8 8-8 8-.. e... ...8. .....8 88.8.3. 8.8 88-8. 8.-.... 3.8 88 8-.. m... ...8. .....8 r 8...... .8 8 HA... ...-WW8... -...w....2.m...._-. ...-<- ....N > x -. ...-2.. 8.88 .685. www.mwcoo. mIMmHP-zwmmox I .5. m. x kt... $.58 ...om: 8.8.8.880. mmkwpmgmot Ii 1 .5. 8.2258031 8:32.883 SEQ. .2 ...9 3w 6.0: 8.8 58888.61 88.. .. 588.? 166 25.5.8. 8.8 8.8 8.8 8-.. ....a o8. 0.5.5.8. 8.8 8.8. 8.8 8.8 8.2 2-.. 8.8 o8. . 8.830 58.28.... 85.5.8. 8.8 8.8. 8.8 8.8 8-.. .... ...o... 8.8.. 85.5.8. ....8 8.8 8.8 8.2 2-.. ....- ...8. 8...... 85.5.8. ...8 8.8. 8.8 8-8 8-.. .. 8.8. ......é 0052.53. 0.8 Nmz-vm g-ov wv-ON omé o 0.8 0.03. 85.5.8. «.2 8.8 8-8 8-2 2-.. .... Q8 .88 85.5.8. ..8 ...-8 8.-.... 8-8 8.8 8-.. ...o ...... ...9... 855.8. .8 8.8 8.8 8.8 8-.. w... o8 8.8.. 85.5.8. ..2 82-8 8-8 88 8-8 8-.. o ...o ...9... 85.5.8. 8.2 88.8. 8.8 8-8 8... .... 8...... 8.8.. 25.5.8. 8.8 8.8 8.8 8.8 8-.. ..o- ...8. o8. 0.5...8 8... 8.8... «...-8 8.8 8-.. ..o. ...8. o8. 85.5.8. 8.8 8.8. 8.-.... 8-8 8-.. 8.... ...8 o.8.. 85.5.8. 8.8 8.8 5-:- ..-8 8-.. ....- ...8 98.. 85.5.8. ...8 82.8. 8.8 8-8 8-.. .. ...... Q88 8......8 «.2 8.8. 8.8 8.8 82 2-.. .. 8.8 8.8.. 85.5.8. ...8 8.8 5-8 82 2-.. .... o... 8.8.. I 8.8 8.0 8. r tom.- om WWW-H .... m .8: IN > x ... <55 8:88 ...08 .emfisufiwme-Ffimpzwmgz in .5. 8558001.”- mcoaatooon :o~...o.. .2 ...2 com .202 5.5 588.85.... 88.. ._ 588.... 167 855.8. ...8 8.8. 8.8 8-8 8-.. - ...8. ...... 0.5.80 ...2 288 8-8 8-.. - ...8. ...8. .5... 8.8 28-8. 8.-.... .....8 8.8 8... ...8. ...8. 85.5.8. ...8 87.2 .28 8-8 8-8 8-.. ...8. ...8. 85.5.8. 8.8 82.... ...-8 8-8 8-8 8-.. ...8. ...8. 85.5.8. ...8 ......2 ...-... .....8 8-.. ...8. ...8. 855.8. ...8 8.8. 8.8 8-8 8... ...8. ...8. . 85.5.8. ..8 8.8. «...-8 8-8 8-.. ...8. ...8. a $830 $82.91... 9580 «.2 8.8.. «...8 8.8 8.8 8... ...8 8...... 85.5.8. ...8 .....8 8.8 8-2 2-.. ...8 ...8... 85.5.8. ...8 8.2. 2.8 8-8 8.8 8... ...8 ...8... 85.5.8. 8.8 8.8. .28 8-8 8-8 8-.. ...8 ...8... 85.5.8. ..8 8.8 8.8 8-8 8-.. ...8 ...8.- 85.5.8. ...8 8.8 8-8 8-8 8... ...8 98.. 85.5.8. ..8 8.8... 8.8 8.8 8-8 8... ...8 ...8. 85.5.8. 8.8 ......2 .28 .88 8.8 8-.. ...8. 8.8.. 85.5.8. ...8 8.8. 8.8 8-8 8... ...8 ...8. t 8.58 8.08 0.. 08 «.8 .8 lm:...im.k-. IN > x be... 8.88.. ...8 €888.80. 818m... 2080.. I ll. .5. 8.2.0803: 303.88.. 53.0.. .0. ..8. 00m .802 .80 588.88: 88.. .. 588.... 168 888.3. 8.8 ...-.-.8 ...-8. 88. 8.8 - ...8 88 888.5. «.8 8..-... «...8 8-8 88 - ....m ...8 888.8. 8.8 8.8... «...8. 8.8 8-.. - 8.... 8.8 88.8.8. ...8 8...... .28. 8.8 8-.. - ...8 88 888.9. m8 8..-.8 .88 8.8 8... - ...8 ...8 888...... ...-8 8..-... ...-.8 .88 88. 8.8 - 8.8 8.8 88.8.3. 8.8 8.8.8. 8.8... 8.8 88 - ...8 8.8 88.8.5. ...8 8.8. 8.8 8.8 88 - ...8 8.8 88.8.8. ..8 88.... ...-.8 .88 88 - ...8 8.... 88.8.5. ...8 8.88 8.8 , 8-8 88 - ...8 8.... 8888.5. 88 .88... 8.8 8.8 88 - ...8 8.8 8 $530 55.2.8... 888.3. 8.8 8.88 8-8 8-8 88 - 8.... ...... 888.8. 8.8 ...8-8. 8.8. 8.8 8-8 88 - ...8. ...... 888.8. 8.8 8.... ...8 8.8 8-.. - ...8. ...... 888.3. ..8 8.8.8. «...8. 8-8 88 - ....8. 8...... 888.3. ..8. 8.8.. 8:88 8-8 8... - 88. 8.8.. 888 8... mm. 8 8 918-8.... :8 FL ..fl N > ...-...! kt... 8.8.8.8 ...8 88.82%... 8:53 2080: . .5. 8.2.8000} f 'Eg'l'...’ 80......88 25.0.. .0. ...8 com .202 88 588888: 8...... .. .888 APPENDIX II Data Input for Semivariogram Calculations 169 Appendix 2. DATA INPUT FOR SEMIVARIOGRAM CALCULATIONS Note: '-99.0' used for missing '96 clay" when thickness is "O” 611 Data 23"t2'bata Control- Coordinate (m) CiaL Thickness Clay Thickness Sec. Clay x Y (96) (cm) (96) (cm) (96) 0.0 0.0 22.5 18 6.4 35 12.2 0.0 20.0 24.4 15 6.5 56 11.9 0.0 40.0 30.6 84 -99.0 0 30.6 0.0 60.0 23.6 33 6.9 48 17.9 0.0 80.0 24.5 41 7.0 66 21.3 0.0 100.0 26.6 25 6.9 46 16.8 0.0 120.0 21.6 31 7.0 48 16.0 0.0 140.0 29.6 66 7.0 44 29.6 20.0 0.0 25.6 16 9.5 33 14.6 20.0 20.0 22.5 26 6.4 31 15.4 20.0 40.0 22.6 61 -99.0 0 22.6 20.0 60.0 25.4 33 6.9 56 19.1 20.0 80.0 15.6 21 6.4 119 10.3 20.0 100.0 22.5 28 6.0 48 16.1 20.0 120.0 27.5 28 9.9 31 19.6 20.0 140.0 21.6 76 -99.0 0 21.6 40.0 0.0 26.4 33 5.9 26 19.4 40.0 20.0 17.7 25 7.0 76 124 40.0 40.0 26.4 46 6.5 43 24.6 40.0 60.0 20.6 18 8.4 35 12.8 40.0 80.0 27.5 43 10.3 71 25.1 40.0 100.0 27.5 30 13.4 46 21.9 40.0 120.0 26.4 33 7.9 84 20.1 40.0 140.0 23.4 28 7.4 31 16.4 60.0 0.0 24.5 20 9.9 21 17.0 60.0 20.0 324 36 13.4 36 27.6 60.0 40.0 20.5 35 8.9 18 17.0 60.0 60.0 25.4 36 6.5 41 21.3 60.0 80.0 23.4 21 8.6 20 16.2 60.0 100.0 10.8 76 6.6 45 10.8 60.0 120.0 30.4 71 9.1 51 30.4 60.0 140.0 15.7 21 11.5 26 13.3 80.0 0.0 22.6 101 -99.0 0 22.6 170 Appendix 2. DATA INPUT FOR SEMIVARIOGRAM CALCULATIONS Note: “-99.0" used for missing '96 day" when thickness is "O" 811 Data 2mata Contrai- Coordinate (ml Clay Thickness Clay Thickness Sec. Clay X Y (96) (cm) ' (96) (cm) (96) 80.0 20.0 23.4 15 9.0 43 13.3 80.0 40.0 -99.0 0 16.6 114 16.6 80.0 60.0 26.3 41 10.1 20 23.4 80.0 80.0 224 35 10.1 72 18.7 80.0 100.0 29.4 35 7.1 54 22.7 60.0 120.0 -99.0 0 10.8 43 10.8 80.0 140.0 18.5 28 10.1 66 14.8 100.0 0.0 24.6 20 7.1 91 14.1 100.0 20.0 37.5 46 9.0 25 35.2 100.0 40.0 20.5 1 5 10.1 76 13.2 100.0 60.0 20.4 13 11.5 28 14.3 100.0 80.0 17.5 23 11.0 41 14.0 100.0 100.0 -99.0 0 8.8 16 8.8 100.0 120.0 24.4 61 ~99.0 0 24.4 100.0 140.0 29.3 _ 46 10.4 2 27.8 120.0 0.0 24.4 30 10.0 64 18.6 120.0 20.0 26.4 39 10.9 35 23.0 120.0 40.0 17.5 23 10.1 15 14.6 120.0 60.0 22.4 28 9.9 46 16.9 120.0 80.0 28.4 38 11.0 43 24.2 120.0 100.0 30.4 33 9.0 33 23.1 120.0 120.0 28.2 38 8.5 59 23.5 120.0 140.0 28.8 41 8.1 51 25.1 140.0 0.0 20.5 7 6.5 87 8.5 140.0 20.0 29.4 39 8.6 20 24.8 140.0 40.0 22.5 33 7.5 35 17.4 140.0 60.0 225 35 8.0 13 18.6 140.0 80.0 21.6 30 -99.0 0 21.6 140.0 100.0 13.7 89 11.4 33 13.7 140.0 120.0 -99.0 0 11.7 97 11.7 140.0 140.0 -99.0 0 5.0 99 5.0 160.0 0.0 -99.0 0 5.6 63 5.6 160.0 20.0 21.4 31 7.5 38 16.1 171 Appendix 2. DATA INPUT FOR SEMNARIOGRAM CALCULATIONS Note: "-99.0" used for missing '96 clay" when thickness is "0" 8110313 2812 Data W Coordinate (m) Clay Thickness Ciax Thickness Sec. Clay X Y (95) (cm) (96) (cm) Q) 160.0 40.0 18.6 39 8.4 33 16.4 160.0 60.0 16.6 41 11.5 97 15.7 160.0 80.0 ' 26.4 46 10.5 89 25.1 160.0 100.0 26.6 38 6.1 61 21.7 160.0 120.0 22.5 46 17.5 76 22.1 160.0 140.0 18.5 20 9.6 31 13.2 180.0 0.0 -99.0 0 15.2 33 15.2 180.0 20.0 -99.0 0 10.4 53 10.4 180.0 40.0 -99.0 0 5.0 89 5.0 180.0 60.0 16.2 15 10.5 46 12.2 180.0 80.0 26.0 46 -99.0 0 26.0 180.0 100.0 22.5 23 14.8 28 18.3 180.0 120.0 24.5 25 10.5 21 18.1 180.0 140.0 23.3 41 13.3 25 21.5 200.0 0.0 27.9 41 7.0 69 24.1 200.0 20.0 28.2 79 -99.0 0 28.2 200.0 40.0 27.1 33 8.0 91 20.6 200.0 60.0 26.0 26 10.5 45 18.6 2WD 80.0 29.0 26 17.4 43 23.4 200.0 100.0 . 21.0 18 12.5 48 15.6 200.0 120.0 24.1 38 14.6 13 21.8 200.0 140.0 27.0 28 12.5 46 20.6 220.0 0.0 19.5 18 -99.0 0 19.5 220.0 20.0 -99.0 0 7.5 94 7.5 220.0 40.0 -99.0 0 9.4 77 9.4 220.0 60.0 28.2 23 8.9 61 17.8 220.0 80.0 28.0 33 10.5 71 22.1 220.0 100.0 25.1 38 8.5 58 21.1 220.0 120.0 30.0 41 -99.0 0 30.0 220.0 140.0 26.9 30 10.4 46 20.3 240.0 0.0 29.0 43 7.0 28 25.9 240.0 20.0 19.2 16 12.8 15 16.1 240.0 40.0 -99.0 0 5.0 21 5.0 Appendix 2. DATA INPUT FOR SEMIVARIOGRAM CALCULATIONS 172 Note: "-99. 0" used for missing "96 clay“ when thickness is "O" Coordinate (m) X Y 240.0 60.0 240.0 80.0 240.0 100.0 240.0 120.0 240.0 140. 0 260.0 0. 0 260.0 20. 0 260.0 40.0 260.0 60.0 260.0 80.0 260.0 100.0 260.0 120.0 260.0 140.0 280.0 0.0 280.0 20.0 280.0 40.0 280.0 60.0 280.0 80.0 280. 0 100.0 280. 0 120.0 280. 0 140.0 300.0 0.0 300.0. 20.0 300.0 40.0 300.0 60.0 300.0 80.0 300.0 100.0 300.0 120.0 300. 0 140.0 320. 0 0.0 320.0 20.0 320.0 40.0 320.0 60.0 Bti Data 237 Data Control- Clay Thickness Clay Thickness Sec. Clay (96) (cm) (96) (cm) (96) 36.1 43 12.9 20 329 24.3 58 -99.0 o 24.3 26.1 48 11.0 46 25.5 18.3 38 -99.0 o 18.3 33.0 31 9.0 99 23. 9 20.8 20 -99.0 o 20. 8 25.5 25 11.4 48 19.3 28.4 28 10.3 13 22.7 22.4 23 11.4 33 16.5 20.6 45 10.0 N 41 19.5 24.4 20 7.6 125 14.3 99.0 o 9.8 84 9.8 28.4 26 7.1 81 18.2 21.6 56 8.6 68 21.6 24.6 35 9.4 23 20.0 25.4 23 ' 8.0 35 16.0 99.0 o 9.5 125 9.5 29.4 56 7.1 56 29.4 20 5 43 10.7 46 19.1 27. 4 56 11.8 53 27.4 27.4 31 10.3 25 20.9 21.1 48 23.7 38 21.2 -999 o 8.8 18 8.8 25.4 33 10.7 23 20.4 20.6 41 10.2 91 18 7 -99.0 o 8. 3 23 8. 3 -99.0 o 12.0 26 120 20.6 74 99.0 0 20-6 29.5 41 11.2 25 26.2 35.3 46 10.4 25 33.3 23.9 25 1o. 7 13 19.4 28.7 61 5. 4 41 28.7 22.5 15 12. 7 112 15.6 173 Appendix 2. DATA INPUT FOR SEMIVARIOGRAM CALCULATIONS Note: "99.0" used for missing "91. Iciay“ when thickness is "o" 811 Data 2mm c_on__troi- Coordinate (m) Clay Thickness Clay Thickness Sec. Cl____1y X Y (‘36) (0M) (96) (0m) (96__)__ 320.0 80. o 225 31 122 81 18.6 320.0 100. 0 19.5 28 8.3 89 14.6 320.0 120.0 30.4 38 10.8 ' 8 27.0 320.0 140.0 25.6 33 11.2 18 20.7 340.0 0.0 32.7 38 6.8 49 26.5 340. 0 20. 0 32.0 41 9.7 69 28.0 340. 0 40. 0 25.4 39 8.8 55 21.7 340.0 60.0 -99.0 0 6.4 92 6.4 340. 0 80.0 28.7 21 9.2 73 17.4 340. 0 100.0 15.5 15 12.2 59 13.2 340.0 120.0 -99. 0 0 8.8 26 8.8 340.0 140.0 28. 9 26 10.2 35 18.9 , 360.0 0.0 28.8 25 17.6 18 24.1 360.0 20.0 26.6 36 6.9 36 23.2 360.0 40.0 26.7 33 10.4 1 24.3 360.0 60.0 27.2 36 7.0 61 21.5 360.0 80.0 28.7 33 9.0 46 22.0 360.0 100.0 28.5 30 7.5 18 20.6 360.0 120.0 30.2 23 8. 0 23 19.1 360.0 140.0 -99.0 0 ‘ 10. 2 97 10.2 380.0 0.0 30.4 31 13.8 15 25.0 380.0 20.0 28.6 30 12.4 26 22.1 360.0 40.0 30.2 33 7.0 99 22.3 380.0 60.0 28.9 20 8. 4 38 16.6 360.0 80.0 27.3 30 15.4 28 22.5 380. 0 100.0 31.8 46 10.8 16 30.1 380. 0 120.0 225 41 11.9 17 20.6 380.0 140.0 33.3 40 9.4 46 28.5 400.0 0.0 28. 6 30 10.4 56 21.3 400.0 20.0 27.1 34 9.4 43 21.4 400.0 40.0 30.4 30 7.9 49 21.4 400.0 60.0 33.3 31 8.9 23 24.0 400.0 80.0 28. 8 36 ~99!) '0 28.8 174 ' Appendix 2 DATA iNPUT FOR SEMIVARIOGRAM CALCULATIONS Note: “99.0” used for missing “96 clay' when thickness is "0" 811 Data 285—Data W Coordinate (m) Clay Thickness ‘ Clay Thickness Sec. Clay X Y 1%) (cm) 06) (cm) (96) 400.0 100.0 28.8 33 9.9 20 224 400.0 120.0 28.7 25 8.9 56 18.8 400.0 140.0 34.7 46 8.4 “ 32.6 420.0 0.0 33.3 33 9.0 46 25.0 420.0 20.0 23.7 21 11.1 73 16.4 420.0 40.0 320 51 9.4 51 32.0 420.0 60.0 27.3 46 9.8 26 25.9 420.0 ‘80.0 31.6 54 6.4 51 31.6 420.0 100.0 17.9 49 10.7 63 17.8 420.0 120.0 27.1 31 10.8 33 20.9 420.0 140.0 27.1 31 7.8 142 19.8 440.0 0.0 23.8 31 11.2 15 19.7 440.0 20.0 34.1 33 15.2 21 27.7 440.0 40.0 25.1 28 -99.0 o 25.1 440.0 60.0 28.8 20 11.2 31 18.2 440.0 80.0 28.7 28 10.3 46 20.6 440.0 100.0 32.3 35 11.3 28 26.0 440.0 120.0 27.3 28 11.9 40 20.5 440.0 140.0 32.1 38 11.9 77 27.3 LYSIMETER CLUSTER 1 420.0 20.5 25.6 45 7.8 54 23.8 420.0 21.0 21.5 61 -99.0 0 21.5 420.0 22.0 26.5 61 10.2 46 26.5 420.0 24.0 24.5 61 9.3 30 24.5 420.0 28.0 32.5 46 9.6 23 30.7 420.0 36.0 25.4 41 8.7 27 22.4 404.0 20.0 24.5 38 10.2 28 21.1 4120 20.0 28.5 59 14.6 38 28.5 418.0 20.0 25.4 43 14.6 40 23.9 419.0 20.0 28.5 61 10.1 20 28-5 419.5 20.0 21.6 33 5.8 33 ~ 16.2 175 Appendix 2 DATA lNPUT FOR SEMIVARIOGRAM CALCULATIONS Note: '-99.0" used for missig "96 clay“ when thickness is "0" ' _ Btt Data 2812 Data ControT- Coordinate (m) Clay Thickness Clay Thickness Sec. Clay _X Y fi (96) (cm) 1%) (cm) (96) LYSIMEI'ER CLUSTER 2 174.0 133.0 31.0 46 15.2 23 29.7 173.5 133.0 29.4 49 8.0 30 29.0 173.0 133.0 26.5 44 13.9 g 53 25.0 172.0 1 33.0 26.6 45 15.5 36 25.5 170.0 133.0 24.6 38 13.5 61 21.9 166.0 133.0 26.5 68 18.5 18 26.5 158.0 133.0 16.6 48 12.5 130 16.4 174.0 132.5 32.6 39 10.6 35 27.8 174.0 132.0 22.9 36 11.5 53 19.7 174.0 131.0 26.4 46 9.6 26 25.1 174.0 129.0 24.4 39 ' 9.0 17 21.0 174.0 125.0 23.6 79 -99.0 0 23.6 174.0 117.0 17.7 13 22.9 56 21.5 LYSIMETER CLUSTER 3 32.0 43.0 27.1 36 9.9 48 22.3 31.5 43.0 25.5 36 9.5 43 21.0 31.0 43.0 26.5 38 12.4 36 23.1 30.0 43.0 28.4 36 9.1 48 23.0 24.0 43.0 28.2 84 -99.0 0 28.2 32.0 43.5 25.4 36 11.5 30 21.5 32.0 44.0 29.4 41 8.0 27 25.5 32.0 45.0 23.4 56 8.5 31 23.4 32.0 47.0 27 .3 56 7.5 26 27.3 32.0 51.0 22.5 49 8.5 43 22.2 32.0 59.0 27.4 35 8.1 38 21-6 APPENDIX III Graphs of Lysimeter Nitrate Concentrations. 1987-89 [WHY .2 (””0 291881318 32‘. 88 110 113 118 125 134 140 201 208 215 242 249 259 274 281 288 295 301 309 316 [JATE (yr/mo/day) 870702 870710 870717 870724 870731 870814 870821 870828 670904 870911 870918 870925 871009 871020 871023 871028 871104 871113 671119 880119 880126 880202 880209 880216 880223 880301 880308 880318 880326 880402 880409 880416 880423 880429 860507 880514 176 [MVY 516 529 575 607 617 651 653 693 701 707 713 728 [MNTE (yr/mo/day) 880521 880528 880604 880611 880621 880629 880706 880713 880720 880727 880816 880917 880925 881001 881008 881015 881022 881029 881105 881114 881121 881130 881213 881220 890128 890204 890301 890311 890324 890414 890416 890526 690603 890609 890615 890630 "03 (new 1103 (mm 177 Lysimeter 1.1 80 7O 80 - SC " I ‘1 .II I 40 H II . 11,1 . m. D........ " '1‘. 11,‘ 1 I 11 i , 7;. I" ‘ I. .II m-IOOMOOO 'l;; t J 00.0-- I ' f, .1 11,1, ,11 1111 i. . ,. 1111‘ :1 5111: 1111' i 1' 1!. I ' . .' 1y 1o-..m.....'r 1111111 ....... 111111,: . 1111 1;“. 1‘- II: 1‘ $1111 1,1111: iiiilifi' I 11,411" 1 I .I 11:1. 21111111 11 - 1111 i 1” . 1.1 .‘mr‘ I I.I . ,1 1. 131111 11': .' i if H i! 1.11-,1111. 1111:: 131.1, 11 ' oJ i 1 ‘1" i . i . . O 100 200 3:30 400 500 600 Days (Day 0:901:12. 1957) 910: 1'6: 1 5' ..mo opteelDecBSJMnW—pm I Lysimeter 1.2 60 1 , 1 70 l 1; ll 1.;- I. f. ‘3: 80 :1 ..;nj ll',. M, ii, 111! l $110.00“... I! ’0‘ to. '11“, 11.21 1“ ,1 11,11, 11.11 iJ'..1 1.;113. 11":1 11.11 1 1511;. __ - I 2 40- 1111 1'11" 1,13, 1:1, f1 1 . 1111 1411 1'11 .311 ‘1 it .' .11 JI. ‘ ‘.‘ 119'1 111,1 :4), ,,,1,. . . .1 SID-”mm” ,1 1111111 111.. w. I." ~' II." 1151 1411'. 1.41. ,r,.1' ” .."..‘I' i i. , Hi I 11,1 111 16141411,...11" r- 11811 I r1 1,111 30...»...m1111 11,1111. . i',‘iI ,‘11.‘,ii 1 1 1L-1 :1: .....1.1111 1,11 1121117111‘1' 1:1,. 3:11.! 11,011}? 11.1'511‘11. Li H 1L.1 11.1 ”J,” l1 1.11 11211 413le.”‘11 (ii II,uLa'I 11.111} I 1111 11111.1 1.1. .. ‘ "1111111121 I r. 1:111:11“ ...... W” I I I' 10"“..."111119111’1. 1,1,: Him .1; 1101.! i i I?" I 1' 11111 111,1 1,1. gm, .11.. ,1}:11;.411,1 1L1 1111 11:1 um 1:31. .,;~'.11't,. .4 f, H 1,“! C'I II ”1“ I 01 . I ' O 100 200 3C!) 400 500 800 ’00 "03 (ppm) "03 (mm Lysimeter 1.3 178 10“ I mum-"1:, - 141. . 11 11 1 .......... H311 .';111: 11-411 1' , 11111, 111'." 11111141 11311 .......... 11,11 1.11 ,j 1,141; 1,’ 111,111 11;:11,:;11;. . 141111,- 1 I1 11:,111'1111 44144.1: . 1113-1 111 11411 1 .151. 114.1: ......... 1'11 11'11141I'1. 41141.4 1:1111' .- 11: 41 11'111I11'111a1 . 1414.713 11411 141141 411541: 41 '1‘ 1111'; .......... 114111411141 .1‘1~ I1. ....... ............... 1::3':1 1' 1141111111 4.4413141 . . .5111. 114111411414 ,1. 1' ~414114 '14: 1141 1.41114: I41141, 4.414;,1,H .141111 1 O 100 200 300 480 Days (Day 0 = JUfy 2. 1 937) 11,1 ”3.1 70 HZIII 14:1 11'}. ”Li a 66 114311: , H41! 1,,1 11:41 14' 50 11,. 1., ”1‘,“ Hi ”41 I III,“ 11... ' 1111 ao-..........”;” I1 11".« ”III I ”(44 l ”#1 HM 3°""”""°11r,r1 41 111411' 111?! ,1 11.~ 1141 41} 4114.1 20“" ------ 11. s ,11 1' ,111,11». 1114. $1 1 ,1 c ..,11','1 11.1 ,11114111. ..1..11.,. , 1111.11 111 1111 ..... .111I11r.-1. 141114111 , 1°"”°"""11...1 I11. 11,;1 :';111 """" 1”” ; .11 411111.“ '1 ‘111’111 ; 11.: .'11 111.1141 111 . 1.4.11: 1111. 11,1 ,11111 114.11 1 111.11 11111 11411. 4 ‘ OI 0 100 200 300 400 500 600 700 Day: {Day o = July 2, 1 9371 [Sept 57 {Doc 5711111.: 85 Farm 5515981511311» 5941.” '53] "03 (ppm) "03 (mm 179 Lysimeter 2.1 BC ' — ‘ ? .1;; J H. 1‘5; 1 -I II I. I ”q -..oo..oo. .L '. . l H 'w ' I II . '.- I 40 NJ] _ I: '. IIiII If} I I ‘3 ”:3" I" I um .' 30. .......... ”3 i 1 '5; ' .’1 . I”, ”fin! Ix. I ”j!!! . j, j Ii" I P _v.......O. ooooooooooooooooo _‘ I ........ ‘ fl_ 0 ‘0 in. ' .3 t' g: - l?‘ b " ~ ‘4. . V,“ 3) . I .. I .'t t ’{ VI" HI"; . .1 If ' III I. I.” . may. mum I I , III "I .1: .9 g H 1o. ooooooooo 0";' ‘1 a... 'M: ....... “‘ :‘$'I 3'? ri"l.: ’I5.'- 0 u 0 IL}: mu. I I426 .75 ffizg‘vfhfingnifi I - r4 h ‘ we]: ‘5 v 4' II It I “II ”’I’I' “a, MW?“ IIIIIIF‘. I I IN! WIN IIII; w“ zrt'wufl ‘ I 4N3 1 I 0 I II I" I.) I1.: ‘. .1, 100 {DRyO=Jury2.1987) Days 630 760 w- 70 m. ......... I a" °°°°°°°°° , .9 I l I I- l i do. .......... ' "1‘ ‘ 1 : a w“ """""" y l I! mun-uncu.- {u ...................... on . on oooooooooooooooooooooo I 10- ooooooooo I ‘ ooooooooooooooooooooo u . on”- c 200 3x '400 500 600 700 100 Days {pay = My 2. 19577. {Sonar:zecwjuyaspunaspmaspsEW—W ar me N03 (0me 179A Lysimeter 2.4 , I 7 ac " l ' H H II I m‘Iu.““.." II I" II ‘I I; w H ‘ I“ "H oooooooo o . ...-ouooooct II H‘II'_ I IEIVI II I ‘- 1. IIH'I Iil‘” I V If EL “ I 'I' , ‘l x H l j! .I I 1.1;. I‘ . It‘lIlfll’i. I H I {TIMIII' " I";ll IITII'W ii? “I' II " ‘1 I '3 II I i. ‘I' ”“382“... Ii gal“. Ii Jim I ‘I II I' ‘I :fIirI: ‘33" I-fg'II‘IIf I'M; :I m-I.”.....- '3 no. l.o :1 1:1" :1 :' ooooooo ‘11.”3" f‘é} "i.‘:::';."’I.r"l o II I I II ijgII :I:‘ IfflIfIfl‘ III"! I ' I'D“ j": fIf'fl II II IN MI: MP“? I'.'.,I'I|‘JI,A.IIIL:. = {JI’ fII'QI‘jI 101vooooooooo "- ' ' I‘ll 7" :‘f ....... wil“ "34"4.l"‘ .f.” . . o..- IIIII Iq‘IIqI 'thiI." “§3tI§.l‘»f?€;i I.I{II .'I I ‘ I III II If _ Hi1" H‘MI‘II. "' "Ink "It’ll”. It??? I‘ll!" ISL]! ‘ i III‘II I'Il M3: rim" I} gs“ I'II" :I ‘ o II: II M (I . ‘ O 100 200 ‘ 300 400 Days 500 (Day 0: My 2 1937) “03 (0me "03 (ppm) 180 Lysimeter 3.1 1G o-I Ir?!" IzIIa ' If! i _ o 1002602360'460 500600700 Days anO=Juzy21953 [Ea-p: 57101: 67 [Mar 5'." pane 5559p: 85li 35 1M8 WP.” 39'! Ag Lysimeter 3.2 ' HI! ..I 6 ‘160r2EX3'300 400 soo'eéo :'oo Day: (Dayo=JuIya1QSZ) _.ep: IA are I .a! Jurafmeepocsswm N03 (ppm) N03 (mm 181 Lysimeter 3.3 o 100 200 ' 300 400 500 ' 600 700 Days oayouuqrzwen Sep: 5: LI' cc 5': Ina: &: 11:30 55 Egg 86 100: 53-1]?ng Lysimeter 3.4 ”03 tannin ~03 (pprrfl 182 Lysimeter 4.1 8G 70 EC . 50 40 3G 20 10- -------- l 04—, I , , l ...—‘11: L-‘ o 100 200 300 400 500 am 700 Days {Dayo=Jury2.198?} EFF: 3’ [50601 Inga P“ 53% #85550: 85 [Mar BE Um BE 80 7c ............. 60- n I I III II] 40_ __________ ”'4' H 1' II; I 41. HI. I {fir II II: I, aw ..II 30 II‘I'I, """"""""""" 3:,II ' In. It Ii 12’“ III I H“! l" Il+lj¢llllm 30- --------- HI . I".-........--.......-. gIIrIIfII-IIIIIIJ III I“ Himaléfll”!!! H] I, III It‘flfillllflfl I 1 _ .......... III‘I . ............. IIIiulrIlL‘IIHHII l 7 0 [III I' I. ‘3';II$I;IIIIII;‘I I I III Ir LII "flIHJ' "I“: t I J II; II Ij, lhIHIILHHHHf‘I II I I O I , . O 100 200 300 400 500 I300 700 Day: (Day 0 = July 2. 195?) WIWEFQ gamma“; Baum] ., N03 02me ”03 (ppno 183 Lysimeter 4.3 80- 70-...... - 80'" .... 50. ......... 40 ..... 30.1 .......... .... zo-mu... .............. - 10- ................................ I! I I o-J ' O 100 200 300 400 530 600 700 Days (Dayo = Ju!y 2.1982} op: lwo’!a~ fiufl-msspxaaluxagpww ’ LySImeter 4.4 BC- 70- 60 50 .0 fl I 33 I all“! I’lflfifrtllll 39 IJSI ll{mmz III 13:. II‘IIII I 3mm IIII II . l ,‘IIIIJUI'IIJIIIHII‘IH ll" :1 1m 1.11m IIIIII I; 10 171 IngllIEIHIfl ..... ”min IIIIrIIII J 331;" fmmfmI .. O . O 1m 200 300 Days (Dag/O: ..IIIIy a 193:": [S era! 3? [5°C 51 Inc: Epwmg 35.195; 55 [yarn—‘53], A n _ "03 (1113111 ~03 1, ppnj 184 Lysimeter 5.1 70 I m . oooooooooooooooo 1 .‘I md-gcouuo-o H " ll l ‘7. : 11+ II I: 1 117 ' ‘ *- ... 17| I71 77 1‘ 717 I ”C'OOOOIO... it“). ' " .0... I. ..... ”kg"? | ill” a“ 11:11 11111 1 I. I“ 1 . 1 IIITII II . I 7: 1 1111 1 I 11311 1:) I 71,717. 113,173,” I . ‘1'" II I; ‘ . a. \|:..‘ DY | m“......-OD l"‘b' iv" ' I? 4‘ ‘ O ...... ‘I'i‘71’fil‘ml?.£..’............ I.' 11.311111 1' 11:11.12 rm 1' .511" ‘171111~.1:1 11 I '1 1 1 T“ {1 1'1 11111 1.311 I 11,11,17 1133.111 1:111} 1211111511 I .11. 1111111711“ 11111 ”I’mwr'flx. 1.17111 1511 111:1 1» 201-mm»- 113111 1171 1. 31.713 ........ . ,1I 77‘1” 1‘71“} 1":11‘1 ,11111 1 11.111 11:11 1 9111 3117.11: 17,118} 141111 I I, 111111 11.11 13111: 11111 ,‘11111‘17 i ll""1=1l:4!|:511'1511. 11111111 10-..... ..... "W '1“ " “”1?" ,,,,,,, 11111111115131 i..111131'11‘: ...... 1,: "11:111. 1111 1.71 I “11:11: 3:711:11: {3111137.L11131I‘11fl 17.7.1131?" 1' 11.711 1311 1- 1.1:“ 111'1111',,117‘WII’1-17I.1 1 1117111111311 6 ‘ , I ”if” h" I M? 1:“ It‘I'kI ~17,1.1 I1'1,1 .1'. 17111131111 0 1 1111 ‘111‘ - I . V O 100 400- ...oo= 400 500 600 700 Days (Day 0 = Jury 2. 193.71 13.5: 3710.: 87 {Mar as 1.th as pee 831m.- 6;. Fun: 8:1 ..1 Lysimeter 5.2 1 I ' l 50 II . H 1 11:1. 40 “J” ‘ ‘1‘,“ I I 11:11 I I II t 1 111,11 I . . _1‘11 m-I.........III;1 I I) ‘1‘", " ‘ I 1 11.1 1111 71711,: 11 I II-,‘I 1:11 11.11" I 7 I: ' 1 , 1 m-Iooocuoo- ”1“ ‘f' II“. 0...... . o. 0.... o... ‘1' ...-o. oo- ‘ 0‘ 11111111 2211'. _7 I 1‘- I I I I 1_‘H I ‘l " ' . . I ' : 11.111 ”‘1‘“ l" 1",7'7. I7 ‘I ‘1':“:.. ' !' 71:' I 7 ‘1 ' Ii" ‘ |". ~‘H1‘l”;.-:lil,h' . ‘£' . I ' 0..... “| I ‘I' I 0.0 “...-ocoooooooooooo- 110.000.... ....... 7 7 V 7 10 11:11 111111 .71 ‘17 ,771 11“ 7,, 13:1 I7‘I I 1,. 17:11 II7I 1 11,11 1111 .1111 . i :1 7117.11 111 1:11.111 I II'1 1‘1 1. "I:‘I~' “71”,, 1,717., I 11.11 111 IIII O 20° ' 330 .100 530 33:: 700 38115 {DB'O=JUI}'2.':957_I Wop: I .1.: I a.- =. 1.11%.: as .1.: as war E? pug] 1"” "03 (1313110 "03 (1:1pr 185 Lysimeter 5.3 'l '71: ':'|'1 .1‘131 '7141‘7 ".74‘ I 7' {1191: -':'.I ‘ 11; 1 11' 0| I 11" ‘f’ffi117gs «171' nuance-0...... ' ' ....... 77‘:' 37:7" .. ' 71.7 77775,,1 77 7. "#111711“ 12.4, ’1,"IU'~1""""’1|5|,"'I .'7'“. l I747 ””711'O-f 6‘ I 1" 1 13337, ‘1'. 71'7”: 1:?7' 1111:1171 l l 7 1111-7 .: '1 17'. 771:11' 11:11 '11711 . l’ 1171, 111:1},3111'finl7111111‘111 1,..1 : ‘ ' ' r-i 13"; 1 1 - 1,7117, ,7 1:11: 7771,11 :1:11-'If1 I 1.1 I 1,111.51 7,;1r,. 11711211311111, . 1 t 111 I . , 0...... .... g ‘ 11111 I! 1111141,: 1177117 111111211 71311111 1‘ ,1 1171". 7 , 71,. 1,,27147111'11 :11 I 17:11 1 1 11 1 171"." 1 1,17. 7 7111,77117,7,171I1 1:1,17111 1 1,71 1,111 1' 1 206 .333 400 500 600 700 Days {Day Jug"?- .. 1987) 1’50: 57 7130:. 57 [Mar :15 Lane 31': 1139p: 83 7109: 3'5 War 5:1 Fun: E ,1 Lysimeter 5.4 20- ......n. o 100 Days (may = .:uIy 2. 1 9571 [Sept 8715:: 57 71Ma'EFJJune 86 7’5.“ safe: 85 War 65 Emmi] N08 (0pr N03 (norm 186 Lysimeter 6.1 80 70 BO 50 40 E If . l I. 33‘ """""" II I 'II II I I. I 1“]; - ‘ II I, I - It . IEI'. P , . I f." ‘ m- ---------- H-00 E” -- I‘ll!“ ------- . I ”I.” I H I'I'Hr‘ lr’l‘I‘III 2;!" 'l 1“ 6 II III}. In!“ IIIITIIIII . .’lm I‘I 10' .......... IImIflls'IgC .lf'FJ' ....... Ir‘szIICHIII IIII W" I II IIII‘EI‘I: :‘Ififlé II'LII s‘II-nm IIII I I; I I I I, I; ’ ‘ ' IIIII IIIr‘mI III Ira“. II I I II II III‘II .IIIIII IIII mt", oi ‘ . I . a O 100 200 300 AGO 500 6'00 700 Day: (Day 0 = Juty 2. 193?) - Bones .0:. 631W Ly imeter 6 2 80 70 I 60 H II I 50 II. ., l|.l ,}I ”I. J.’ (H. L-r' m. .......... 't‘ ‘I, I VI" ”‘1' 171:??? up! 3,93 I . 30"“"""' H I "I“ IIII1 HI 1‘ ,‘IIIer 4”!" III :‘I I lllll' 30...........III,..1, IIEIIII; ....... III I II I #I I III II . I II I; .f’nfifiu I,IIIII IIII [‘22 . I. II ‘ II I: ; L1 H'fllll ........ Hill I IL; I! H - 1°" °°°°°° III :';”! szm‘ IIII IEQI; "L II II I ”'3’? ”:3 HI” IIIIJ-'II, If; I.I III J II I' rI;I I ”III IIII IRE; I?” g I II o ‘ - . _ V v, fi— r 1 o 100 200 :30 A00 500 600 700 Days (38]! C = Jay 2. 19375 [Sept 51' [Doc 51' Wm 85 L-Lneffjgopc 65 Pa: Elgar Hi 'mc fi .1 "03 (WW N03 (ppm) 187 Lysimeter 6.3 80 7C 65 so I 40 , ‘ wq .......... ‘ 204 ...................................... 1 1 .- 10-1.......... ....... . - . a O- . . _ v 1 O 100 200 300 400 :100 600 700 Days (Day 0 = Jury 2. 192?} {52:11 ’67 [Doc 67 [Mar 5 p111» 555.“ 6833:: 65. 1E3: 613 13ij I 5* Lys1mc1e16.4 l 1 1' 1. 11 11 111 I l' 33 11 11 11 H 111 , 1 m ”1 , ""1 ‘ 11. 1" 1 1, . 111 111,1, 11‘ 1 1 1m 1141 40 11 1 11111 (1' 1 1 (nu-'11. 11’ 11 1’ 1115111, 33“ """"" 111”( ,- "$111,", 111 1‘1 1 ,. 1 11131:}, 11,1, 111, 1,11 1,11 .;. 1'1 , (,1; 1 (1 m...........1(r1'1.1—;‘1 ....1,11 ....... , 11:11, ...... ,3!” 1.. . ($1,111 1“" 11, 1f11.'111111,1:,1,"(,11111 1 $113.” 1 (111,111 11', 111% 131111 1111111151111 1‘1,“ :1 1111 11111,,1 1°_,mmm.11,111111 1111,11 ....... 11‘1111.'1 ”L1 " ... (1112,1151 111:1 113111111 ,1,:§11‘,111111-,111,. 1 1 11,311‘11 111.111 11 1,111 ,1131111 1: .1 "i 111,; O“ 1111 11111111111111 ¢,.1;11,(1111 ,_,1,-, 1 11,111,, ' O 100 200 ’3“) 400 500 6a) 700 Day: (Day O=Ju1y-‘=‘ 1937) [Sept 57 Po: 6? “dub: 7Uune 56 Egg: 8.515151: 6: [name MP 3E! N03 (ppm) N03 (ppm) 188 Lysimeter 7.1 70 h _l I A r r fi- 1éoj200'360‘460‘soo'eéoT7oo Days (Day 0 == Jury 2 198?) gap: 5: [50: in" [Mar 35:53” 8515.; wEosiéjMu 3. En Sb 1’ 1G -1 n] Y 1 v 6 wéojzéofaoo '450 500 800 700 Days (DayO:Juty2.1932-') [gen 51' [Dec 57 [Us 53 LEWIS-Egg: 85 Ba: 65 1M3! a Flu-ac 5P 1 N03 (Wm) N03 (ppm) 189 Lysimeter 7.3 -1. _ f fiffififio ‘ séEreéo ‘ 765 Day: {Dwo=Juty2.198:-') Lysimeter 7.4 7O h ‘ Y O4 zéofifiaoiaéosoo eéo‘7éo' Days (Dayo=.Ju!y2.195?) ' "use '3 09:66 ac "‘iifilzurflafiwmwmmmfl‘5