This file was created by scanning the printed publication. Errors identified by the software have been corrected; however, some errors may remain. Consideration of Spatial Variability in the Management of Non-Point Source Pollution to Groundwater W. Woldtl, F. Goderya2, M. Dahab3, and I. ~ogard? Abstract. - Geostatistical simulation and unsaturated zone modeling are combined to evaluate the impact of spatial variability of selected parameters on groundwater nitrate contamination from agricultural production. Three management scenarios involving spatially variable application were investigated with consideration of spatial variability in residual soil nitrate, yield, and hydraulic conductivity. The process is applied to three conceivable situations, differing in the extent of spatial variability. One decision input, the fertilizer amount, is distinct for different scenarios as well as spatial location for one scenario. Modeling results indicate that variable application of nitrogen, based on spatially variable parameters, not only reduced the over-application of nitrogen, but also reduced the overall groundwater contamination potential. The modeling results indicate that spatial management techniques hold promise for maintaining production while simultaneously protecting the environment. INTRODUCTION The natural and management induced variabilities of field soils and hydrologic formations are extensively acknowledged as dominant factors influencing fluid and mass transport through the subsurface zone. The variations of soil and crop properties have led to attempts to understand those variations and to manage production accordingly. The management of production according to localized conditions is variously known as spatially variable, site specific, soil specific, precision, or prescription production. A consensus does not appear to have been reached within the scientific community on the proper term or definition. However, the methodology presented in this research complements production and environmental concerns according to localized conditions and will be referred to as spatial management technology. Assistant Professor, Depamnent of Biological Systems Engineering. Universityof Nebraska - Lincoln, Lincoln NE. Graduate Research Assistant, D e p a m n t of Civil Engineering, University of Nebraska-Lincoln, Lincoln, NE. 3Professor, Department of Civil Engineering, University of Nebraska-Lincoln, NE. The key factor driving spatially-variable control is field variability. Spatial variability may be random or auto-correlated, may be long distance or short distance, and may be small or large. However, all affect the feasibility of spatially-variable control and the design of a particular system to achieve such control. In addition, variability may result from both natural processes and management practices. If there were no variability, production through traditional practice with proper adjustment for field conditioils would be adequate. For all its importance, spatial variability and it's impact on crop production and the environment has not been studied to a large extent. Furthermore, the procedures for developing an understanding of the variability in a particular field and effective utilization of that information to reduce NPS groundwater pollution, while at the same time maintaining production are still not clearly defined. The purpose of this research is to develop a framework for exploring the effect of spatial variability related to different field scale parameters in managing crop production and minimizing groundwater contamination. METHODOLOGY The methodology outlined in this research incorporates spatial variability of various parameters on a field scale in estimating nitrate contamination to groundwater. Unsaturated zone modeling in combination with conventional measurements and simulation is utilized to quantify the effect of field spatial variability on contaminant loading for quantifying environmental benefits of spatial management technology. The approach involves gemtatistical simulation to generate a number of realizations reflecting differing degrees of spatial variation. The model is used simulate crop growth and maintain a nitrogen mass balance in the system. Primary elements of the methodology are discussed below. Management Scenarios and Variability Three application scenarios were developed in the methodology leading to the spatial management technology. The spatial management scenario is based on appropriate management of nitrogen to meet, but not to exceed production needs. Phosphorus and other nutrients are not assumed to be limiting factors. The scenarios include traditional practices and are modified to reflect possible advancement through application of spatial management technology (Figure 1). The first scenario is designed to consider uniform input of nitrogen based on traditional practices with the input rate set at a "typical" level and held constant through time. The second scenario assumes the same crop and uniform application, but the amount of application is modified based on soil and yield information from a single location in the field. Thus, the application is uniform over the field, but is variable with respect to time. Scenario 1 Scenario 2 UA-TP UA-16 Scenario 3 VA-16- Legend: UAWT Uniform Applicstioil ova Time VAWT Variable Appkafion over Time Figure 1. UA-TP Unifgm Application, Traditioaal Practice UA-1B UniformApplicltion, 1 CoatFd Point VA-16CP Vrtiablc Application, 16 b t r d Point -- Graphical illustration of three application scenarios. The third scenario uses variable application based on the spatial variation of selected model and decision parameters. The selected parameters include residual soil nitrate, crop yield and soil hydraulic conductivity. The field is divided into sixteen sectors, and the application rate is based on a single hypothetical measurement of the decision variables in each sector. Hence, the field will receive an application with respect to position as well as time conforming to decision variables within the sector. The methodology is applied to three different cases which are distinguished by the magnitude of field variability. They are referred to hereafter as; low, medium, and high variability case. In the low variability case, available field information for residual soil nitrate and yield was used to develop the lateral and vertical parameter distributions and geostatistical relations (Goderya et al., 1996; Goderya, 1996). Field conditions were then generated for medium and high variability cases. Mean values for these fields were comparable to the observed field data. However, each case is distinguished by differing coefficients of variation. For this study, values obtained from the literature were used to define these coefficients of variation (Goderya, 1996). Using the subjective prior values of mean, standard deviation and correlation length values for selected parameters, a number of equally likely unconditional realizations were generated. Employing this approach, medium and high variability field conditions were generated for residual soil nitrate and yield values. The general statistical properties for low, medium and high variability fields are given in Table 1. Geostatistical Simulation and Mass Balance A total of 100 field realizations, each realization having 120 locations, were simulated for each identified spatially variable parameter under each variability case. Simulated fields were then used as an input to an unsaturated zone transport model for predicting crop production and nitrate loading to groundwater. Table 1.--Generated field properties for low, medium, high variability cases. Residual NO3 (lcglha) Mean - 165.3 -- Std. Dev. 51.5 88.5 163.5 Coeff.Var. 0.31 0.60 1.06 Minimum 65.7 29.4 5.2 Maximum 362.5 518.8 861.0 Hence, the complete process of gwstatistical simulation and unsaturated zone transport modeling was executed within each variability case for each spatial management scenario. The spatial input data changed between each set of field realizations, but remained constant between each spatial management scenario. On the other hand, decision input changed between each spatial management scenario, but remained identical for various field realizations. For example, spatial inputs of residual soil nitrate, yield, and hydraulic conductivity were different between realizations 1 and 2, but they were identical between realization 1 of scenario 1 and realization 1 of scenario 2 and 3. A quasi-three-dimensional approach to modeling the important processes and maintaining a mass balance in the system is used in this research. In this case, a one-dimensional model is combined with geostatistical simulation to represent the heterogeneity (spatial variability) in a typical field. The intention of this formulation is to investigate contaminant loading and production from a heterogeneous field within a spatial framework. The model, TDNIT, is used in this research for simulating nitrate movement through root- and vadose- zone (Goderya et al., 1995). However, the methodology is not dependent on the type of model. The use of the selected transport model offers the advantage of short computation time and reduced input data demands. The results from this model also compared well with the program Erosion Productivity Impact Calculator (EPIC) in terms of its predictive ability (Goderya et al., 1995). A total of 48 simulation runs, each encompassing 100 field realizations, were completed for this effort. The resulting decision input and output were analyzed for each individual model node, and for each run over a five year period. CASE STUDY A case study is presented to demonstrate the methodology. A farm field in Central Nebraska represents a typical crop production area with the potential for nitrate contamination that may be found in the Midwest. Site selection was based upon availability of data, shallow water table, highly permeable vadose zone, rather uniform soil characteristics, and homogeneity of crops produced as well as the general agricultural practices employed. Here, most of the land adjacent to the site is cropland and nitrogen fertilizer is a major source of crop nutrients and nitrate contamination to groundwater. Continuous corn is selected as the crop to be simulated since it is the major crop grown in the case study area. Site specific information for the model input variables was compiled from a variety of sources. The transport program required meteorological input data in the form of daily precipitation (including irrigation amounts), average monthly temperature, average monthly solar radiation and albedo. Additional input data included depth to groundwater, potential mineralization amounts, the number of soil layers, soil data for each layer, plant data, and the initial moisture conditions for each layer. The soil input data (porosity, field capacity, permanent wilting point, sand content, organic matter content, saturated hydraulic conductivity, and residual water content) and initial conditions (water content and nitrate content) for the field were based on the vertical profile of the soil horizon and the unsaturated zone. The plant data include crop type, four different nitrate uptake coefficients, leaf area index values and root depth values as a function of time, dry mattedyield ratio, and potential maximum yield. The nitrogen fertilizer data include application amount, date and depth of application. RESULTS AND DISCUSSION The model data were evaluated in terms of area-wide leaching potential and subsequently for leaching as a function of spatial management practice. For all evaluations, output variables were used for relative comparisons. These results represent short-term impacts of the spatial management technology and were calculated for the following variables of interest: annual fertilizer input, annual crop uptake of nitrogen, and nitrogen losses to groundwater. Figure 2 summarizes the primary nitrogen input. In the low variability case, nitrogen fertilizer input was set at 200 kg/ha/yr (178 bulaclyr) for the traditional practice scenario, which compares favorably with the average corn fertilization rate in Nebraska (Follet et al., 1991). Fertilizer nitrogen inputs were lower for second scenario because of the information contained in the field sample which was used to adjust application rate. The overall amount of nitrogen fertilizer applied to the crop in the third scenario was reduced further due to the added level of information and use of spatial application methods. The simulated crop nitrogen uptake was used as a measure of yield potential. The annual uptake was similar under management scenarios 1 and 2 for all three cases of variability (Figure 3). However, there was a detectable decrease in uptake for scenario 3 (ie., spatial management technology). Based on these results, it appears as though production based on spatial management technology compares reasonably well with the traditional baseline output. Hence, while the spatial management scenario reduced the fertilizer requirements, it also resulted in simulated reduction in the uptake. The predicted amount of nitrate-nitrogen loading to the groundwater is presented in Figure 4. Potential nitrate loading to groundwater is significantly Scenarios - Figure 2. Annual nitrogen input for variability cases and management scenarios. - Figure 3. Simulated annual crop uptake for variability cases and management scenarios. Scenarios Figure 4. - Simulated nitrate leaching for variability cases and management scenarios. reduced for all three scenarios. However, the reduction is especially evident in the third scenario with high levels of field variability. The results of these simulations are consistent with the magnitude of the responses reported from experimental studies in the Midwest (Ferguson et al., 1994; Phillips et al., 1993). These results suggest that increased information for different parameters and conditions in a spatial management system may not necessarily result in a significant decrease in nitrate loading to groundwater. In fact, it appears that the success of spatial management technology is highly dependent on the degree of spatial variability of the primary field parameters that influence nitrate fate and transport. Thus, the effort associated with spatially variable application in a field with inherent low variability of key parameters would most likely be too costly to justify the benefits. However, as the field variability of the key process dependent parameters increases, the apparent environmental benefits of spatial management technology become clearer as depicted in Figure 4. SUMMARY AND CONCLUSIONS The methodology presented in this research incorporates spatial variability of various parameters in estimating groundwater contamination potential. It employs a combination of geostatistical simulation and unsaturated zone transport modeling. The methodology outlined in this research was applied to three conceivable variability cases, differing in extent of spatial variability. In all the cases, management practices of spatially variable application were evaluated on the basis of sustaining agricultural production and minimizing environmental pollution. Three management scenarios, including one using spatial management technology were developed and evaluated given the spatial distribution of residual soil nitrates, yield, and hydraulic conductivity in the fields. The first case was defined as one exhibiting low variability. Results of scenario modeling for this case indicate that while there is a slight reduction in nitrate leaching, the use of spatial management technology may not necessarily result in substantial benefits. It appears that uniform application, given realistic yield goals and composite soil samples, may achieve results comparable to spatial management technology in fields exhibiting low spatial variability. However, use of variable rate application not only reduced the over-application of nitrogen, but also reduced the non-point source pollution to groundwater due to agricultural practices for fields exhibiting both medium and high variability. The methodology described in this study provides a framework to ascertain if the use of spatially variable fertilizer application, or spatial management technology, is environmentally friendly and also provides a basis to evaluate the economics of the technology. ACKNOWLEDGEMENTS This paper is supported, in part, by the Center for Infrastructure Research, the Water Center, and the Agricultural Research Division at the University of Nebraska-Lincoln and, in part, by the Cooperative State Research Service (CSRS) of the U.S. Department of Agriculture. REFERENCES Ferguson, R.B., G.W. Hergert, J.E. Cahoon, T.A. Peterson, C.A. Gotway, and A. H. Hartford, 1994, "Managing spatial variability with furrow irrigation to increase nitrogen use efficiency," Second International Conference on Site-Specijic Management for Agricultural Systems, Bloomington 1 Minneapolis, Minnesota, March. Follet, R.F., D.R. Keeney, and R.M. Cruse, 1991, "Managing nitrogen for groundwater quality and farm profitability," Published by Soil Science Society of America Znc. , Madison, Washington. Goderya, F. S., 1996, "Evaluation and estimation of groundwater potential using spatial parameters, " Ph. D. Dissertation, Department of Civil Engineering, University of Nebraska, Lincoln, Nebraska. Goderya, F.S., M.F. Dahab, W.E. Woldt, I. Bogardi, 1995, "Comparison of Two Transport Models for Predicting Nitrates in Percolating Water", Submitted for publication in Transaction of the ASAE. Goderya, F.S., M . F . Dahab, W.E. Woldt, I. Bogardi, 1996, "Spatial Patterns Analysis of Field Measured Residual Soil Nitrate", In Geostatistics for Environmental and Geotechnical Applications, ASTM STP 1283, R . Mohan Srivastava, Shahrokh Rouhani, Marc V. Cromer, A. Ivan Johnson, Eds., American Society for Testing and Materials, Philadelphia (In press). Phillips, D.L., P.D. Hardin, V.W. Benson, and J.V. Baglio, 1993, "Nonpoint source pollution impacts of alternative agricultural management practices in Illinois: A simulation study", J. Soil and Water Conservation, 48:5:449-457. BIOGRAPHICAL SKETCH Wayne E. Woldt is an assistant professor in the Department of Biological Systems Engineering at the University of Nebraska-Lincoln. His interests include risk analysis, spatial management technology, and environmental systems analysis. Farida Goderya is a graduate research assistant in the Department of Civil Engineering at the University of Nebraska-Lincoln. She is currently completing her Ph.D. dissertation on spatial management technology. Mohamed F. Dahab is a professor in the Departments of Civil Engineering and Biological Systems Engineering at the University of Nebraska-Lincoln. His interests include pollution prevention and nitrate treatment methods for municipal water supplies. Istvan Bogardi is a professor in the Department of Civil Engineering at the University of Nebraska-Lincoln. His interests include risk assessment, systems analysis for environmental issues, and global climate change.