ABSTRACT: Recent evidence that agriculture in general, and animal waste in particular, may be an important factor in surface and ground water quality degradation has induced a strong interest in nutrient management research. The presence of nitrogen and phosphorus in surface water bodies and ground water aquifers is recognized as a significant water quality problem in many parts of the world. A Generic Interactive Dairy Model (GIDM) has been developed as a tool for creating alternative dairy waste management plans and evaluating the effects of such plans on surface and ground water quality degradation. GIDM utilizes a GIS-based interface to develop field level management plans, run the GLEAMS water quality model and analyze the results obtained by means of tabular reports and thematic maps. GIDM runs on SUN SPARC stations using the ARC/INFO GIS software.
KEY TERMS: Dairy; nutrient management; nitrogen; phosphorus.
The main animal waste constituents which can impair water resources are nutrients and pathogens. The United States Environmental Protection Agency (USEPA) has assessed that 25% of more than 12 million lakes across the country surveyed in 34 states were at least partially threatened by water quality problems, and 20% were threatened by pollution from nutrients and sediments (Cooper et al., 1992). Some forms of nitrogen and phosphorus, such as nitrate nitrogen and soluble phosphorus, are readily available to plants. If these forms are released into surface waters, eutrophic conditions that severely impair water quality may result. The physical and chemical changes caused by advanced eutrophication may interfere with recreational and aesthetic uses of water. In addition, possible taste and odor problems caused by algae can make water less suitable or desirable for water supply and human consumption (Novotny and Chesters, 1981). Most water quality problems related to phosphorus result from transport with sediment in surface runoff into receiving waters. However, continuous high loadings from animal waste on very sandy soils with low retention capacity may contribute significant quantities of labile phosphorus to subsurface drainage. Ground water aquifers may also become polluted due to recharge of high loadings of nitrogen. Domestic consumption of water with nitrate nitrogen concentrations in excess of 10 mg/L may lead to methemoglobinemia in infants. In Florida, the Department of Environmental Regulation installed monitoring wells on nine dairies and of the 34 wells installed, 16 had nitrate concentrations above 10 mg/L (Darling, 1991).
Animal waste management has always been a part of farming, but historically has been relatively easy due to the buffering capacity of the land. In fact, land application of animal waste at acceptable rates can provide crops with an adequate level of nutrients, help reduce soil erosion and improve water holding capacity. However, as the animal industry attempts to meet the food requirements of a fast growing population, it tends to apply new technologies that lead to a reduced number of producers, but larger and more concentrated operations. That, in addition to the decreasing amount of land available for waste application, has resulted in an increased potential for water quality degradation. The successful planning of an animal waste management system requires the ability to simulate the impact of waste production, storage, treatment and utilization on the water resources. It must address the overall nutrient management for the operation, including other nutrient sources such as supplemental fertilizer applications.
GIDM integrates the GLEAMS (Groundwater Loading Effects of Agricultural Management Systems) (Leonard et al., 1987) water quality model and a Geographical Information System (GIS) allowing the user to assess alternative waste management plans for various performance criteria such as long-term trend of nitrogen and phosphorus movement into surface water bodies and aquifers.
By using models we can better understand or explain natural phenomena and under some conditions make predictions in a deterministic or probabilistic sense (Woolhiser and Brakensiek, 1982). A hydrologic model can be defined as a mathematical representation of the transport of water and its constituents on some part of the land surface or subsurface environment. Hydrologic models can be utilized as planning tools for determining management practices that minimize nutrient loadings from an agricultural activity to water resources. The results obtained depend on a good representation of the environment through which water flows and of the spatial distribution of rainfall characteristics. These models have been quite successful in dealing with time but they lack a spatial context and are often spatially aggregated or lumped parameter models. Recently, hydrologists have turned their attention to GIS for assistance in studying the movement of water and its constituents in the hydrologic cycle. A GIS can represent the spatial variation of a given field property by means of a cell grid structure in which the area is partitioned into regular grid cells (raster GIS) or using a set of points, lines and polygons (vector GIS).
There is obviously a close connection between GIS and hydrologic models and tremendous benefits in integrating them. Parameter determination is currently one of the most active areas in GIS related to hydrology. Parameters such as land surface slope, channel length, land use, and soil properties of a watershed are being extracted from both raster and vector GIS, with most work up to this time in raster-based systems. The spatial nature of a GIS also provides an ideal structure for modeling. A GIS can be a substantial time saver that allows different modeling approaches to be tried, sparing manual encoding of parameters. Further, it can provide the tool for examining the spatial information from various user-defined perspectives (Tim et al., 1992). It enables the user to selectively analyze the data pertinent to the situation and try alternative approaches towards the analysis. In fact, one of the most successful application areas for GIS has been in addressing problems of the environment.
Approaches for Integrating GIS and Models
The various approaches for integrating GIS and hydrologic models were described by Fraisse et al., 1995a. There has been a significant amount of work integrating both raster and vector GIS with hydrologic/water quality models. Several strategies and approaches for the integration have been tried. Initial work tended to use simpler models such as DRASTIC (Whittemore et al., 1987) and the Agricultural Pollution Potential Index (Petersen et al., 1991). In these cases the models were implemented within the GIS themselves. More complex models require a form of connection in which a common interface and transparent file or information sharing and transfer between the model and the GIS is provided. The interface between GLEAMS and the GIS was developed using ARC/INFO's AML which is a higher level application language built into the GIS. A subset of functions of a full-featured GIS, such as creation of maps (including model output) and tabular reports, as well as model-related analysis are embedded in the application, giving the system great flexibility and performance. LOADSS (Lake Okeechobee Agricultural Decision Support System) (Negahban et al., 1995), recently developed to evaluate the effectiveness of different phosphorus control practices in the Lake Okeechobee basin located in southern Florida, is an extension of this type of application since it includes an optimization module that enables the system to select the best phosphorus control practices based on goals and constraints defined by the user.
The choice between integrating a water quality model with a raster or vector GIS depends on the importance of spatial interactions in the process being studied and the nature of the model itself. Some water quality models such as GLEAMS are field scale models that provide edge of the field values for surface runoff and erosion as well as deep percolation of water and its constituents. In this case, spatial interactions between adjacent fields are ignored and a vector GIS can be used to describe the system. Moreover, important factors in the simulation process such as land use and management practices are normally field attributes and are thus, better represented in a vector structure. However, other factors playing an important role in the hydrologic process, such as field slope, aspect and specific catchment area are hard to estimate in vector systems. Watershed models in which the process of routing is important and spatial interactions are considered are better handled by raster based GIS. For those, several algorithms for estimating important terrain attributes are often incorporated in commercially available raster-based GIS.
GENERIC INTERACTIVE DAIRY MODEL
GIDM is designed to be an additional tool for answering questions related to the environmental impacts of dairy operations. A design schematic of the GIDM is given in Figure 1. It operates on an individual dairy basis and incorporates the GLEAMS water quality model for simulating nutrient transport of both nitrogen and phosphorus for specific fields of a dairy. It is designed to be generic, so that any dairy represented by a coverage for which relevant data such as topography, soil characteristics, weather and field boundaries are available can be simulated.
GLEAMS is a field scale water quality model which includes hydrology, erosion/sediment yield, pesticide and nutrient transport submodels. GLEAMS was developed to utilize the management oriented CREAMS (Knisel, 1980) model and incorporate more descriptive pesticide subroutines and more extensive treatment of the flow of water and chemicals in the root zone layers. The water is routed through computational soil layers to simulate the percolation through the root zone, but the volume of percolation in each layer is saved for later routing in the pesticide and nutrient components. A minimum of 3 and a maximum of 12 layers with variable thickness may be used. Soil parameter values are provided by soil horizon, and the crop root zone may have up to 5 horizons. The values for parameters such as porosity, water retention properties and organic content are automatically fitted into the proper computational layers. A modification of the SCS curve number method (U.S. Soil Conservation Service, 1972) is used to simulate runoff from daily rainfall. The daily time step is used for hydrologic computations, and thus all processes in nutrient transformations and fate use the same time increment. Two options are provided in the model to estimate potential evapotranspiration. The Priestly-Taylor method (Priestly and Taylor, 1972) and the Penman-Monteith method (Jensen et al., 1990). The nutrient component treats both phosphorus and nitrogen as nearly alike as possible, that is mineralization from crop residue, from soil organic matter, and from animal waste, immobilization to crop residue, solution and adsorbed phases for transport and routing, and crop uptake. Nutrient specific processes such as nitrogen fixation by legumes, denitrification and ammonia volatilization from animal waste are also taken into consideration by the model. The model simulates land application of animal wastes by creating appropriate nitrogen and phosphorus pools for mineralization. Ammonia volatilization from surface applied animal waste is estimated by using a relationship developed by Reddy et al.(1979).
The coverage representing a dairy to be simulated must be prepared so that the Polygon Attribute Table (PAT) associated with each polygon in the coverage contains attributes relevant to GLEAMS or a foreign key relating to other info tables or external files. Soil, weather, topography and field boundaries coverages must be combined into a single coverage before assigning practices and running GLEAMS. The methodology for combining the coverages is described in the GIDM users guide (Fraisse et al., 1995b). First field boundaries, soil type, slope, aspect and weather coverages are clipped to the area of interest and overlaid into a single coverage. Second, a spatial generalization process is used to decrease the number of polygons within each field. Third, the maximum overland flow length is calculated for each polygon.
While the soil type and weather coverages may be easily obtained from other sources, several steps may be needed to create slope and aspect coverages. Initial data may come in the form of a digital elevation model (DEM) which can be imported into ARC/INFO as a grid coverage. Generally the data will need to be smoothed using a low pass filter before the Arc LATTICEPOLY command can be used to generate percent slopes and aspect polygons. The final number of polygons in the simulation coverage is sensitive to these steps. Less filtering and smaller slope bins will result in larger number of polygons to be simulated. The resulting number of polygons can be further reduced by using map generalization techniques. McMaster and Shea (1992) describe two generalization methods for reducing the number of polygons in a vector-based coverage: classification and amalgamation. While classification consists of redefining the thematic attributes into fewer classes, amalgamation is the general term to describe the merging of adjacent polygons. Typically there are two ways of performing amalgamation: 1) elimination, by merging with adjacent polygons, all polygons below a set size threshold (e.g. the ARC command ELIMINATE, ESRI, 1991) and 2) merging adjacent polygons that share some attribute (e.g. the ARC command DISSOLVE, ESRI, 1991). Stallings et al. (1995) discuss the effects of the elimination technique on model results when the polygon size threshold is based on polygon sizes within agricultural fields. A one-percent threshold, for instance, would result in each polygon that consisted of less than one percent of the field area being merged with an adjacent polygon.
The PAT of every coverage used by GIDM must contain the following set of items: AREA, PERIMETER, COVERAGE#, COVERAGE-ID, AREA_ACRES, SLOPE, FLOW_LENGTH, ID_SOIL AND WSTA. The first four items are automatically written in the PAT by ARC/INFO, the other five items must be added by the user during the preparation of the coverage. Other items such as LANDUSE and COUNTY may be added by the user but are not required by GIDM. COVERAGE# is the key item to relate the PAT to other INFO tables. Figure 2 gives an overview of the INFO databases in GIDM. ID_SOIL and WSTA contain codes to connect with external files containing soil and weather data.
The graphical user interface is designed to help the user in planning a balanced nutrient management program for the dairy being simulated Figure 3. The very first activity that the user must perform is to select a dairy. A dairy is in fact a coverage at any scale containing the attributes required for running the model. All other activities are blocked if no dairy has been selected. Pressing the [Dairies] button in the main menu gives a list of available dairies (coverages) in the system. The user can select a dairy or cancel the menu by selecting 'none'. If the dairy is successfully selected, a base map of that dairy is displayed showing the contours of the dairy and its field boundaries. The field selection, simulation and map utilities menus also appear at this time on the left side of the screen. Now the user can create thematic maps of the selected dairy or define, modify, and simulate field management practices. Once a dairy is selected, the user can either select an existing plan to work with (Management plans sub-menu) or start defining new field and crop management practices. A plan includes a set of field management practices assigned to one or more fields in the dairy and the associated simulation results. However, the latter need not always be present.
Pressing the [Field Management] button in the main menu gives a pulldown menu that allows the user to define and modify field management practices and assign them to any set of selected fields. For each field, a sequence of crops can be defined in the Field Management Table and for each crop, the sequence of practices or field operations is defined in the Crop Management Table. Every time a waste application operation is defined or a field is used as pasture for a certain period of time, the corresponding amount of nutrients will be decreased from the amount available for assignment and the total available for future assignment updated. Once practices are assigned to one or more fields, the GLEAMS model can be run (Simulation menu) and results displayed in maps (Figure 4) or tabular reports (Result Reports sub-menu). The Maps sub-menu can also be used to print or save any thematic map being displayed on the screen. The Utilities sub-menu allows the user to use several Unix OpenWindows tools. The Fields menu provides the user with different ways to select fields and display information about a selected field. The Map Utilities menu contains a variety of options for displaying a dairy coverage.
Alternative plans can be designed and saved for comparison and selection of best management options. It should be mentioned that the best solutions in terms of reducing nutrient loadings to surface and groundwater must also consider economic aspects. The producer decision between competing waste management practices is ultimately economically motivated. A tool for economic analysis of the alternative management options will eventually be added to the system.
The search for solutions to the many problems related to nutrient management that affect water resources implies a continued demand for the development of modeling systems which can be used to analyze, in a holistic approach, the impact of alternative management policies.
A Generic Interactive Dairy Model (GIDM) has been developed as a tool for helping decision and policy makers to analyze the effects of alternative dairy waste management practices at a farm level. GIDM integrates the GLEAMS water quality model with ARC/INFO GIS software and utilizes a GIS-based interface to develop field level management plans for dairies, run the GLEAMS simulation model and analyze the results obtained. GIDM represents a different approach in integrating water quality models and GIS in the sense that it is designed to be generic and focused mainly on the farm level. The framework used in designing the interface can easily be adapted to handle different types of animal wastes (such as beef cattle and poultry) as well as to simulate the impact of other crop management practices such as pesticide applications.
This project is a contribution from the Institute of Food and Agricultural Sciences, University of Florida, as a part of the Southern Region Project S-249 of the USDA-CSRS with support from the U.S. Department of Agriculture, Economic Research Service.
Cooper, A. B., C. M. Smith, and A. B. Bottcher. 1992. Predicting runoff of water, sediment, and nutrients from a New Zealand grazed pasture using CREAMS. Transactions of the ASAE 35(1):105-112.
Darling, W. A. 1991. Status of Florida regulations of dairy farm waste management. Proceedings the National Workshop on National Livestock, Poultry and Aquaculture Waste Management. American Society of Agricultural Engineers. St. Joseph, MI. pp. 67-70.
ESRI. 1991. ARC/INFO Command References. Environmental systems Research Institute. Redlands, CA.
Fraisse, C. W., K. L. Campbell, J. W. Jones, W. G. Boggess, and B. Negahban. 1995a. Integration of GIS and Hydrologic Models for Nutrient Management Planning. Proceedings of the National Conference on Environmental Problem-Solving with GIS. EPA/625/R-95/004. pp. 283-291.
Fraisse, C. W., K. L. Campbell, J. W. Jones, and W. G. Boggess. 1995b. GIDM User's and Developer's Manual. Research Report AGE 95-3. Agricultural and Biological Engineering Department, University of Florida. Gainesville, FL. 112 p.
Jensen, M. E., R. D. Burman and R. G. Allen (Eds.). 1990. Evapotranspiration and Irrigation Water Requirements. American Society of Civil Engineers, Manuals and Reports on Engineering Practice, No. 70. 332 p.
Knisel, W. G. 1980. CREAMS: A Field-scale Model for Chemicals, Runoff, and Erosion from Agricultural Management Systems. Conserv. Res. Rep. No. 26. U.S. Department of Agriculture. Washington, D.C.
Leonard, R. A., W. G. Knisel and D. A. Still. 1987. GLEAMS: Groundwater Loading Effects of Agricultural Management Systems. Transactions of the ASAE 30(5):1403-1418.
McMaster R. B. and S. K. Shea. 1992. Generalization in Digital Cartography. Association of American Geographers. Washington, D.C.
Negahban, B., C. Fonyo, W. G. Boggess, J. W. Jones, K. L. Campbell, G. Kiker, E. Flaig and H. Lal. 1995. LOADSS: A GIS-Based Decision Support System for Regional Environmental Planning. Ecological Engineering 5(2+3):391-404.
Novotny, V. and G. Chesters. 1981. Handbook of Nonpoint Pollution: Sources and Management. Van Nostrand Reinhold Company, New York. 555 p.
Petersen, G. W., J. M. Hamlett, G. M. Baumer, D. A. Miller, R. L. Day and J. M. Russo. 1991. Evaluation Of Agricultural Nonpoint Pollution Potential in Pennsylvania Using a Geographic Information System. Environmental Resources Research Institute - ER9105. University Park, PA. 60 p.
Priestly, C. H. B. and R. J. Taylor. 1972. On the Assessment of Surface Heat Flux and Evaporation Using Large-Scale Parameters. Monthly Weather Review 100:81-92.
Reddy, K. R., R. Khaleel, M. R. Overcash and P. W. Westerman. 1979. A Nonpoint Source Model for Land Areas Receiving Animal Waste: II. Ammonia volatilization. Transactions of the ASAE 22(6):1398-1405.
Stallings, C., S. Khorram and R. L. Huffman. 1995. Spatial Simplification of Input Data for Hydrologic Models: Its Effect on Map Accuracy and Model Results. Proceedings of the 61st Annual Convention & Exposition of the American Society of Photogrammetry and Remote Sensing. Charlotte, NC.
Tim, U. S., M. Milner and J. Majure. 1992. Geographic Information Systems / Simulation Model Linkage: Processes, Problems and Opportunities. Paper No. 92-3610. American Society of Agricultural Engineers, St. Joseph, MI.
U.S. Soil Conservation Service. 1972. National Engineering Handbook: Section 4, Hydrology. Washington, D.C. 548 p.
Whittemore, D. O., J. W. Merchant, J. Whistler, C. E. McElwee and J. J. Woods. 1987. Groundwater Protection Planning Using the ERDAS Geographic Information System: Automation of DRASTIC and Time-related Capture Zones. Proceedings of the NWWA FOCUS Conference on Midwestern Ground Water Issues. Dublin, OH. pp. 359-374.
Woolhiser, D. A. and D. L. Brakensiek. 1982. Hydrologic System Synthesis. In Hydrologic
Modeling of Small Watersheds. Edited by C. T. Haan, H. P. Johnson and D. L. Brakensiek.
ASAE Monograph No. 5. St. Joseph, MI. pp. 3-16.
home page