Fate and Transport of Nitrogen What Models Can and Cannot Do
Working Paper No. 11
M. J. Shaffer
USDA, Agricultural Research Service
Great Plains Systems Research Unit
Fort Collins, Colorado
General types of applications and associated models
Availability of data for use in nitrogen models
Variability of field data needed to drive and test nitrogen models
Capabilities of soil root zone models to predict soil nitrogen status and potential nitrate leaching
Potential impacts on groundwater quality
Capabilities of soil root zone models to help design best management practices for N applications
Capabilities of groundwater nitrate-N models
Limitations of nitrogen models
What a Simulation Model Is and Isn't
Over the past several decades, attempts have been made to develop integrated theories (i.e., models) of the carbon and nitrogen cycle in soil-crop-aquifer systems. These models represent approximations to (i.e., simulations of) the actual processes, process interactions, and matter and energy exchanges that take place in the real world. The amount of detail contained in simulation models varies widely, depending on the needs and objectives of the projects under which they were developed.
How Decision Support Systems, Expert Systems, and Simulation Models Differ
Decision support systems (DSS's) provide their users with integrated tools to help them make better management decisions. Generally, these tools may include databases, simulation models, and expert systems. An important feature here is that making the final decision is left up to the user.
Expert systems attempt to capture the knowledge and decision-making logic of experts in a limited subject area and to place this capability into a computer program capable of making decisions similar to those that would have been reached by the original expert.
Simulation models attempt to approximate real-world processes and their interactions at the mechanistic level. They are extremely important components of decision support systems, and some expert systems may also contain simulation components. Simulation models usually contain logical relationships derived from the subject knowledge base and also may include expert systems.
The overlap and interrelationships of DSS's, expert systems, and simulation models can be quite involved, depending on a particular application. However, examination of the primary purpose of the system or model can be used to reveal whether it is a DSS, expert system, or simulation model.
Interrelationships between Modeling and Classical Field and Laboratory Research
A common misconception frequently stated in the literature is that classical field and laboratory research provides the basic knowledge, while modeling packages this knowledge for use by the end user. The reality of the situation, however, is somewhat different. This is particularly true of simulation models but also applies to DSS's and expert systems. Modeling generally involves integration of subsystems. Extremely important subsystem interactions cannot be studied or tested outside the context of an integrated model. This means that classical research that isolates processes for study has a difficult or impossible task in making real progress in areas where multiple process interactions are involved--such as soil carbon and nitrogen transformations and plant environmental stress--unless an integrated model is developed. Classical field research alone has not been able to adequately address these badly needed research areas. Evidently, modeling and classical research methods need to be combined at all stages of the research process where whole systems are involved.
Additional Model Benefits
Once a suitable model (or models) has been developed and tested, long-term simulation studies and interpolation of results between field research stations are possible. In addition, knowledge gaps concerning whole systems and subsystem interactions can be identified for further study. Education is another area where models can play a significant role. For example, students can use models to learn how systems respond to environmental and managerial inputs and which parameters and state variables are the most important.
General Types of Applications and Associated Models
Point and Field-scale Simulations
These model applications are used to address local impacts of various management, soil, and climate scenarios. Individual fields, research plots, or soils within a field are the spatial units of interest here. In general, most soil nitrogen models have been designed to address this range of field scales. Geographical Information System (GIS) technology can be used to spatially reference individual soil simulation analyses within a field and provide a link to "farming by soil" methods.
Farm-scale and Ranch-scale Analyses
Multiple fields and management enterprises are considered simultaneously. Models designed to make fertilizer recommendations or predict nitrate-N leaching at the point and field scale may also have application at the whole-farm or whole-ranch scale by aggregating results obtained for fields or smaller areas through the use of spatially referenced databases and GIS technology.
Regional or Basinwide Analyses
This approach combines GIS, remote sensing, and simulation technology to address large-scale spatial and temporal impacts of management, soil, and cllimate. The models being used here either are field-scale models that have been adapted for use at these larger scales or are large-scale, 2- or 3-dimensional models designed primarily for surface runoff calculations with some provision for subsurface flows.
Soil, Aquifer, and Combined Soil-Aquifer Models
Models also can be grouped into (1) those that address the crop root zone, (2) specific aquifer models, and (3) combined approaches that take an integrated look at the soil-aquifer system. Most nitrogen models are limited to the crop root zone, but some consider N transport and limited N-fate processes in aquifers, and a few models look at the combined effects of the soil root zone, deep vadose zone, and aquifer.
Examples of Nitrogen Models and Associated Scales
Soil nitrogen models have been developed at various levels of resolution and for various purposes (Hansen et al. 1994). Probably the most common type comprises the fertilizer recommendation models developed by the individual State Experiment Stations and by agribusiness. These models generally are based on results obtained from field trials and may use crop types and yield goals, soil NO3-N tests, leaf tissue and chlorophyll meter tests, soil organic matter levels, manure and legume credits, and other information sources to help calculate soil nitrogen budgets and make fertilizer recommendations to producers. Nitrogen models of this type are normally applied at the field scale and are limited to the crop root zone.
Another significant group of soil nitrogen models has been developed that can make assessments of nitrate-N leaching below the crop root zone as a function of soils, climate, and management. Examples include EPIC, Williams et al. (1984); GLEAMS, Knisel (1993); NLEAP, Shaffer et al. (1991); NTRM, Shaffer and Larson (1987); LEACHM-N, Wagenet and Hutson (1989); CENTURY, Wetherell et al. (1993); and RZWQM, USDA(ARS (1992). These models include soil process mechanisms at varying degrees of complexity for computing soil water and nitrogen budgets, and transport of nitrate-N through and out of the root zone. These models were initially developed for use at the point and field scales, but some of them, such as NLEAP, CENTURY, and LEACHM-N, have also been applied at the farm and regional scales through the use of GIS and related techniques (Wylie et al. 1994; Burke et al. 1989; Bleecker et al. 1990).
Another group of models has been designed primarily to estimate transport of nitrogen and other chemicals in surface runoff. These include general models such as CREAMS (Knisel 1980) and more detailed 2-dimensional models such as AGNPS (Young et al. 1989) and SWRRB (Williams and Nicks 1985). Other models such as EPIC, GLEAMS, NLEAP, NTRM, RZWQM, and LEACHM have 1-dimensional surface runoff components that include soil nitrogen.
Availability of Data for Use in Nitrogen Models
On-site Data and Their Relative Importance
Quantitative or semiquantitative model predictions of site-specific plant-available soil nitrogen, soil nitrate-N leached from the root zone, gaseous losses of N, soil carbon levels, and residual soil nitrate-N require local data on soil physical, chemical, and biological properties. For example, local measurements of soil properties such as plant-available water-holding capacity, percentage of soil organic matter (SOM), fraction of SOM in the fast mineralization pool, initial soil water content, and initial soil nitrate-N are needed to make site-specific predictions of nitrate-N leached and residual soil nitrate-N in field research plots and farm fields.
USDA Natural Resources Conservation Service (NRCS) Soil Databases
Typical NRCS soil databases used in nitrogen modeling include the SOILS 5/6 and the Pedon databases. These databases contain information on soil properties such as texture, drainage class, hydrologic group, bulk density, pH, plant-available water-holding capacity, percentage of organic matter, and percentage of coarse fragments. In addition, the Pedon database contains more detailed information on soil properties such as water retention relationships and soil chemistry.
NRCS databases such as the 1:24,000 SURRGO and the 1:250,000 STATSGO provide georeferenced data on soils for use in GIS and modeling applications. Soil property attribute types available in the SOILS 5/6 database are also generally available in the STATSGO and SURRGO databases.
The STATSGO database is available for the entire United States. However, the SURRGO database is under development and has been completed in only a limited number of States and localities. The relative usefulness of these databases in nitrogen models depends on the objectives and required resolution of a particular study. For example, the 1:24,000 SURRGO database has a rasterized resolution of about 28 m on the ground, while the STATSGO resolution is about 290 m. Also, generalization of local State soil survey data has been done in the national SOILS 5/6 database and in the SURRGO and STATSGO databases. Application of soil nitrogen models to specific fields may require remapping of the fields with an order-1 survey and accompanying soil sampling. Model applications at larger scales may be able to make use of existing SURRGO and STATSGO databases.
National Climate Databases
The National Climate Data Center (NCDC) database contains historical weather records at numerous stations across the United States. Daily data are available for precipitation, air temperature, pan evaporation, and snow. The database is available on CD-ROM from commercial companies, and the NRCS has its own version of the database on its computer system.
Local Soil and Climate Databases from Research Plots
Detailed soil and climate databases are frequently collected by researchers working on field plots. These data represent the most detailed information available for use in nitrogen models. Access to this information is through the individual research scientists.
These databases represent summaries of management practices commonly practiced in different regions of the country. They are really summaries of management systems that include cropping practices, tillage and fertilizer methods, irrigation practices, pest control, erosion and leaching control methods, and others.
Various models often have supporting databases. For example, the NLEAP model has regional databases for soil and climate information. These databases are designed to function with the specific model or models but may also have other applications.
Variability of Field Data Needed to Drive and Test Nitrogen Models
Model Output No Better Than Field Measurements
In general, the accuracy of model predictions cannot exceed the accuracy of the input data used in the analysis. This is particularly true of the more sensitive state variables. Also, in comparisons of model predictions with observed field data, the model cannot be tested beyond the accuracy of the field measurements.
Spatial and Temporal Variability in the Field
Field variability associated with measurements of soil residual nitrate-N and nitrate-N leached is known to be quite high. The reasons for this are numerous and include the complex, interrelated processes associated with the carbon and nitrogen cycles, the spatial variability of the soil and the management practices, and temporal variability of management as well as state variables such as temperature and precipitation.
Capabilities of Soil Root Zone Models to Predict Soil Nitrogen Status and Potential Nitrate Leaching
Status of Crop Residue, Manure, and Other Organic Amendment Pools
Many simulation models such as RZWQM, NTRM, NLEAP and CENTURY track the nitrogen and carbon contents of residue additions during the decay process. This information is useful in determining the stage of decay, potential contributions to and immobilization from the soil mineral N pool, contributions of carbon and nitrogen to the soil organic matter (humus) pools, and production of CO2.
Status of Soil Humus Pools (Fast and Slow)
The size of these pools helps determine how much soil organic N is potentially available for mineralization in a given year. Models such as CENTURY, RZWQM, NLEAP and others are designed to track the carbon and nitrogen contents of these pools over seasonal as well as longer time periods. This can provide valuable information relative to trends in the readily mineralizable (No) nitrogen pool, in the more stable nitrogen pools, and in the soil organic carbon levels.
Nitrogen Uptake by Crops
Soil nitrogen models have the capability of estimating the amount of mineral nitrogen (NH4-N and NO3-N) available for crop uptake. This information can be used in conjunction with a crop growth model or curve to estimate N uptake by the crop.
Gaseous Losses of Nitrogen
Soil nitrogen models have the capability of estimating soil gaseous losses from denitrification (N2 and N2O) and ammonia (NH3) volatilization (Hansen et al. 1994; Shaffer et al. 1991; Shaffer et al. 1992). Some soil models can also estimate fluxes of carbon dioxide (CO2). These capabilities have implications relative to studies involving greenhouse gases. Soil models can make predictions of gas fluxes for a variety of soil, climate, and management conditions that are difficult or too costly and time consuming to measure in the field.
Status of Nitrate-N Available for Leaching (NAL) and Residual Soil Nitrate-N
NAL is defined as the mass of soil nitrate-N in the root zone that is available for leaching after sources and sinks other than nitrate-N leaching have been considered. Residual soil nitrate-N is NAL minus nitrate-N leached from the root zone. NAL and residual soil nitrate-N in the root zone are valuable indicators of potentially leachable nitrate-N as well as important components of soil fertility status. Soil nitrogen models include these components as part of their nitrogen budget calculations. In particular, the models are capable of tracking these components over time during the growing season and the off-season periods (Shaffer et al. 1994).
Nitrate-N leached from the crop root zone is estimated in most soil nitrogen models by combining an estimate of nitrate-N dissolved in the soil pore water with estimates of soil water flux. The effects of dispersion and diffusion are accounted for by the introduction of appropriate coefficients into the solute transport equations. The effects of soil macropores on nitrate-N leaching are included in some of the research level models such as RZWQM, USDA-ARS (1992).
Potential Impacts on Groundwater Quality
Studies have shown that the mass of nitrate-N leached from the crop root zone often is positively correlated with nitrate-N concentrations in shallow underlying groundwater aquifers (Wylie et al. 1994). Nitrate-N leaching models can be used in conjunction with appropriate cropping system, soil, and climate data to make long-term estimates of annual nitrate-N leaching across broad geographical areas (Shaffer et al. 1993; Shaffer et al. 1994b; Wylie et al. 1994). Geographical Information System (GIS) maps of simulated nitrate-N leached can be used to help identify potential leaching hot-spot areas across an agricultural landscape. For example, a shallow alluvial aquifer along the South Platte River near Greeley, Colorado, was evaluated by Shaffer and Wylie (1994). Results for irrigated agriculture in the region showed that long-term steady-state predictions of the NLEAP model nitrate-N leached (NL) index were correlated with nitrate-N concentrations in the underlying shallow aquifer (figure 1).
** Missing Image **
Capabilities of Soil Root Zone Models to Help Design Best Management Practices for N Applications
Models can rapidly make long-term analyses as opposed to expensive and time-consuming field experiments. A range of potential best management practices (BMP's) can be evaluated using models, and the most promising ones can be field tested. Model results can be used in conjunction with field demonstration sites to help producers develop BMP's for their farms.
Examples of BMP Studies Using Models
Nitrogen models can be used to help determine management strategies that reduce leaching of nitrate-N while maintaining crop yields. For example, NLEAP simulations were used to determine the periods during the year when nitrate-N leaching is most likely to occur for sites in Ohio, Colorado, and North Dakota (Shaffer et al. 1994). This type of information is extremely valuable from the standpoint of strategic planning of nitrogen fertilizer applications and other N management techniques.
In another example (figure 2), NLEAP was used to simulate nitrate-N leaching under a sandy loam and a loam soil for furrow, surge, and sprinkler irrigation (Shaffer et al. 1994b). This type of model application provides a rapid method of determining relative potential leaching under alternative management scenarios.
** Missing Image **
Capabilities of Groundwater Nitrate-N Models
Groundwater models simulate water and solute transport, but processes such as denitrification and N uptake by riparian vegetation are not well quantified. These models are capable of simulating solute mixing and transport effects within the aquifer. Losses in nitrate-N are simulated using empirical degradation coefficients determined by calibration. Model examples include USGS-2D-Transport/MOC (Konikow and Gredehoeft 1978) and MODFLOW (McDonald and Harbaugh 1988). A major input to nutrient simulations in shallow aquifers often is nitrate-N leached from the root zone.
Limitations of Nitrogen Models
Input Data Limitations on Model Applications
Most model applications are limited by the availability of input data. For example, high-resolution field simulations (i.e., a few meters) of soil nitrogen status and nitrate-N leaching are generally limited to field research plots or fields where appropriate soil, climate, and management data are available. Simulations involving larger areas such as whole farms, drainage basins, and regions are limited to predictions of trends in a qualitative and/or relative sense. For example, would higher or lower leaching be expected under a given set of conditions as opposed to others? In such large-scale situations, values for nitrate-N leached or residual soil nitrate-N cannot be predicted at specific locations.
Use of Root Zone Models to Predict Aquifer Nitrate-N Concentrations
Root zone nitrogen models predict mass of nitrate-N leached and nitrate-N concentrations in the leachate. They do not, however, account for processes in the deep vadose zone and aquifer that can modify nitrate-N concentrations in a shallow underlying aquifer. For example, denitrification, dilution and mixing in the aquifer, aquifer sideflows, N uptake by deep-rooted riparian vegetation, travel times through the deep vadose zone, and other factors can make significant contributions to nitrate-N concentrations measured in the aquifer. Root zone leachate volumes and nitrate-N concentrations must be considered in conjunction with other factors in the deep vadose zone and aquifer before predictions can be made of nitrate-N concentrations in an associated shallow aquifer.
Quantification of Deep Vadose Zone and Aquifer Processes
Methods do not yet exist to adequately quantify certain processes such as denitrifica-tion in the deep vadose zone and aquifer, or N uptake from a shallow water table by deep-rooted vegetation.
Limitations in Testing and Evaluation of BMP's
Potential best management practice (BMP) benefits to nitrate-N leached cannot be distinguished better than the resolution of the model and its associated input data. For example, studies have shown that existing nitrate-N leaching models applied using feasible levels of research plot and farm-field level input data have a predictive resolution of about 20 to 50 lb N/ac/yr for residual soil nitrate-N at the end of the growing season and for annual soil nitrate-N leached below the root zone (Radke et al. 1991; Shaffer et al. 1991; Khakural and Robert 1993; Follett et al. 1994; Shaffer et al. 1994b; Hoffner and Crookston 1994). This means that BMP's for farm fields that are expected to alter nitrate-N leaching less than about 50 lb N/ac/yr probably cannot be tested using a simulation model.
Numerous models are available that calculate nitrogen budgets and simulate soil nitrogen processes within the crop root zone. These models have been shown to be useful in making fertilizer recommendations, in estimating leaching of nitrate-N below the crop root zone, in helping to design BMP's for efficient use of soil nitrogen inputs, and in estimating and maintaining soil carbon levels. Nitrogen models have also been developed that are useful in estimating agricultural loading of N to surface streams and water bodies. Nitrogen modeling associated with the deep vadose zone and shallow aquifers has been limited primarily to conservative routing and dispersion of nitrate-N without adequate consideration of source-sink processes within those regions.
Bleecker, M., J.L. Hutson, and S.W. Waltman. 1990. Mapping groundwater contamination potential using integrated simulation modeling and GIS. In Proceedings of: Application of Geographic Systems, Simulation Models, and Knowledge-based Systems for Landuse Management. Virginia Polytechnic Institute and State University, Blacksburg, Virginia. Pp. 319-328.
Burke, I.C., C.M. Yonker, W.J. Parton, C.V. Cole, K. Flach, and D.S. Schimel. 1989. Texture, climate, and cultivation effects on soil organic matter content in U.S. grassland soils. Soil Sci. Soc. Am. J. 53:800-805.
Follett, R.F., M.J. Shaffer, M.K. Brodahl, and G.A. Reichman. 1994. NLEAP simulation of residual soil nitrate for irrigated and non-irrigated corn. J. Soil and Water Conserv. 49:375-382.
Hansen, S., M.J. Shaffer, and H.E. Jensen. 1994. Developments in modelling nitrogen transformations in soil. In Nitrogen Fertilization and the Environment (chapter 3). Marcel Dekker, Inc.
Hoffner, G. and M. Crookston. 1994. 1993 Nutrient and Irrigation Management Education Program. Northern Colorado Water Conservancy District, Irrigation Management Service. 73 pp.
Khakural, B.R. and P.C. Robert. 1993. Soil nitrate leaching potential indices: using a simulation model as a screening system. J. Environ. Qual. 22:839-845.
Knisel, W.G. (ed.). 1980. CREAMS, a field scale model for chemicals, runoff, and erosion from agricultural management systems. Conservation Research Report 26, USDA-ARS. 640 pp.
Knisel, W.G. (ed.). 1993. GLEAMS: Groundwater Loading Effects of Agricultural Management Systems, Version 2.10. Univ. of Georgia, Coastal Plain Expt. Sta., Biol. & Agric. Eng. Dept. Publ. No. 5. 260 pp.
Konikow, L.F. and J.D. Gredehoeft. 1978. Computer model of two-dimensional solute transport and dispersion in ground water. Techniques of Water-Resources Investigations, Book 7, ch. C2. U.S. Geological Survey.
McDonald, M.G. and A.W. Harbaugh. 1988. A modular three-dimensional finite-difference ground-water flow model. Techniques of Water-Resources Investigations of the USGS, ch. A1. U.S. Geological Survey.
Metherell, A.K., L.A. Harding, C.V. Cole, and W.J. Parton. 1993. CENTURY soil organic matter model environment. Technical Documentation. Agroecosystems Version 4.0. Great Plains Systems Research Unit, Technical Report No. 4. USDA-ARS, Fort Collins, Co.
Radke, J.K., M.J. Shaffer, and J. Saponara. 1991. Application of the Nitrogen-Tillage-Residue-Management (NTRM) model to low-input and conventional agricultural systems. Ecol. Modelling 55:241-255.
Shaffer, M.J., B.K. Wylie, R.F. Follett, and P.N.S. Bartling. 1994a. Using climate/weather data with the NLEAP model to manage soil fertility and nitrate leaching. Agr. and Forest Met. 69:111-123.
Shaffer, M.J., A.D. Halvorson, and F.J. Pierce. 1991. Nitrate leaching and economic analysis package (NLEAP): Model description and application. Chapter 13, pp. 285-322. In R.F. Follett et al. (eds.), Managing Nitrogen for Groundwater Quality and Farm Profitability. Soil Science Society of America, Inc., Madison, Wisconsin. 357 pp.
Shaffer, M.J. and W.E. Larson (eds.). 1987. NTRM, a soil-crop simulation model for nitrogen, tillage, and crop-residue management. Conservation Research Report 34-1, USDA-ARS, National Technical Information Service, Springfield, Virginia.
Shaffer, M.J., K. Rojas, D.G. DeCoursey, and C.S. Hebson. 1992. Summary Documentation for OMNI, the organic matter/nitrogen cycling component of the Root Zone Water Quality Model (RZWQM). Pp. 118-148. In Volume 1, Technical Documentation for the Root Zone Water Quality Model. USDA-ARS, Great Plains Systems Research Unit, Fort Collins, Colorado.
Shaffer, M.J., M.K. Brodahl, and B.K. Wylie. 1993. Integration and use of the Nitrate Leaching and Economic Analysis Package (NLEAP) in the GIS environment. In Proceedings of the Federal Interagency Workshop on Hydrologic Modeling Demands for the 90's. June 6-9, 1993, Fort Collins, Colorado.
Shaffer, M.J., B.K. Wylie, and M.K. Brodahl. 1994b. NLEAP as a predictive tool for regional nitrate leaching in Colorado. In Great Plains Soil Fertility Conference Proceedings. March 7(9, 1994, Denver, Colorado. Pp. 197-202.
Shaffer, M.J. and B.K. Wylie. 1994. Identification and mitigation of nitrate leaching hot spots using NLEAP/GIS technology. In Transactions of 15th World Congress of Soil Science, July 10-16, 1994. Acapulco, Mexico. Volume 5a: Commission IV: Symposia. Pp. 151-164.
Wagenet, R.J. and J.L. Hutson. 1989. LEACHM: Leaching Estimating and Chemistry Model--a process based model of water and solute movement, transformations, plant uptake and chemical reactions in the unsaturated zone. Continuum Vol. 2. Water Resources Institute, Cornell University, Ithaca, New York.
Williams, J.R., C.A. Jones, and P.T. Dyke. 1984. A modeling approach to determining the relationship between erosion and soil productivity. Trans. ASAE 27:129-144.
Williams, J.R. and A.D. Nicks. 1985. SWRRB, a simulator for water resources in rural basins: an overview. In D.G. DeCoursey (ed.), Proc. of the National Resources Modeling Symp., Pingree Park, Colorado, October 16-21, 1983. USDA-ARS, ARS-30, pp. 17-22.
Wylie, B.K., M.J. Shaffer, M.K. Brodahl, D. Dubois, and D.G. Wagner. 1994. Predicting spatial distributions of nitrate leaching in northeastern Colorado. J. Soil and Water Conserv. 49:288-293.
Wylie, B.K., M.J. Shaffer, and M.D. Hall. 1995. Regional assesment of NLEAP NO3-N leaching indices. Water Resources Bulletin 31:399-408.
Young, R.A., C.A. Onstad, D.D. Bosch, and W.P. Anderson. 1989. AGNPS: A nonpoint-source pollution model for evaluating agricultural watersheds. J. Soil and Water Conserv. 44:168-173.
< Back to RCA Publication Archive