CURRENT EVALUATION EFFORT

The team met in National Headquarters in January, 2000, to begin the process of developing a matrix of model attributes and the inventory of current models as identified in the first two steps of the five step process included as Appendix A. The team was assisted by input from other members of the two centers and the institute, as well as some model developers who were kind enough to assist us evaluate model attributes.

Thirty some water quality related models of current interest to NRCSers were identified and listed in the matrix. The team recognized there are many more tools being used for water quality purposes across the country not on our list. These other models are generally proprietary in nature, and in many cases require detailed site information (much more information than the suite of tools included) not readily available.

Some of the models included in the matrix were older tools such as EPIC and GLEAMS that are being used by individuals and groups in the agency. Some such as SWAT have been available for a number of years, but there has been no concerted technology transfer activity by the agency. Newer tools such as Specially Protected Areas and Wildlife (SPAW) and AGNPS98 are recently operational and to some extent still being developed and updated. A limited number of European models were included as well.

Model attributes were broadly grouped into categories of hydrology, sediment, nutrients, and pesticides. Other water quality parameters included as attributes were salinity, pathogens, temperature, and dissolved oxygen. Members of the team completed the matrix for each of the models, and in some cases more than one team member completed the matrix for a tool. Where multiple ratings were available, these were consolidated into a single rating. Further, the matrix was examined to insure models with similar components (such as the nutrient and pesticide components of EPIC and GLEAMS) were given similar ratings.

Attributes were rated as high, medium, and low if the tool had the attribute. High indicated the model element was present with a very detailed, sophisticated, physically based representation. Medium was an indication the model element was present with a conceptual or moderately complex representation. A low rating indicated the model element was present, but had a very simple, empirical representation. Question marks were used to indicate the model element was present, but the quality of the representation was not known. Blanks meant the element was not present.

Other elements in the matrix were grouped into categories of data requirements, model output, application time and spatial scale, the NRCS level at which the technology is most appropriately applied, and information about model development and support. Most of these elements are rated with either a "yes" or a "no." A blank can be considered as a "no."

The team was not familiar enough with some of the tools to identify elements with which to complete the matrix. In those cases, the model developer and/or specialists within NRCS were contacted to obtain help with the matrix. Where model developers or other NRCS specialists were able to assist, the matrix was completed. There were several models where no one could be located to complete the matrix and these tools were removed. The final matrix posted on the National Water Management Center web site at http://wmc.ar.nrcs.usda.gov/ includes twenty-six models. The definitions of the attributes and other matrix elements are also included on the web site.

Models are complex, and there are many pitfalls and places for error. The application of a model at any NRCS office level requires a great deal of attention to detail and watchful care; it is not something that can be done hastily, halfheartedly, or carelessly. This, naturally, requires a significant commitment of resources. If properly done, however, this can save money in the long run by correctly identifying water quality problems and source areas; if done improperly, it will just lead to confusion and a waste of resources.

Most of the tools identified were rated for use at a state or national level. Exceptions were the irrigation tools, SPAW, and Revised Universal Soil Loss Equation (RUSLE) that are intended for the field or area office use. Water quality models require a significant amount of input data preparation, and they require a certain level of expertise to apply them meaningfully and to interpret the results properly. Consequently, as a whole, they are not well suited as field office tools per se.

Some decisions at the field office level can undoubtedly be made with simple methods and do not require extensive analysis or sophisticated tools. Other decisions, however, conceivably involve more complex processes or spatial interactions that cannot be adequately addressed with simple methods. For example, hydrologic, erosion, and water quality processes for a watershed cannot, in general, be properly described only at the point or field scale because they are in reality watershed scale processes. That is, a field does not exist in isolation, but rather there are interactions with other fields up and downslope, and, of course, the water quality in a stream is the result of the integrated effect of all of the land area in its watershed. Because of these complex processes and spatial interactions, one cannot expect to have a simple water quality tool that can be an adequate basis for all land management and conservation practice decisions.

Field scale tools are useful when looking at the impact of specific conservation measures. The field scale allows the user to examine in some detail the natural processes that occur in the field and at a point. This detail would be difficult to examine in a large-scale tool that tends to integrate the small-scale point processes into the larger picture.

Even if a model is used at a state office, this does not mean it has no relevance to field office operations. In the more complex, watershed scale issues, for example, it would seem to be the proper role of the state office to assist the field offices by running a model and providing them with the results. Many times the results can be in the form of charts, graphs, or tables that relate field conditions with the impacts modeled. In effect, the chart, table, or graph then becomes the field office or field level tool. In this regard the rating system used in this matrix which keys in on the expertise of the user may be somewhat misleading.

From the inventory of twenty-six water quality models, fifteen were identified with the highest potential for further development and technology transfer. The fifteen are grouped by scale and use as much as possible. It becomes evident many of the tools have like attributes with similar levels of complexity. In some cases components were purposefully matched (EPIC and GLEAMS share nearly identical nutrient and pesticide components). In other cases the components utilize a somewhat different technology, but technologies that have similar complexity. The matrix for the fifteen models is included as Appendix B, and the definitions of the attributes and other matrix elements is included as well. In addition, a brief description of the fifteen models and a point of contact for each is included.

No single existing model is adequate in all areas of importance to NRCS or is generally applicable everywhere in the US. For example, the best erosion models do not include water quality, all of the water quality models reviewed are deficient in some way in their hydrologic component, and some of the models are applicable only to certain regions where one can assume the dominance of particular processes. Regarding the hydrology, most of the current USDA water quality models, with the exception of Root Zone Water Quality Model (RZWQ), use the Soil Conservation Service (SCS) runoff curve number process to describe the relationships between rainfall, runoff, and infiltration. This process is dated and in some instances can be greatly in error. The basic water fluxes must be simulated adequately to be able to simulate erosion and water quality, yet hydrology is weak in some models.