Criteria to select basins for hydrologic model development and testing

In light of a plethora of hydrologic models in existence today and interest in their development, there is a real need for formulating objective criteria to evaluate the models. Hydrologists have generally relied on retrospective testing of models over a few basins to evaluate model adequacy. This paper is concerned with the question of how many and what kind of basins and data are needed to conduct effective retrospective testing. Clearly, there should exist sufficient length of data period and adequate density of precipitation gages. Our recent study has indicated that at least 10 years of data are needed to observe a reasonable range of soil water storage variability and more years are preferred. The paper also suggests criteria for a minimum gage density standard. This implies minimum drainage area requirements depending on available gage density. Test basins should be representative of a variety of climates and of different hydrologic response characteristics in a given climate. Basins with single dominant vegetation and soil characteristics should be included to understand the effects of vegetation and soils on hydrologic processes. Basins with complex vegetation and soils are needed as well. Lastly, with availability of radar data, consideration must also be given to the location of test basins in relation to radar sites. Preferably, test basins should be located within 30-150 km range of the radar sites to avoid some of limitations of radar coverage. We will present examples of using these criteria for selecting basins for the international Model Parameter Estimation Experiment (MOPEX) project and for retrospective testing in the GCIP Land Data Assimilation Schemes (LDAS) project.