Evaluation of Biogeochemical Watershed Models: How Do We Know When A Model is Wrong?

点击翻译按钮获取中文摘要

0094312MeixnerBiogeochemical watershed models have become disconnected from rigorous hypothesis testing due to a lack of available evaluation tools, thereby limiting the usefulness of these models. I propose to improve the utilitty of biogeochemical watershed models by developing improved techniques of hypothesis testing using recently developed multi-criteria calibration techniques. Three questions will be addressed: (1) What are the benefits of distributed models of catchment biogeochemistry? (2) How do we incorporate multiple fluxes and state variables into our judgment of how well a model is performing? (3) How do we use different types of data in evaluating what is wrong with a model? Preliminary results with multi-media methods have shown promise in solving the hypothesis-testing challenge of biogeochemical watershed models. I will foster the development of these methods through both education and fundamental research into these techniques. I will develop an innovative group project on watershed mass balance models for the general hydrology course I teach. The graduate education portion of the project will involve lecture and collective learning in small groups using a simple bucket model of watershed processes. The group project will take the students from the rudiments of model construction to the tools available for model calibration and model evaluation of lumped as well as distributed hydrologic models. This course will form the foundation for the biogeochemical watershed modeling group I hope to create. The research I propose will focus on the development of multi-criteria model calibration and evaluation tools for multi-process biogeochemical watershed models. The proposed work includes testing the AHM model of the Emerald Lake watershed using "true" (testing for parameter uniqueness and identification) and real data sets using the available multi-criteria evaluation tools. Model calibration will be conducted for several scenarios comparing the level of spatial aggregation, time series duration, and the kinds of data available (outflow time series in addition to spatial data of soil conditions). These studies will investigate the utility of distributed biogeochemical watershed models and what sort of data is needed to properly parameterize and evaluate these models. I will also conduct similar modeling studies with agricultural non-point source models with the assistance of a post-doctoral associate trained in agricultural engineering. Finally, I will establish a biogeochemical watershed processes field site at the San Diego Experimental Forest.