Date: Thursday, September 24, 2009 4:30 PM
Location: 366 Hollister Hall
As we build more and more realistic and detailed models of environmental systems, we must also develop methods powerful enough to evaluate (test) and correct these models. In particular, such methods must be 'diagnostic', meaning they must help illuminate to what degree a realistic representation of the real world has (or has not) been achieved and (more importantly) how the model should be improved. It is suggested that many current strategies for confronting hydrologic system models with observational data are inadequate in the face of the highly complex models becoming central to modern environmental modeling, and steps are proposed towards the development of more robust and powerful evaluation approaches while considering the uncertainty present in data and model. Through the use of so-called signature indices, which measure theoretically relevant environmental system process behavior, we can address the issue of the degree of system complexity resolvable by a model. Using signatures as a main component in uncertainty and sensitivity analyses enables us to evaluate dynamic models in new ways. In this presentation I will outline the suggested approach and will provide examples of how we can apply it to both an individual model - to test its consistency with underlying system behavior, as well as during model comparison studies - to identify which degree of model complexity is required to reflect the functional behavior of a particular hydrologic system. The results shown are based on several watershed modeling studies across climatic gradients in the US.
Refreshments will be served at 4:15 PM
