A TOP-DOWN FRAMEWORK FOR WATERSHED MODEL EVALUATION AND SELECTION UNDER UNCERTAINTY

Open Access
Author:
Bai, Yaoling
Graduate Program:
Civil Engineering
Degree:
Master of Science
Document Type:
Master Thesis
Date of Defense:
July 17, 2008
Committee Members:
  • Thorsten Wagener, Thesis Advisor
  • Patrick M Reed, Thesis Advisor
Keywords:
  • model selection
  • top-down approach
  • hydrologic model
  • uncertainty
Abstract:
This study presents a novel top-down strategy for model evaluation and selection under uncertainty. It extends the top-down approach suggested by Klemes and formalized by Sivapalan and colleagues through a Monte Carlo framework in which model ensembles are tested for four different watershed response signatures across time scales. Lumped watershed model structures of increasing complexity have been applied to twelve watersheds across a range of hydro-climatic conditions within the US. The necessary minimum complexity and the related model assumptions provide indicators of the dominant controls on the watershed response at each temporal scale represented by different water balance signatures. Probabilistic measures of model performance with respect to reliability (Is the model ensemble capturing the observed signature?) and with respect to shape (Is the model structure capable of representing the signature variability?) have been developed to distinguish the ability of the models to represent watershed response behavior for each time-scale. The probabilistic measures are combined in a fuzzy rule system to guide model selection. Results suggest that the framework can be tuned to function as a screening tool that formalizes our model selection process. This fuzzy model selection framework therefore enhances our ability to objectively and automatically select parsimonious model structures for large databases of watersheds.