Date of Award
Master of Science (MS)
Mathematics and Statistics
Physical models in the hydrological sciences are often calibrated using methods that do not formally quantify uncertainty in the model parameters. Additionally, many competing hydrological models exist and are used to model the same processes. Considering existing mechanistic models of rainfall-run off in a statistical context can assist hydrologists in understanding the true physical process taking place. This paper introduces a data assimilation mixture model of runoff that yields statistical estimates of hydrological mode l parameters and predictions. This statistical model incorporates two commonly used hydrological models, each with strengths and weaknesses. The mixture framework allows comparisons between models as well as combines the strengths of both. Results from three implementations of the mixture model are summarized and additional generalizations of the models are suggested.
Flake, Darl D. II, "Mixtures of Truncated Normal Data Assimilation Models for Parameter Estimation and Prediction in Hydrological Systems" (2008). All Graduate Plan B and other Reports. 1290.