Choosing among alternative hydrologic regression models

Bayesian theory provides for the explicit accounting of both parameter and model uncertainties. It is used in this work to derive a procedure for discriminating among alternative hydrologic regression models. In particular, the procedure was used to discriminate among alternative exogenous variables in regression models. Controlled experiments with hydrologic models were designed to test the proposed procedure under different assumptions on model prior probabilities, length of sample, and model subset. These examples showed that besides its theoretical advantages, the use of the Bayesian procedure unambiguously selects the correct model in most of the applications.