School of Meteorology (Defense)
Thesis Defense: Probabilistic Flood Prediction Using Ground-Based and Space-Based Sensors
OU School of Meteorology
01 May 2012, 11:00 AM
National Weather Center, Room 4140
120 David L. Boren Blvd.
University of Oklahoma
This study develops a Bayesian framework for evaluating the skill of precipitation estimates in the context of flood prediction. The framework makes use of archived United States Geologic Survey stream gauge observations for determining flood events, and then simulates streamflow using a distributed hydrologic model run with a-priori parameters and forced with precipitation from NEXRAD and satellite-based (TMPA 3B42RT, 3B42V6) rainfall products. The time series of simulated streamflow is converted into return period and compared against observed return periods. The degree of correlation between these time series indicates the prior predictability of flooding, conditioned on the QPE forcing. Results based on forcing from the satellite-based estimates, 3B42RT & 3B42V6, at flood estimation are analyzed as a function of basin area and hydro-climatic regime. These results are compared and contrasted to the reference hydrologic skill that used forcing from the higher resolution, gauge-calibrated NEXRAD rainfall product. A demonstration will be shown how this framework can be extended to provide probabilistic predictions of flooding for gauged locations.