NH42A-07
Coupled Bayesian hierarchical modeling of streamflow and precipitation extremes
Thursday, 17 December 2015: 11:50
309 (Moscone South)
Cameron Bracken, University of Colorado at Boulder, Civil Engineering, Boulder, CO, United States, Balaji Rajagopalan, University of Colorado at Boulder, Boulder, CO, United States, Linyin Cheng, University of California Irvine, Irvine, CA, United States and Subhrendu Gangopadhyay, Bureau of Reclamation Denver, Denver, CO, United States
Abstract:
There is strong evidence for the nonstationarity of streamflow and precipitation extremes in the western US both due to trends in time and large scale atmospheric oscillations. In the literature, there are many examples of nonstationary spatial precipitation and streamflow models, yet these analyses are typically conducted separately. It is reasonable to assume, especially in rainfall-runoff dominated regions, that annual or seasonal maximum precipitation and streamflow are closely related and can be estimated simultaneously. We present a spatial Bayesian hierarchical model for simultaneously estimating nonstationary precipitation and streamflow extremes. Extremes are coupled at the parameter level, allowing return levels to be estimated in space and time as well as under future climate conditions with the use of appropriate covariates. We apply the model to a small basin in the western US.
