Uncertainty quantification of daily precipitation extremes

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Thursday, 8 January 2015
Michael F. Wehner, LBNL, Berkeley, CA; and P. Pall, M. Duffy, D. Stone, C. Paciorek, and W. D. Collins

We examine a variety of ways of quantifying uncertainty in the estimation of long period precipitation return values. Using a large ensemble from the climateprediction.net public participation computing project, we can estimate simulated long period Western US precipitation return values directly. We use this estimate to test the accuracy of generalized extreme value (GEV) statistical methods using samples of varying lengths. Critical to determining acceptable sample size is the quantification of uncertainty in the GEV return value estimates. We estimate this uncertainty from both Maximum Likelihood Estimator and L-moments approaches to fitting GEV distributions by a variety of parametric and non-parametric bootstrapping methods.