Characterizing moisture delivery mechanisms for extreme precipitation in large geographic regions

Understanding dominant moisture delivery sources for extreme precipitation events is extremely important for characterizing their statistical behavior and behavior under specific climate regimes. Typically, for a given region, the largest extreme events occur in specific seasons but events occurring in off seasons can be just as socio-economically devastating. A complete picture of how and where events originate in all seasons paves the way for statistical forecasting and simulation of extreme precipitation. We present a data driven methodology applicable to large geographic regions that can partition heterogeneous areas into subregions and then characterize the moisture delivery mechanisms for each subregion under specific climate regimes (e.g., ENSO phases, PDO, etc.) and in each season. Extreme subregions are defined using a new nonparametric extreme value clustering method and moisture delivery characterization is done using the HYSPLIT storm backtracking algorithm. We apply this methodology to the Western United States where the nature of extreme events varies widely due to complex terrain, teleconnections and climate interactions.