Modern Statistical Techniques for Characterization of Extreme Precipitation During Atmospheric River Events

Monday, May 12, 2014 - 07:00

Extreme value theory concerns the application of statistical methodologies to understand low-frequency but high-impact extreme events in climate data. In particular, spatial modeling of climate extremes has been investigated to account for regional patterns of extremes and to characterize the dependence among locations based on the extreme value theory. However, there has been relatively little study on statistical extreme value analysis of changes in characteristics of extreme precipitation during atmospheric river events from CMIP5 projections. Atmospheric Rivers (ARs) are large spatially coherent weather systems with high concentrations of elevated water vapor that often cause severe downpours and flooding over western coastal United States. We have recently developed TECA (Toolkit for Extreme Climate Analysis) for automatically identifying and tracking features in climate datasets. In particular, we are able to identify ARs that make landfall on the western coast of North America. This detection tool examines integrated water vapor field above a certain threshold and performs geometric analysis. Based on the detection procedure, we investigate impacts of ARs by exploring spatial extent of AR precipitation for CMIP5 simulations, and characterize spatial pattern of dependence for future projections under climate change within the framework of extreme value theory. The results show that AR events in RCP8.5 scenario (2076-2100) tend to produce heavier rainfall with higher frequency and longer duration than the events from historical run (1981-2005). Range of spatial dependence between extreme precipitations is concentrated on smaller localized area in California under the highest emission scenario than present day. This study was funded under the CASCADE Science Focus Area funded by DOE/BER as part of the Regional and Global Climate Modeling (RGCM) Program.

Presentation file(s):