570 Climate extremes in uncoupled and coupled regional climate models

Wednesday, 9 January 2013
Exhibit Hall 3 (Austin Convention Center)
Tarun Verma, Texas A&M University, College Station, TX; and J. S. Hsieh, C. M. Patricola, R. Saravanan, and P. Chang

Handout (6.8 MB)

The study of climate extremes has acquired increased importance in recent years as attention has shifted from the effects of climate change on mean properties of climate to its effect on the higher-order properties of climate, such as heat waves, droughts, floods etc. Improved simulations and predictions of these events could help mitigate the catastrophic effects of these extreme events.

Global climate models that use relatively coarse horizontal grid resolution, often have difficulties in simulating the strong localized fluctuations associated with the climate extremes. To overcome this difficulty, regional climate models (RCMs) with fine horizontal resolution are used. Finer spatial resolution can provide better simulation of smaller or local scale phenomena often associated with extremes of the temperature and precipitation. However, RCMs are usually integrated in an uncoupled mode, using a regional atmospheric model forced by prescribed sea surface temperatures. This approach ignores potential feedbacks between the atmosphere and the ocean that could have an impact on climate variability.

To incorporate the potential impacts of air-sea feedbacks on climate statistics, we have constructed a coupled regional climate model (CRCM), where the Weather Research and Forecasting (WRF) model is coupled to the Regional Ocean Modeling System (ROMS). In our study, we consider both the CRCM and an uncoupled RCM, using the WRF model alone. We will analyze the statistics of climate extremes in the RCM and the CRCM, focusing on continental U.S/Mexico, and adjoining oceanic regions. We will compare these model results to each other, as well as to the observed properties of extremes as derived from NCEP Reanalyses. In addition to basic statistical analysis such as comparison of the probability distribution functions and return values, we will also carry out statistical modeling using extreme value theory.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner