EAPS

“Predictability of regional climate variability from subseasonal to multidecadal scales”

Speaker: Fuqing Zhang, Professor of Meteorology, Director, Center for Advanced Data Assimilation and Predictability Techniques, Penn State, University

Abstract: Accurate predictions of regional climate across all temporal and spatial scales can have important economic and societal impacts but equally important are the uncertainties associated with such predictions. This talk will give an overview of the regional climate predictability through examining sensitivity and ensemble simulations with state-of-the-art regional and global climate models for a few exemplar regional-scale climate phenomena that include the intra-seasonal Madden-Julian Oscillations during DYNAMO, the great US midwest flood of 1993, and the midlatitude winter weather extremes in response to Arctic sea ice loss. Also examined will be the predictability of regional climate at multi-decadal scales explored through examining the ensemble variability among CMIP5 simulations over North America. We demonstrate a fundamental apparent spatial scale limit below which useful additional refinement of climate model predictions may not be possible. That limit varies among climate variables and from region to region. We show here that the uncertainty (noise) in surface temperature predictions (represented by the ensemble spread among a large group of global climate model simulations) generally exceeds the ensemble mean (signal) at horizontal scales below 1000 km throughout North America, implying poor predictability. More limited still is the predictability of regional precipitation, which is also shown by limited number of convection-permitting RCM simulations. These findings highlight the challenges in predicting regionally-specific future climate anomalies, especially for hydroclimatic impacts such as drought and wetness.