Temperature Extremes and Associated Large-Scale Meteorological Patterns in NARCCAP Regional Climate Models
Paul LoikithPostdoctoral Scholar
California Institute of Technology
|Date:||Friday, July 11, 2014|
|Time:||11:00 a.m. – noon|
|Location:||Argonne National Laboratory|
TCS Building 240
In an investigation on the fidelity with which regional climate models (RCMs) simulate temperature extremes, large-scale meteorological patterns (LSMPs) associated with extreme-temperature days are evaluated for a suite of RCMs that are part of the North American Regional Climate Change Assessment Program (NARCCAP). LSMPs are composites of surface air temperature, sea level pressure, and 500-hPa geopotential height anomalies concurrent with extreme-temperature days. Six of the RCMs (from a hindcast experiment) are driven by observed boundary conditions, while 11 RCMs are driven by one of four global climate models (GCMs). Four case studies are analyzed in detail. Model fidelity is high for cold winter extremes near Chicago but weaker for cool summer extremes near Houston and extreme heat events in the Ohio Valley. The RCMs have fundamentally different LSMPs associated with extremely warm days over much of California in the winter, suggesting a need for higher resolution. The results also suggest that the ability of an RCM to reproduce a realistic temperature distribution shape, especially at the tails, is related to model fidelity in simulating LMSPs. The entire suite of simulations is evaluated over the entire domain, and each ensemble member is ranked. Overall, the multi-RCM ensemble mean LSMPs resemble observations better than does any individual ensemble member. The methodology developed here provides a framework for identifying locations where further process-based evaluation would improve the understanding of simulation error and help guide future model improvement and downscaling efforts.