HR: 1340h
AN: A53A-0852    [Abstracts]
TI: A Comparison of Statistical and Dynamical Downscaling for Surface Temperature in North America
AU: * Spak, S
EM: snspak@wisc.edu
AF: Center for Sustainability and the Global Environment, University of Wisconsin-Madison, 1710 University Avenue, Room 207, Madison, WI 53726 United States
AU: Holloway, T
EM: taholloway@wisc.edu
AF: Center for Sustainability and the Global Environment, University of Wisconsin-Madison, 1710 University Avenue, Room 207, Madison, WI 53726 United States
AU: Lynn, B
EM: bhl7@columbia.edu
AF: Center for Climate Systems Research, Columbia University, 2880 Broadway, New York, NY 10025 United States
AU: Goldberg, R
EM: rgoldberg@giss.nasa.gov
AF: NASA Goddard Institute for Space Studies, 2880 Broadway, New York, NY 10025 United States
AU: Hogrefe, C
EM: chogrefe@dec.state.ny.us
AF: New York State Department of Environmental Conservation and Atmospheric Sciences Research Center, University at Albany, 625 Broadway, Albany, NY 12222 United States
AB: A multiple linear regression statistical downscaling model is applied to estimate summer monthly mean surface temperatures over eastern North America. The model is calibrated with time series of $0.5\deg$ x $0.5\deg$ gridded observations from the U.S. Historical Climate Network (the predictand field) and a transient climate simulation with the NASA Goddard Institute for Space Sciences coupled ocean/atmosphere General Circulation Model (GCM), using forcings from the SRES A2 scenario (the predictor field). Three scales of predictor domain are assessed, ranging from regional to continental, in addition to predictor domains selected pointwise by objective analysis. The downscaling model is found to be much more sensitive to predictor domain than to the inclusion of mean sea level pressure as an additional predictor variable. Performance is compared to a dynamical regional climate simulation with the PSU/NCAR Mesoscale Model v.5 (MM5) for both current conditions (1997-1999) and future projections (2020s, 2050s, 2080s) with a standard suite of diagnostic statistics. The statistical and dynamical models exhibit comparable skill, and all consistently project greater warming than the GCM. Predictions from statistical and dynamical models approach convergence when adjusted for model bias. This experiment illustrates the research potential for employing GCM-derived surface temperature as a predictor field and highlights the importance of predictor domain selection in downscaling climate change scenarios.
DE: 9350 North America
DE: 1694 Instruments and techniques
DE: 1600 GLOBAL CHANGE (New category)
SC: Atmospheric Sciences [A]
MN: 2004 AGU Fall Meeting