4.2
Regional Climate Modeling of Interannual Variability: EOF Analysis
Jan F. Dutton, Penn State University, Univeristy Park, PA; and E. J. Barron
Many regional climate model (RCM) studies have shown that increased orographic, topographic, and numerical resolution of an RCM tends to improve the simulation of mean climates relative to a forcing global climate model (GCM). The simulation of interannual variability by regional climate models has only recently been explored. Either these studies have used small ensembles of a few months or simulations of a few years to characterize regionally simulated interannual variability. In this study, a 10 year six member ensemble RCM/GCM simulation is used to study the simulation of interannual variability. The overall purpose of this project is validation of the RegCM2 regional climate model simulation of interannual climate variability. A six-member ensemble of the NCAR CCM3 GCM at T31 resolution is initiated using different initial conditions but forced by the same 10 year NCEP observed monthly varying SST data set. Each of these CCM3 simulations provides the 12 hourly one-way boundary conditions for 6 RegCM2 simulations centered at 45N, 100W with a 108 km resolution. The simulated period for the CCM3 and RegCM2 ensemble is 1969 to 1993.
The analysis presented focuses on empirical orthogonal functions of 60 years of simulated and gridded observed January and July precipitation and surface temperature. The results for winter precipitation variability indicate that the primary mode of Western U.S. variability is simulated by both RegCM2 and CCM3, although the magnitude of RegCM2 variability is improved relative to CCM3. The primary mode of winter variability in the observed data, centered in the Southeastern U.S. is not simulated by the models.
The simulation of winter surface temperature variability is well simulated by the models. Whereas the winter precipitation EOF1 and EOF2 between models and observed data are substantially different, the EOF1 and EOF2 patterns between RegCM2, CCM3, and the CRU data show similarities. However, the models under predict the magnitude of the EOF1 surface temperature variability.
The analysis of summer precipitation variability indicates that little similarity exists when comparing RegCM2 and CCM3 in both space and time. Likewise, little similarity exists between the models and observations. While the surface temperature variability patterns are geographically different, the variation of the RegCM2 and CCM3 time coefficients suggests CCM3 is forcing the July surface temperature variability.
Session 4, Regional Modeling and Downscaling (Parallel with Session 3)
Monday, 15 January 2001, 3:30 PM-5:15 PM
Previous paper Next paper