Low (web) Quality - BALTEX
Low (web) Quality - BALTEX
Low (web) Quality - BALTEX
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
85<br />
Present climate simulations (1982-2003) of precipitation and surface<br />
temperature using the RegCM3, RSM, and WRF<br />
Yoo-Bin Yhang, Kyo-Sun Lim, E-Hyung Park and Song-You Hong, ybyhang@yonsei.ac.kr<br />
1. Introduction<br />
Interannual variations of precipitation and surface air<br />
temperature are integral parts of the climate system, where<br />
remote controls by planetary circulation and global surface<br />
anomalies act together with local influences by regional<br />
mesoscale surface characteristics. These two broad factors<br />
(remote and local) can be distinguished by using a regional<br />
climate model (RCM), in which planetary signals are<br />
integrated through lateral boundary conditions while<br />
mesoscale impacts are internally resolved (e.g., Giorgi et al.,<br />
1993; Liang et al., 2001). Given the inadequacies of general<br />
circulation models (GCMs) to simulate regional climate<br />
variability, the RCM downscaling has become a powerful<br />
alterative and is widely applied in studies of seasonalinterannual<br />
climate prediction and future climate change<br />
projection.<br />
It is imperative that any RCM must be rigorously validated<br />
in reproducing historical observations, including both mean<br />
climate and temporal variability, before credible application<br />
for climate change projection. In the published literature,<br />
numerous studies have demonstrated the RCM skill<br />
enhancement for downscaling regional characteristics,<br />
especially of precipitation and surface air temperature,<br />
focusing on mean climate (long-term averaged) biases such<br />
as in the annual cycles (Pan et al., 2001; Roads et al., 2003).<br />
In this study, three different RCMs, the Weather Research<br />
and Forecasting model (WRF), National Centers for the<br />
Environmental Prediction (NCEP) Regional Spectral Model<br />
(Juang et al., 1997), and Regional Climate Model version 3<br />
(RegCM3, Pal et al., 2007), are chosen to examine<br />
downscaling skill. The downscaling sckills of RCMs are<br />
compared with the driving reanalysis, against observation of<br />
interannual variations of precipitation and surface<br />
temperature during 1982-2003 over East Asia. This is<br />
facilitated by using empirical orthogonal function (EOF) and<br />
correlation analyses.<br />
2. Model and experimental design<br />
Three regional climate models used in this study are the<br />
RSM, WRF, and RegCM3. The RSM is a primitive equation<br />
model using the sigma-vertical coordinate. The model<br />
includes parameterizations of surface, boundary layer (BL),<br />
and moist processes that account for the physical exchanges<br />
between the land surface, the boundary layer, and the free<br />
atmosphere. The RegCM3 is a primitive equation,<br />
compressible, sigma-vertical coordinate, and limited area<br />
model of which the dynamical core is based on the<br />
hydrostatic version of the fifth-generation Penn State<br />
University-National Center for Atmospheric Research (PSU-<br />
NCAR) Mesoscale Model (MM5; Grell et al. 1994). The<br />
Advanced Research WRF (ARW; Skamarock et al. 2005) is<br />
a community model suitable for both research and<br />
forecasting.<br />
Three-month-long simulations for the summer season (June-<br />
July-August; JJA) were performed for 22 years from 1982 to<br />
2003. The model grids consist of 109 (west-east) by 86<br />
(north-south) grid lines at 60 km horizontal separation with<br />
the polar stereographic map projection. The model<br />
configurations used for each model are summarized in Table<br />
1. The simulations are performed with reanalyses-derived<br />
boundary forcing.<br />
RegCM3 RSM WRF<br />
vertical σ-23 layers σ-28 layers σ-28 layers<br />
levels<br />
dynamics hydrostatic hydrostatic nonhydrostatic<br />
numerics finite<br />
difference<br />
spectral<br />
computation<br />
finite<br />
difference<br />
CPS<br />
Grell<br />
SAS 2005 Kain-Fritsch<br />
(Byun and Hong, (Kain and Fritsch,<br />
(Grell, 1993)<br />
2007)<br />
1993)<br />
PBL<br />
LSM<br />
RAD<br />
Holtslag<br />
(Holtslag ,1990)<br />
BATS<br />
(Dickinson et al.,<br />
1986)<br />
CCM3<br />
(Kiehl, 1998)<br />
YSUPBL<br />
(Hong et al., 2006)<br />
OSU<br />
(Mahrt and Pan,<br />
1984)<br />
GSFC<br />
(Chou and Suarez,<br />
1999)<br />
YSUPBL<br />
(Hong et al., 2006)<br />
Noah<br />
(Chen and Dudhia,<br />
2001)<br />
simple cloudinteractive<br />
(Dudhia, 1989),<br />
RRTM<br />
(Mlawer et al.,<br />
1997)<br />
3. Dominant interannual variation patterns<br />
Figure 1. The first EOF mode and corresponding<br />
principal component of surface temperature.<br />
Figure 1 shows the first EOF mode of summer<br />
temperature and the corresponding principal component<br />
(PC) reanalysis data and simulated by the RegCM3, RSM,<br />
and WRF, where summer is defined as the average of JJA.<br />
The first dominant mode of the reanalysis explains 36% of<br />
the total variance in the model domain, which is<br />
characterized by positive values over the whole domain.