On the twenty-first-century wet season projections over the Southeastern United States
The RSM is forced to generate the downscaled climate for the twentieth century (hereafter, RSM20th) with output from the CCSM3 historical simulation for the years 1969-1999 (hereafter, CCSM3-20th). The downscaled region is shown in Fig. 1. The time interval of the lateral boundary forcing from the CCSM3 integration is six hours. Similarly, the RSM is forced to generate the downscaled twenty-first-century climate (hereafter, RSM21st) with output from the CCSM3 SRESA2 run (hereafter, CCSM3- 21st) for the years 2040-2070.
Kanamaru and Kanamitsu (2007a) showed that the scale-selective bias correction and interior nudging of small wave numbers used by the RSM make the regional solution insensitive to the size of the regional domain and the location of lateral boundaries. This allows the nesting ratio to be much greater than that typically used in other regional models and eliminates the need for intermediate nesting. Because of this, we are able to downscale CCSM3 simulations and reanalysis data with grid resolutions on the order of 200 km directly to 10 km (Kanamaru and Kanamitsu 2007b; Kanamitsu and Kanamaru 2007; Stefanova et al. 2012). The results compare well with station and gridded observations, as well as with other high-resolution regional reanalyses. Furthermore, Misra et al. (2011a) and Stefanova et al. (2012) showed that the RSM was well suited for depicting the phase and amplitude of diurnal rainfall in the SEUS when compared with observations.
We also use the National Centers for Environmental Prediction (NCEP) Stage IV observed rainfall data available at 4-km grid resolution for the period 2002-2010 (Lin and Mitchell 2005) at hourly intervals for validation. This is a merged analysis of rain gauge and radar-based rainfall estimates available for the continental United States. This data set is used to validate the diurnal variations in the summer season rainfall in RSM20th.