High-Resolution Image

NASA's finite-volume General Circulation Model (fvGCM) at 0.625 X 0.5 deg. lat. X long. resolution is used for experimental 10-day forecasts. Components added to a new version of fvGCM, a cloud scheme called McRAS (Sud and Walker, 1999) working in concert with a radiation scheme by Chou et al. (1994, 1998, & 1999), outperform the standard fvGCM in climate simulations (Sud and Walker, 2003). We tested these changes over the continental U.S. using 10-day forecasts for May and June 2003, during which the eastern half of the United States was noticeably wet (main frame). The blue and purple areas received significantly more rainfall during these months than the long-term average for May and June from the Global Precipitation
Climatology Project (GPCP).

The inset panel shows the anomaly correlation for rainfall prediction over the United States for these months with both the standard (red line) and McRAS (green line) versions of fvGCM. The higher value of anomaly correlation implies a better agreement with the daily rainfall estimates from NOAA's Climate Prediction Center (CPC). The "Day 1" value represents the anomaly correlation of the average rainfall from the 1-day advance lead-time forecast for these two months, and so on for "Day 2" through "Day 9". McRAS with new radiation has improved the prediction for nearly all lead times even though the forecasts tend to get poorer with time. The blue line shows a small addition in skill when observed sea surface temperatures (SSTs) are used in place of climatological SSTs. The results of this study will be the
subject of a journal article.

Submitted by Yogesh Sud and David Mocko (SAIC).