Some of the material in is restricted to members of the community. By logging in, you may be able to gain additional access to certain collections or items. If you have questions about access or logging in, please use the form on the Contact Page.
This study examines the predictability of the Florida State University/Center for Oceanic and Atmospheric Prediction Studies (FSU/COAPS) climate model, and is motivated by the model's potential use in crop modeling. The study also compares real-time ensemble runs (created using persisted SST anomalies) to hindcast ensemble runs (created using weekly updated SST) to asses the effect of SST anomalies on forecast error. Wintertime (DJF, 2 month lead time) surface temperature and precipitation forecasts over the southeastern United States (Georgia, Alabama, and Florida) are evaluated because of the documented links between tropical Pacific SST anomalies and climate in the southeastern United States during the winter season. The global spectral model (GSM) runs at a T63 resolution and then is dynamically downscaled to a 20 x 20 km grid over the southeastern United States using the FSU regional spectral model (RSM). Seasonal, monthly, and daily events from the October 2004 and 2005 model runs are assessed. Seasonal (DJF) plots of real-time forecasts indicate the model is capable of predicting wintertime maximum and minimum temperatures over the southeastern United States. The October 2004 and 2005 real-time model runs both produce temperature forecasts with anomaly errors below 3°C, correlations close to one, and standard deviations similar to observations. Real-time precipitation forecasts are inconsistent. Error in the percent of normal precipitation vary from greater than 100% in the 2004/2005 forecasts to less than 35% error in the 2005/2006 forecasts. Comparing hindcast runs to real-time runs reveals some skill is lost in precipitation forecasts when using a method of SST anomaly persistence if the SST anomalies in the equatorial Pacific change early in the forecast period, as they did for the October 2004 model runs. Further analysis involving monthly and daily model data as well as Brier scores (BS), relative operating characteristics (ROC), and equitable threat scores (ETS), are also examined to confirm these results.