Drought has significant social and economic impacts that could be reduced by preparations made possible through seasonal prediction. During the convective season, when the potential of extreme drought is the highest, the soil moisture can provide a means of improved predictability through land-atmosphere interactions. In the past decade, there has been a significant amount of work aimed at better understanding the predictability of land-atmosphere interactions. One such approach classifies the interactions between the land and the atmosphere into coupling states. The coupling states have been shown to be persistent and were used to demonstrate the existence of strong biases in the coupling of the NCEP Climate Forecast System, version 2 (CFSv2). In this work, the attribution of the coupling state on the seasonal prediction of precipitation and temperature and the extent to which the bias in the coupling state hinders the prediction of drought is analyzed. This analysis combines the predictions from statistical models with the predictions from CFSv2 as a means to isolate and attribute the predictability. The results indicate that the intermountain region is a hotspot for seasonal prediction because of local persistence of initial conditions. In addition, the local persistence of initial conditions provides some level of drought prediction; however, accounting for the spatial interactions provides a more complete prediction. Furthermore, the statistical models provide more skillful predictions of precipitation during drought than the CFSv2; however, the CFSv2 predictions are more skillful for daily maximum temperature during drought. The implication, limitations, and extensions of this work are also discussed.
All Science Journal Classification (ASJC) codes
- Atmospheric Science
- Atmosphere-land interaction
- Numerical weather prediction/forecasting
- Seasonal forecasting