Development of Operational Streamflow and Soil Moisture Forecasts for the Southeastern U.S

Skillful seasonal streamflow and soil moisture forecasts offer a great potential in water management, planning of hydroelectric power and agricultural operations. The uncertainty in utility of climate forecasts poses a major challenge in implementation of seasonal streamflow and moisture forecasts in real time operations of water and energy systems. Several studies have shown that multimodel combinations of Atmosphere-Ocean-General Circulation Models perform better than results from a single model. Therefore, the objectives of this study are to: 1) Develop operational streamflow and soil moisture forecasts for the southeastern U.S. by utilizing NASA's Land Information System (LIS) consisting of multiple Land Surface Models (LSMs), and 2) Reduce model uncertainty in monthly to seasonal streamflow forecasts by combining multiple General Circulation Model (GCM) forecasts and several land surface models including Noah, Catchment, and Variable Infiltration Capacity (VIC). First, monthly updated precipitation and temperature forecasts ensembles from multiple GCMs will be combined based on skill levels of candidate models contingent on relevant predictor state. These multimodel forecasts will be statistically downscaled at 0.25° by 0.25° spatial resolution using canonical correlation analysis and then temporally disaggregated at daily time step using K-Nearest-Neighbor (KNN) approach. Subsequently, multimodel forecasts ensemble will be used to implement NASA's LIS which includes widely applicable LSMs such as Noah, Catchment and VIC. Initial daily land surface conditions will be updated (prior to the forecasting period) from LSMs forced with actual NLDAS-2 forcings. Finally, monthly to seasonal streamflow and soil moisture forecasts will be developed for the entire southeastern U.S.