A regional climate-hydrology modeling system is being used to downscale dynamical seasonal climate forecasts provided by the NCEP global model to produce ensembles of regional climate and streamflow forecasts. To test the performance of the Regional Climate Model (RCM), a simulation has been performed for October 1991 - September 1998. The simulation was driven by large scale conditions from the NCEP/NCAR reanalysis and has been evaluated using a 1/8 degree gridded temperature and precipitation dataset for the eastern U.S. Global simulations have been performed with the NCEP Global Spectral Model (GSM) driven by observed sea surface temperature for October 1996 - May 1997, and October 1997 - May 1998. There are 10 ensemble simulations for each period. These simulations are being used to drive the RCM to examine the value of downscaling in simulating the ENSO signal. Both the GSM and RCM simulations will be used to drive a distributed hydrology model of the Tennessee River to determine how well the modeling system simulates the ENSO streamflow signal. The model simulates streamflow realistically at the Clinch River when driven by observed meteorological data. The model is being extended to the Tennessee River. Simulations will be performed with observed and simulated meteorological data.

An analysis framework has been established to assess the value of improved inflow forecasts in reservoir management. The simple mass-balance model of the Norris Reservoir is combined with historical inflow measurements, the ESP forecasts, and a pareto genetic algorithm. The pareto genetic algorithm defines the tradeoff relationship between two reservoir management objectives by defining pareto optimal "rule curves" for reservoir releases. Reservoir releases are assumed to be a function of the reservoir's current storage and inflow forecasts for multiple lead times. Currently, objectives for baseload hydropower and flood control are being evaluated. The tradeoff curves based on the perfect forecasts, ESP forecasts, and those derived from the coupled regional climate-hydrologic forecast system will provide the upper and lower bounds for the extended forecasts that utilize seasonal climate predictions.