HR: 17:15h
AN: H14A-06 INVITED     [Abstracts]
TI: Seasonal Hydrologic Prediction System over the Eastern U.S.
AU: * Wood, E F
EM: efwood@princeton.edu
AF: Princeton University, Dept. Civil adn Envionmental Engineering, Princeton, NJ 08544 United States
AU: Luo, L
EM: lluo@princeton.edu
AF: Princeton University, Dept. Civil adn Envionmental Engineering, Princeton, NJ 08544 United States
AB: The development of seasonal hydrologic prediction systems faces a number of challenges, including the removal of systematic bias in seasonal climate model forecasts, scale inconsistencies between climate and hydrologic models, the merging of multiple seasonal climate forecast information that include multi-model forecasts and other climate information, and creating forecast ensembles that are both skillful and reflective of the total uncertainty in the seasonal forecasts. A seasonal hydrologic ensemble prediction system is being developed and applied over the Eastern U.S. that addresses the above issues. This prediction system primarily utilizes seasonal climate forecasts from the NCEP Climate Forecast System (CFS) but within a multi-model framework is also using forecasts from other seasonal forecast products. For the hydrologic seasonal prediction element, the system currently utilizes the VIC (Variable Infiltration Capacity) hydrological model, which is used to produce ensemble predictions of soil moisture, snow and streamflow with lead times up to 6-month. The system consists of four basic components: (i) a pre-prediction processor that generates the initial condition for the hydrologic prediction; (ii) a hydrologic ensemble processor that performs the model integrations in an ensemble fashion from the initial condition; and (iii) an atmospheric forcing preprocessor that gathers information from multiple forecasting sources and produces an outlook of precipitation and air temperature for the forecast period at a spatial scale that is suitable for hydrologic applications. In this component, a Bayesian approach is implemented to merge multi-model climate prediction into one probability distribution that reflects our best estimate of monthly precipitation and its uncertainties. During the merging process, in effect the dynamic climate model seasonal prediction is bias-corrected and downscaled. Finally, (iv) a product post-processor produces user-friendly predictions based on the ensemble hydrologic prediction. The VIC model is used in both the pre-prediction processor and hydrologic ensemble preprocessor. The seasonal hydrologic ensemble prediction system is being applied over the Eastern U.S., and produces real-time seasonal prediction on a monthly basis for the region. The performance of the system is evaluated over selected basins with hindcasts for the 19-year period (1981-1999), during which seasonal forecasts from NCEP Climate Forecast System (CFS) and ECMWF DEMETER project are available. The evaluation shows that this system and its forecast approach is capable of producing reliable precipitation, soil moisture and streamflow predictions at seasonal timescale. The talk will present a current evaluation and planned improvements to the forecast system.
DE: 1816 Estimation and forecasting
DE: 1847 Modeling
DE: 1880 Water management (6334)
SC: Hydrology [H]
MN: Fall Meeting 2005