Tuesday, 15 January 2002: 2:44 PM
Simplified Short Term Precipitation Ensemble Forecasts: Theory
The National Weather Service (NWS) has been requested by a variety of users to provide hydrologic forecasts which explicitly account for the uncertainty in the forecast. One major component of the uncertainty in hydrologic forecasts is the uncertainty in the forecast precipitation. Therefore, efforts to assess the uncertainty in a river forecast have focused on assessing the uncertainty in the future precipitation. While a variety of methods exist for quantifying uncertainty, the NWS has generally focused on ensemble methods. Several sophisticated methods have been explored in the past, however none has been found to meet the complex mix of operational and scientific needs that exists in a forecast environment. A new method for generating short term precipitation ensembles has been developed. This method relies on existing NWS data streams of deterministic Quantitative Precipitation Forecast (QPF) and observed precipitation, and is intended to be simple to calibrate. The method requires deriving a forecast distribution for a given deterministic forecast from the joint distribution of historical QPF and observed precipitation. Once the forecast distribution is known, the climatological distribution is mapped into the new forecast distribution to create ensembles of precipitation. A complete description of the new process and a sensitivity analysis of the parameters will be presented.