Potential Impact of Climate Forecast Information on Ensemble Streamflow Predictions Made Using Trace Weighting
Abstract
At the National Weather Service River Forecast Centers, Advanced Hydrologic Prediction Services (AHPS) are being implemented for seasonal streamflow forecasting. AHPS uses ensemble streamflow predictions (ESP) to make forecasts of the probability distribution of streamflow variables over a forecast window of up to 90 days. The approach uses historical weather data to simulate streamflow time series (traces) conditioned on the current hydroclimatic state. The ensemble traces are then weighted to produce the probability distribution forecast. In the absence of information on future climate conditions, each trace is treated as an equally likely outcome. However, if climate forecast information is available when the streamflow forecast is made, individual traces can be weighted differently, depending on the correspondence of their weather inputs to forecast climate anomalies. Using a 42-year sample of retrospective forecasts (or hindcasts) for AHPS for a Des Moines River tributary, we examine the potential effect of climate forecast information on forecast quality based on trace weighting. Hypothetical skillful climate forecasts are synthetically generated to study how streamflow forecast quality varies with climate forecast skill. Verification results are presented using both a distribution-oriented and (traditional) summary measures, to illustrate how inferences on the impact of climate forecast information differ by approach.
- Publication:
-
AGU Spring Meeting Abstracts
- Pub Date:
- May 2004
- Bibcode:
- 2004AGUSM.H33A..05B
- Keywords:
-
- 1812 Drought;
- 1833 Hydroclimatology;
- 1860 Runoff and streamflow;
- 1869 Stochastic processes