|Van Liew, Michael|
|Lall, Upmanu - COLUMBIA UNIVERSITY|
Submitted to: Environmental and Water Resources Institute World Congress Proceedings
Publication Type: Proceedings
Publication Acceptance Date: February 12, 2004
Publication Date: June 27, 2004
Citation: Garbrecht, J.D., Schneider, J.M., Van Liew, M.W., Lall, U. 2004. Utility of streamflow forecasts derived from seasonal precipitation forecasts. In: Proceedings of the Environmental and Water Resources Institute World Congress, June 27-July 1, 2004, Salt Lake City, UT. 2004 CDROM. Interpretive Summary: In this exploratory study the size of the streamflow response to a given precipitation forecast is examined to establish the conditions under which a precipitation forecast is likely to translate into useful streamflow information. This is a critical step to promote a better understanding and wider acceptance and utilization of NOAA's forecasts in the water resources user community. Rainfall-runoff computer simulation was used to determine streamflow response to several hypothetical one-month precipitation forecasts under alternatively saturated, average and dry soil conditions in a small experimental watershed. The findings suggested that precipitation forecast, in combination with initial soil saturation, produce a range of streamflow responses that is broad enough to provide useful decisions information for water resources management. It was also found that expressions of likelihood of streamflow to exceed a specified threshold provided the best format for decision making. Based on the success of this focused study, it was recommended that this investigation be expanded to include larger watersheds and persistent forecasts spanning several months.
Technical Abstract: In this study, the magnitude of streamflow response to a precipitation forecast is examined to establish the conditions under which a probabilistic precipitation forecast, if skillful, is likely to translate into useful streamflow information. The hydrologic model SWAT was used to simulate streamflow response to several hypothetical precipitation forecasts and antecedent hydrologic conditions. The test watershed was a 32.9 km2 subwatershed in the Little Washita River watershed in central Oklahoma. The streamflow characteristics were expressed in terms of runoff volume, change in runoff volume, odds of exceeding a specified threshold volume, and probability of exceedance curves of volume. While the numerical findings of this study are specific to the climatic and physiographic conditions of the watershed, the findings suggest that combinations of antecedent hydrologic and precipitation forecast conditions produce a broad range of streamflow responses that are suitable for decision making. It was also found that strong wet/dry antecedent hydrologic conditions can dominate over the impact of typically modest precipitation forecasts. In addition, dry antecedent hydrologic conditions tend to limit the potential utility of wet/dry forecasts, because any forecasted precipitation is mostly absorbed in the soil and produces little immediate impact on streamflow, though it may influence the antecedent conditions for subsequent precipitation-runoff events (delayed impact). Finally, streamflow response to a forecast is best expressed in probabilistic terms, and decisions framed in terms of change in odds of success or failure for key indicators. Recommendations are also given for further research.