Assessing Relative Sources of Streamflow Prediction Uncertainty across the Contiguous United States

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Monday, 3 February 2014: 11:30 AM
Room C209 (The Georgia World Congress Center )
Levi D. Brekke, U.S. Bureau of Reclamation, Denver, CO; and M. Clark, A. W. Wood, A. J. Newman, K. Sampson, T. Hopson, J. R. Arnold, and D. Raff

National Center for Atmospheric Research - Research Applications Laboratory, Department of Interior's Bureau of Reclamation (Reclamation), and the U.S. Army Corps of Engineers (USACE) are partnering to understand the impact of different sources of short-term forecast uncertainty on water management decisions.

Operational approaches to streamflow forecasting typically involve (i) running a hydrologic model up to the start of the forecast period to estimate basin initial conditions; and (ii) running the model into the future with weather forecasts and climate outlooks. Forecast skill depends on (i) the accuracy of the basin initial condition estimates and their impact on the basin response; (ii) the accuracy of the weather and climate forecasts, and (iii) the capability of the hydrologic model to accurately portray hydrologic processes. A key shortcoming of this approach is that it only accounts for uncertainty in weather and climate forecasts, and neglects uncertainty in the estimates of the basin initial conditions and uncertainty in the hydrologic model. Several studies have focused on the relative importance of different sources of uncertainty on hydrologic prediction. There have also been efforts to improve estimates of and reduce forecast uncertainty (e.g., methods for hydrologic data assimilation, probabilistic quantitative precipitation estimation, and post-processing of forecast ensembles). We, as a community, thus recognize there are different opportunities to evaluate and reduce uncertainties. However, we have yet to comprehensively evaluate the relative importance of uncertainties in basin initial conditions, hydrologic model structure, data assimilation choices, weather and climate forecasts, forecast post-processing techniques, and how these uncertainties depend on the type of forecast. Determining this relative importance may help identify where resources may be best allocated to reduce uncertainties in the most efficient manner in the future.

The purpose of this study is to undertake a comprehensive predictability assessment to quantify and document the relative importance of the myriad of uncertainties in hydrologic monitoring and prediction products. The overall goal of the proposed project is to quantify the impact of different sources of uncertainty on different types of forecasts (e.g., 1-day stage forecasts, 3 month volume forecasts), at different forecast initialization times throughout the year (e.g., forecasts initialized on October 1st versus April 1st), and in different hydroclimate regions (e.g., regions with/without substantial snow storage; regions with varying degrees of climate predictability).

This presentation will focus on experimental design and highlight findings from year one of this three-year effort. This study is expected to indicate opportunities to improve hydrologic prediction products, and help identify and prioritize future research projects.