J2.3
Verification of ENSO-weighted long-range ensemble streamflow forecasts in the Blue Nile River

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Wednesday, 26 January 2011: 11:00 AM
Verification of ENSO-weighted long-range ensemble streamflow forecasts in the Blue Nile River
611 (Washington State Convention Center)
Mohamed A. Habib, University of Iowa, Iowa City, IA; and A. A. Bradley, M. Elshamy, and D. Amin
Manuscript (354.0 kB)

Ensemble streamflow forecasting of major rivers around the globe has been of increasing interest to researchers and decision makers for the last two decades. Often, hydrologic model initial conditions, describing the current land surface moisture state and river conditions, provide sufficient long-range predictability of future streamflow. Long-range predictability may be enhanced by exploiting the relationship between large-scale climate variations and regional hydrologic conditions. This presentation will evaluate contributions of initial conditions and climate information in ensemble forecasting skill for the Blue Nile River. The Nile Forecast System (NFS) is used to produce long-range ensemble streamflow forecasts for locations on the Blue Nile. Two sets of model states are used as initial conditions for all the forecasts: simulated flow, which is the output of a well calibrated model, and assimilated flow, in which simulated flow is modified to match the observed flow. Forecast periods examined are out to one year. Retrospective forecasting (or hindcasting) is used to generate forecasts for the last 18 years (1992-2010), using approximately 70 years of observed meteorological conditions (each year used to generate an ensemble member), for the chosen forecast locations. To examine the potential contribution of climate information, established linkages between ENSO and Nile River flows are used to selectively weight ensemble members. Using an ENSO index for the forecast year, ensemble members are weighted based on the ENSO index for the year corresponding to the weather inputs. Diagnostic verification measures are then used to assess the effects of initial model states and data assimilation, and the use of ENSO climate information, on the skill of long-range flow forecasts for the flood season.