Extending operational probabilistic forecasts of Ganges and Brahmaputra river flow to 15-day and monthly time scales

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Tuesday, 19 January 2010: 11:30 AM
B215 (GWCC)
Jun Jian, Georgia Institute of Technology, Atlanta, GA; and P. J. Webster, T. M. Hopson, H. R. Chang, and C. D. Hoyos

In 2007, our new probabilistic 1-10 day forecast system of Ganges and Brahmaputra flow into Bangladesh was introduced operationally. The system uses ECMWF ensemble forecasts that are statistically rendered and used to drive the basin-scale lumped and distributed models. At the 5-day forecast horizon, correlations between the ensemble mean and observations had verifying correlation of about 0.9 and Brier scores of 0.03. At 10 days these statistics were 0.8 and 0.04. The system showed high risk of flooding 10 days in advance of three major flooding events along the Brahmaputra allowing mass evacuation and actions to reduce the social and economical impacts of the flooding.

The system has now been extended to 15 days and 31 days taking advantage of the full 1-15 day ensemble runs of ECMWF EPS and the new 51-ensemble ECMWF forecasts that are run each week that incorporate, after 15 days, a full coupled system. At 15 days, the forecasts still verify strongly. The monthly forecasts, starting in 2009 are encouraging and show strong predictability of Brahmaputra river flow accounting for the intra-seasonal component.