J15.3
Seasonal numerical forecasts of the Ganges and Brahmaputra river flow

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Tuesday, 19 January 2010: 4:30 PM
B212 (GWCC)
Peter J. Webster, Georgia Institute of Technology, Atlanta, GA; and J. Jian, H. R. Chang, and T. M. Hopson

Earlier diagnostic studies have indicated that there is predictability in the Ganges and the Brahmaputra, especially in the former basin, related to slowly evolving patterns of sea-surface temperature in the Pacific and Indian Ocean. With these clues, we employ the ECMWF syetem-3 couple ocean-atmosphere climate prediction model to forecast river discharge into Bangladesh on 1-6 month time scales. Inaugurated in 2007, System-3 possesses 41 ensemble members of 7 months duration made every month. Post-processing is applied to the ECMWF model using quantile-to-quantile techniques and the model precipitation is used to run the hydrological module. Based on the ensemble simulation of the monthly mean Ganges/Brahmaputra flows, statistical downscaling technique is employed to manipulate probability function of the extreme events as flooding.

The model predicted well the late season Brahmaputra flooding in 2007 and 2008. Furthermore, the 2009 forecasts have correctly anticipated low level discharge in both river basins. Specifically, the system managed to predict the late onset of the Ganges flow and the wide-spread drought season as early as from April and May. To examine further the skill in the model, we use hindcasts of the System-3, possessing 11 ensemble members per month, for the period 1980-2006. Overall the forecast scheme contains more predictability for the Ganges River. We examined, in particular, the forecasts for the year of the great Bangladesh flooding in 1998. Encouragingly, a scheme based on forecasted SSTs scheme forecast shows strong predictability for both rivers in time and magnitude. It is important to note that System-3 predicted the exceptional Bay of Bengal SST warming, suggested by other studies to be the cause of the great flood, six months in advance.