Predictability of the Coupled Monsoon System:
Applications to the Forecasting of Bangladesh Floods

Peter J. Webster

Earth and Atmospheric Sciences
Georgia Institute of Technology

The monsoon is described within the context of a coupled ocean-atmosphere-land system that is regulated through negative feedbacks between components of the system. This modulation leads to the minimization of variability on seasonal and interannual time scales and explains why the monsoon rarely exhibits multi-year periods of extreme drought or flood.

We use this knowledge to attempt forecasts of floods in Bangladesh which is the confluence region of the large Ganges and Brahmaputra river systems. Examples of forecasts for short (1-6 days), medium (20-25 days) and long term (1-6 months are shown. The first and last periods use ensemble forecasts whereas the medium range uses new statistical techniques. Considerable skill is found at all three time scales, Finally, we consider methods of communicating probabilistic forecasts to a user community through the generation of user metrics.