Michael Tippett (IRI)
Andrew Robertson (IRI)
Makarand Kulkarni (IIT) 
O.P. Sreejith (IMD)
Palash Sinha (IIT)
Kripan Ghosh (IMD)

Predictability and predictions of Indian summer Monsoon rainfall using
dynamical models

The significant impacts of Indian summer Monsoon rainfall (ISMR) variability on society have motivated the long history of efforts to understand and predict ISMR.  Empirical, statistical, and more recently, dynamical models have been used. We investigate the Indian summer monsoon in a number of dynamical forecast models. The models vary in complexity and include coupled ocean/atmosphere GCMs, an atmosphere GCM coupled to a mixed layer ocean model, and an atmospheric model forced by statistically predicted sea surface temperature ("two-tier"). All the forecasts systems are routinely used to make real-time predictions. Signal-to-noise maximizing EOFs are used to identify the most robust features of the models, that is, the most predictable components in the perfect model sense. The time-series of the predictable components are compared with ENSO and ISMR variability. Potential model skill is compared to the observed forecast skill of the models. Multi-model probabilistic rainfall forecasts constructed from a variety of model combination and correction schemes are evaluated using probabilistic skill scores.