Seasonal Prediction of Indian Summer Monsoon; A dynamical downscaling perspective.

The importance of rainfall associated with Indian summer monsoon and its large scale dependencies are quite well known to the scientific and economic community. Hence the prediction of Indian summer monsoon rainfall at a lead time of 2-3 months is of great importance. However Indian monsoon is a complex system with its complexities in the form of land surface heterogeneities, intra seasonal variability and mesoscale convective activities. GCMs have improved a lot over time in simulating the seasonal rainfall, onset dates etc., but still lack in a skillful prediction. Downscaling of GCM outputs using a Regional Climate Model can help in representing the land surface and sub-grid scale processes better than the parent model and hence help in improving the predictive skill of the summer monsoon. In this study, we have used a dynamical downscaling method by two different regional climate models, RegCM and WRF, forced by CFSv2 initial and boundary conditions to evaluate the predictive skill of the two regional models. The evaluation have been carried out by comparing the downscaled products with observation/reanalysis data sets as well as the parent model used for driving the model. Both the models capture the seasonal rainfall amount and pattern better than the host GCM but still lack in a skillful prediction when compared to the observed data set. Improvements are observed in simulation of VIMT, CAPE and upper air circulations.