Abstract
In this paper, the impact of the SST drifts in the ECMWF system 3 forecasting system on the simulated monsoon climatology is investigated. It is shown that hindcasts initialized in February show paradoxically better climatological rainfall in the early monsoon season compared to the hindcasts initialized in the May. The differences in rainfall and SST evolution in the two hindcast sets point to the SST differences as the crucial factor that improves the February initialized hindcasts. Further experiments with the atmospheric component of the ECMWF system 3 forecasting system confirm this by showing similar rainfall biases in the early monsoon season as the hindcasts initialized in the May. This study points to the potential beneficial impacts of reducing systematic biases in the atmospheric components forecasting systems, and of an anomaly initialization technique to improve Indian Monsoon forecasts.