Seasonal prediction of South Asian monsoon in CFSv2

Abstract

The predictions of the seasonal mean monsoon over South Asia by a coupled model are assessed by analyzing the retrospective forecasts by the Climate Forecast System version 2 of the National Centers of Environmental Prediction for the period 1982–2009. The predictability is assessed by examining the June–September seasonal mean of the outgoing longwave radiation in forecasts from May initial conditions. Although the forecasts exhibit considerable interannual variability, the correlation with the observation is low. While the model is able to predict weak (strong) seasonal monsoon during certain El Nino (La Nina) years, with differences in magnitude, the predictions in the years of Indian Ocean Dipole (IOD) events are not so successful. The relative role of the Pacific and Indian oceans in determining the seasonal mean monsoon is analyzed by extracting the El Nino-Southern Oscillation (ENSO) mode and IOD mode from the forecasts and observation. The model’s ENSO mode captures the observed spatial structure whereas its IOD mode shows incorrect anomalies over India. The model is inadequate in predicting the relative strength and signs of the ENSO and IOD modes. The ocean–atmosphere interaction of the ENSO mode is better represented in the model by capturing the observed relation with the sea surface temperature (SST) and the low-level winds. In the model’s IOD mode, the relation with the SST and the low-level winds is not correct over part of the Indian region while the anomalies over the Pacific Ocean also seem to influence, unlike in the observation.