Impact of convective parameterization on the seasonal prediction skill of Indian summer monsoon

The sensitivity of seasonal predictions of the Indian summer monsoon (ISM) to convection parameterization schemes (CPS) is studied using 37 years of hindcast experiments. The predictions are quite sensitive to changes in these schemes and improve the skill by 18-28%. Though the mean state circulation and rainfall over India improves, the sea surface temperature (SST) biases increase in the sensitivity experiments compared to the control run. The ability of the model to realistically capture the teleconnections associated with monsoon such as the El-Nino Southern Oscillation (ENSO) and Indian Ocean Dipole (IOD) also appears to change with different CPS. It is found that the suitability of a CPS for ISM in the Climate Forecast System version 2 (CFSv2) stems from its ability to capture cloud fractions realistically and keep the SST biases to a minimum. The revised Simplified-Arakawa-Schubert (SAS2, Han and Pan in Weather Forecast 26:520-533. 10.1175/waf-d-10-05038.1, 2011) scheme gives better prediction skill for ISM compared to the skill score obtained from SAS2 with shallow convection (SAS2sc) primarily because it simulates realistic clouds, without aggravating the SST biases, particularly in the tropical Pacific Ocean, and captures the Indian Ocean teleconnections realistically. SAS2sc significantly under-estimates the low-level clouds over global equatorial region, despite simulating better mid and high-level clouds, higher Nino 3.4 skill, and better inter-annual variability of ISM. The cold SST bias in the tropical basins is large in SAS2sc. Therefore, to exploit the merits of SAS2sc, unrealistic suppression of low clouds needs to be addressed, and the cold SST biases need to be minimized.