Comparison of MMCFS and SINTEX‐F2 for seasonal prediction of Indian summer monsoon rainfall

Abstract

AbstractA comparison of the Indian summer Monsoon Rainfall (ISMR) in two different coupled models (viz., Scale Interaction Experiment‐Frontier‐F2; SINTEX‐F2, and Monsoon Mission Climate Forecast System; MMCFS) is carried out to ascertain the predictability sources in these models and their strengths and weaknesses. SINTEX‐F2 has a stronger cold sea surface temperature (SST) bias in the central equatorial Pacific, and it simulates mean ISMR better while underestimating the interannual variability of ISMR. On the other hand, MMCFS has warmer SST bias in the tropical Pacific off the equator, simulates a drier mean monsoon, but has a more realistic ISMR interannual variability. Further, the cold SST bias in the central tropical Pacific adversely affects the ability of the SINTEX‐F2 to capture the El Niño–Southern Oscillation (ENSO) related interannual variability and teleconnection patterns with monsoon by shifting it further westward in the tropical Pacific Ocean. The models' skill in simulating various climate indices are compared by considering criteria such as anomaly correlation coefficient (ACC), spread to root mean square error (RMSE) ratio, and signal to noise ratio (SNR). The prediction skill for ISMR in terms of ACC in MMCFS (.53) and SINTEX‐F2 (.45) are comparable for hindcasts initialized in February month. However, RMSE for ISMR from February initial conditions in SINTEX‐F2 (1.43 mm/day) is small compared to MMCFS (2.34 mm/day). A simple multi‐model ensemble prediction system based on MMCFS and SINTEX‐F2 results in better prediction skill in terms of ACC for tropical SST and ISMR. The spread/RMSE ratio for ISMR is similar in both the models but is better for ENSO indices in MMCFS at a longer lead time than in SINTEX‐F2. Considering the SNR as performance criteria, MMCFS has an advantage due to better predictability of SST and vertical wind shear in several parts of the Indo‐Pacific domain.