El Niño and Indian summer monsoon rainfall relationship in retrospective seasonal prediction runs: experiments with coupled global climate models and MMEs

The relationship between the warm phase of El Niño southern oscillation (ENSO) and Indian summer monsoon rainfall is explored through seven coupled global climate models (CGCMs), which are semi-operational at APEC Climate Center (APCC). The 23-year (1983–2005) hindcast datasets of individual model ensembles derived from May initial conditions for southwest monsoon season (JJAS) are utilized to find out the simultaneous influence of El Niño–ISMR relationship in 1990s, which is observed to be weaker than present decades. The hindcast of ISMR climatology derived from seven individual models viz. APCC, NCEP, POAMA, SINT, SUT1, PNU and UHT1 appears to be reasonably simulated; in particular, about 50 % of mean departure is evident in most CGCMs. In addition, four of six El Niño years during the aforementioned period are well depicted in most of the CGCMs, while the years 1994 and 1997 are not represented well by these seven individual models. The warm SST anomaly aligned with surplus precipitation over tropical equatorial Pacific region simulated using APCC, NCEP, POAMA, SINT and SUT1 is relatively better than that simulated in PNU and UHT1 and it is closer to observation. The El Niño–ISMR teleconnection skills both monthly to seasonal scale are very poor in PNU as well as UHT1 and their RMSEs are 3.84 and 3.77 higher than APCC, NCEP, POAMA, SINT and SUT1 models. The authors developed two Multi-Model Ensembles (MMEs) that were simple composites of ensemble forecast from seven models (APCC, NCEP, POAMA, SINT, SUT1, PNU and UHT1) referred to as MME1, and from five models (APCC, NCEP, POAMA, SINT and SUT1) are referred to as MME2. Importantly, the one-month lead MME2 prediction of anomaly correlation coefficient (ACC) and its adverse impacts is reasonably better than MME1 prediction. However, there are some limitations in capturing SST forcing fields over Indian Ocean region in both MMEs. Among the seven models, SINT has the highest pattern correlation of precipitation over the Indian monsoon region.