Role of the Indian and Pacific Oceans in the Indian Summer Monsoon Variability

Abstract

The role of the Indian and Pacific sea surface temperature (SST) variability in the intraseasonal and interannual variability of the Indian summer monsoon rainfall is examined by performing a set of regionally coupled experiments with the Climate Forecast System (CFS), the latest and operational coupled general circulation model (CGCM) developed at the National Centers for Environmental Prediction (NCEP). The intraseasonal and interannual variability are studied by isolating oscillatory and persistent signals, respectively, from the unfiltered daily rainfall anomalies using multi-channel singular spectrum analysis (MSSA). This technique identifies nonlinear oscillations, its variance and period without preconditioning the data with a filter and also helps to separate the intraseasonal and low frequency climate signals from the daily variability. It is found that, although the model has large amount of daily variance in rainfall, the combined variance of coherently propagating intraseasonal oscillations is only about 7% while the corresponding number in the observations is 11%. The model has three intraseasonal oscillations with periods around 106, 57 and 30 days. The 106-day mode has a characteristic large-scale pattern extending from the Arabian Sea to the West Pacific with northward and eastward propagations. These features are similar to the northeastward propagating 45-day mode found in the observations except for the longer period. The 57-day mode is more dominant in the region, 60°E-100°E and is strictly northward-propagating. The 30-day mode appears to be equivalent to the northwestward propagating oscillation in the observations. The dominant low frequency persistent signal in the region is due to the El Niño-Southern Oscillation (ENSO). The ENSO-related rainfall anomalies, however fail to penetrate into the Extended Indian Monsoon Rainfall (EIMR) region, and therefore, the ENSO-monsoon relationship in the model is weak. Regionally coupled simulations of the CFS have revealed that the northeastward propagating 106-day mode exists in the model with weak amplitude and reduced variance even when the air-sea interaction over the Indian Ocean is suppressed. However, this mode was not obtained when the Indian Ocean SST variability is reduced to climatology. The spatial structure and propagation of the 106-day mode appear to be unaffected by the Pacific SST variability; i.e., a simulation with climatological SST in the Pacific reproduced this mode. The 30-day northwestward propagating mode showed little change with respect to the Indian Ocean SST, but is dependent on the air-sea interactions over the west Pacific. Simulations using prescribed SST in the Indian Ocean showed that the spatial structure of the ENSO mode in the Indian Ocean is dependent on the air-sea interaction in that region. It is argued that the western Indian Ocean in this model is over-sensitive to atmospheric momentum fluxes and therefore cools down quickly in response to the ENSO-induced circulation anomalies. Further, this process creates a dipole pattern with cool (warm) western and warm (cool) eastern Indian Ocean during a La Niña (El Niño) event. This dipole prevents the ENSO anomalies from reaching the EIMR region and causes the incorrect ENSO-monsoon relationship. It is also found that such a dipole pattern, although with less variance is present even in the absence of the ENSO variability. The monsoon rainfall variability in the absence of the ENSO could be dictated by internal dynamics in this model.