Oceanic factors controlling the Indian summer monsoon onset in a coupled model

Despite huge socio-economical impacts, the predictability of the Indian summer monsoon (ISM) onset remains drastically limited by the inability of both current forced and coupled models to reproduce a realistic monsoon seasonal cycle. In the SINTEX-F2 coupled model, the mean ISM onset estimated with rainfall or thermo-dynamical indices is delayed by approximately 13 days, but it occurs 6 days early in the atmosphere-only component of the coupled model. This 19 days lag between atmospheric-only and coupled runs, which is well above the observed standard-deviation of the ISM onset (10 days in the observations), suggests a crucial role of the coupling, including Sea Surface Temperatures (SST) biases, on the delayed mean onset in the coupled model. On the other hand, the key-factors governing the interannual variability of the ISM onset date are also fundamentally different in the atmospheric and coupled experiments and highlight the importance of El Nio-Southern Oscillation (ENSO) and ocean-atmosphere coupling for a realistic simulation of the variability of the ISM onset date. At both interannual and seasonal timescales, we demonstrate the importance of the meridional gradients of tropospheric temperature, moisture and vertical shear of zonal wind in the Indian Ocean for a realistic ISM onset simulation. Taking into account that the tropical tropospheric temperature and the vertical shear are not only controlled by local processes, but also by large-scale processes, we need to examine not only the Indian Ocean SST biases, but also those in others tropical basins in order to understand the delay of the mean onset date in the coupled model. During April and May, the main tropical SST biases in the coupled model are a strong warm bias in the Indian, Pacific and Atlantic Oceans, associated with an important excess of equatorial precipitations, and thus a warmer equatorial free troposphere. In order to identify the keys tropical SST regions influencing the mean ISM onset date, sensitivity coupled experiments have been performed. In these experiments, the SST is corrected separately in each tropical basin. The correction of SST biases in the tropical Indian and Atlantic oceans only slightly improves the onset date in the coupled model and produces "El Nio-like" changes in the tropical Pacific. Conversely, the correction of the Pacific SST biases advances the onset date by 9 days compared to the control coupled run. These results suggest that, while the correction of Indian SST biases improves the rainfall spatial distribution, the delayed mean ISM onset date is mainly control by the tropical Pacific Ocean SST biases.