Simulation of heavy rainfall events over Indian region: a benchmark skill with a GCM

Extreme rainfall events (ERE) contribute a significant component of the Indian summer monsoon rainfall. Thus an important requirement for regional climate simulations is to attain desirable quality and reliability in simulating the extreme rainfall events. While the global circulation model (GCM) with coarse resolution are not preferred for simulation of extreme events, it is expected that the global domain in a GCM would allow better representation of scale interactions, resulting in adequate skill in simulating localized events in spite of lower resolution. At the same time, a GCM with skill in simulation of extreme events will provide a more reliable tool for seamless prediction. The present work provides an assessment of a GCM for simulating 40 ERE that occurred over India during 1998–2013. It is found that, expectedly, the GCM forecasts underestimate the observed (TRMM) rainfall in most cases, but not always. Somewhat surprisingly, the forecasts of location are quite accurate in spite of low resolution (~50 km). An interesting result is that the highest skill of the forecasts is realized at 48 h lead rather than at 24 or 96 h lead. Diagnostics of dynamical fields like convergence shows that the forecasts can capture contrasting features on pre-event, event and post-event days. The forecast configuration used is similar to one that has been used for long-range monsoon forecasting and tropical cyclones in earlier studies; the present results on ERE forecasting, therefore, provide an indication for the potential application of the model for seamless prediction.