Predictability of High Resolution Models in the Indian Monsoon Region

The predictability of the ECMWF model with varying horizontal resolution is assessed in predicting the rainfall and circulation over the Indian monsoon region. The model used is the ECMWF ensemble prediction system, and retrospective forecasts are made with coupled model as a part of the MINERVA project. The ensemble of retrospective forecasts by the high-resolution model (16km resolution in the atmospheric model) is compared with those by lower resolution versions (62km and 31km atmospheric models) to understand the impact of increasing the horizontal resolution. Specifically, the mean conditions over the South Asian monsoon region and the predictability of the rainfall and circulation at intraseasonal and seasonal time scales are examined. The models' ability to simulate the period, spatial structure and propagation of the leading intraseasonal oscillations (ISOs) in the monsoon rainfall and circulation is studied. The ability of the models to properly simulate the rainfall over the central India region is examined. The relative roles of the Indian and Pacific oceans in determining the seasonal mean monsoon is also examined. The predictability is also assessed by estimating the growth rate of small errors and the limit of predictability. The predictability study includes both the forecast errors with respect to observations and the predictability errors in a perfect-model assumption. The predictability of the ECMWF model is compared with that of the NCEP Climate Forecast system (CFS version 2).