A study on the capability of the NCEP-CFS model in simulating the frequency and intensity of high-intensity rainfall events over Indian region in the high and low resolutions

In the current perception of an increase in extreme precipitation events in a developing and densely populated country like India and the demands of high resolution forecast runs are high, the present study compares the statistical skill of free runs from an operational climate model run in two horizontal resolutions in simulating the frequency and intensity of extreme rainfall events over Indian region. The operational climate model is a version of the National Center for Environmental Prediction (NCEP, USA) Climate Forecast System (CFS) version 2. It is run at two horizontal resolutions: CFST126 and CFST382, which is used for seasonal and extended range prediction in real-time in the India Meteorological Department’s (IMD) operational forecast framework. From the analysis of departure and bias of both the components of the model with respect to India Meteorological Department’s (IMD) observation, it is observed that marked dissimilarity in simulating high-intensity rain events exists between this two versions of the same model which should be hypothetically same. Both the models capture intensity and frequency differently. The main conclusions are (a) CFST126 free run gives better estimates of the frequency of rainfall events compared to CFST382, (b) CFST382 free run gives better estimates of the intensity of rainfall events compared to CFST126. These discrepancies indicate the resolution dependence of the statistics of extreme event, which should be statistically corrected and a multi-resolution ensemble version runs of CFSv2 has to be used for operational outlooks of the extreme events.