Performance evaluation of WRF model with different cumulus parameterizations in forecasting monsoon depressions

Three monsoon depression events of 2011 have been simulated using WRF (ARW) model with five different cumulus parameterizations namely Betts-Miller-Janjic (BMJ), Kain-Fritsch (KF), Grell-3dimensional (G3D), Tiedtke (TDK) and new simplified Arakawa-Schubert (NSAS) schemes.The forecast skills of the model have been verified with observed TRMM-3B42 rainfall analysis.The validation of forecasts is conducted through two approaches.The standard categorical skill scores have used for grid-point by grid-point verification over India domain.The contiguous rain area (CRA) method, one of the object oriented verification techniques in recent times also has been applied for further analysis.The rainfall forecasts of the model performed variedly with different cumulus physics schemes.The comparative performance of 5 schemes through categorical have been analyzed over whole India and seven separated zones as well to capture spatial variation.Using CRA method the rain objects in the observed analysis and model forecasts have been compared in terms of displacement, volume and structure errors.Consequently, the percentage match of rain objects between observation and model forecast has been computed for all cumulus physics separately for comparison.As the displacement error is found to be major contributor, the linear displacements of forecast objects from the location of respective observed objects have also been computed for further clarity.The comparison of verification results for all 5 schemes has been completed separately for all selected weather events to bring a generalized view.