Simulation of Monsoon Rainfall over Bangladesh using High Resolution WRF-ARW Model

Abstract

In the present study, the Weather Research and Forecast (WRF-ARW V3.5.1) model have een used to simulate the station wise monsoon rainfall during 2010–2014 over Bangladesh. The initial and boundary conditions are drawn from the global operational analysis and forecast products of National Center for Environmental Prediction (NCEP-GFS) available at 1°×1o resolution. The model was configured in single domain, 6 km horizontal grid spacing with 161×183 grids in the east-west and north-south directions and 28 vertical levels. For the simulation of monsoon rainfall WSM6-class graupel scheme coupled with Kain-Fritsch (KF) cumulus parameterization (CP) scheme has been used. Initially the model was run 137 days for long term prediction starting with the initial condition of 0000 UTC of 17 May up to 0000 UTC of 1 October for the period 2010-2014. The model was also run 72 hours with every day 0000 UTC initial conditions for 124 days for the prediction of 24, 48 and 72 hours lead time rainfall in the monsoon season of 2014. In this research convective and non-convective rainfall have been simulated at 3 hourly interval then made daily and monthly total rainfall data for 24, 48, 72 hour and long term during the studied period. We have compared this data with the observed rainfall at 33 meteorological stations of BMD and TRMM rainfall. It has shown that the performance of the model for 24, 48 and 72 hours predictions are reasonably well except northeast and southeast hilly regions. The distribution of 24 hours lead time predicted rainfall is almost similar all over the country except the higher rainfall area where model simulated rainfall is much more. The prediction deteriorates as the prediction time increases. The long term predictions of simulated rainfall are not matched with BMD observed rainfall. The pattern of TRMM and observed rainfall are almost similar but the value of TRMM observations is much lower than that of observed rainfall. The RMSE shows that the value for 24 hour prediction lies within 15-30 mm range except hilly regions. The MAE shows that the value for 24 hour prediction lies within 10-20 mm range except hilly regions. It has also been observed that where the rainfall has minimum the RMSE and MAE have also minimum.