In this study the improved initial conditions through three-dimensional variational data assimilation (3DVAR-DA) technique of Advanced Research of Weather Research and Forecasting (ARW-WRF) model on a single domain with resolutions of 9 km was used to improve rainfall forecasts associated with tropical cyclones (TCs) over the Bay of Bengal (BoB). The observations assimilated in 3DVAR are satellite and conventional observations and initial-boundary conditions for model integration was derived from NCEP. First, two case (Titli and Mandous) studies was selected with and without DA and results suggested that DA experiments provided good results along with IMD best-fit tracks. The results from 14 case studies suggested that model was able to capture the overall spatial distribution, structure, and evolution of rainfall, including light-to-moderate rainfall, but frequently overestimated heavy rainfall intensity compared with GPM observation (Fani, Mocha, Titli, Mandous, Nivar., and Hamoon). Also second day forecasts of rainfall generally showed good agreement with GPM observation. Overall forecast skill of model demonstrated reasonable to strong spatial agreement with GPM, for Mocha showing the highest spatial correlation (0.76) and Fani (0.68) and useful skill in predicting heavy and very heavy rainfall, with POD values 40% and 69% for Bulbul and Hamoon TCs.

Read original article