Hybrid assimilation on a parameter-calibrated model to improve the prediction of heavy rainfall events during the Indian summer monsoon

Abstract

Heavy rainfall events during the Indian summer mon-soon cause landslides and flash floods resulting in a significant loss of life and property every year. The ex-actness of the model physics representation and initial conditions is critical for accurately predicting these events using a numerical weather model. The values of parameters in the physics schemes influence the accu-racy of model prediction; hence, these parameters are calibrated with respect to observation data. The present study examines the influence of hybrid data assimilation on a parameter-calibrated WRF model. Twelve events during the period 2018–2020 were simulated in this study. Hybrid assimilation on the WRF model signi-ficantly reduced the model prediction error of the varia-bles: rainfall (18.04%), surface air temperature (7.91%), surface air pressure (5.90%) and wind speed at 10 m (27.65%) compared to simulations with default para-meters without assimilation