Improvement on Prediction of Tropical Cyclone ‘Fani’ by Assimilating Quality Controlled Indian DWR Radial Wind: A Case Study

Abstract

Tropical cyclones are among the most destructive natural phenomena and embedded with complex meso-convective systems (MCSs). Better initial conditions are needed for NWP models to accurately predict MCSs associated with the cyclones. The Doppler Weather Radar (DWR) network provides valuable high spatio-temporal data that enhances NWP model forecast accuracy after assimilation, however, observation quality is critical for NWP system performance. In this study, the Python ARM Radar Toolkit (Py-ART) based dealiasing algorithm is applied to improve the quality of radial wind observation before assimilation. The present study evaluates the impact of the assimilation of quality-controlled radial wind in the WRF forecast system on the prediction of extremely severe cyclonic storm ‘Fani’, which hugely affected the eastern coast of India in April 2019. Three sets of assimilation experiments are conducted viz. CNTL: Utilization of conventional observations from GTS; RAD: DWR radial wind plus observation used in the CNTL and RADQC: same observation used in RAD but quality-controlled radial wind observations through Py-ART. Both radar experiments suggest that track prediction and landfall location are more enhanced than CNTL. The statistical analysis shows that compared to the RAD experiment, the assimilation cycle used more radial wind at various stations in the RADQC. The minimum central pressure and wind speed associated with the cyclone are improved by ~ 9–10% in the RADQC. The various statistical measures are considerably improved in the RADQC than RAD and CNTL. The study deduced that incorporating quality-controlled radial wind enhanced the model's forecast skill on cyclone prediction, which can potentially contribute for improving early warning systems and reducing storm impacts.