Evaluation of NCUM-R 4DVAR assimilation technique’s performance on simulation of tropical cyclones over NIO region

Abstract

The present study evaluates the forecasting skill of a regional model (NCUM-R) with the 4DVAR analysis system on the simulation of pre- and post-genesis phases of three tropical cyclones (TCs) over the North Indian Ocean (NIO) region. Specifically, the study aims to determine whether the NCUM-R model can identify pre-genesis of the storm 2–3 days in advance to accurately predict the intensity and track of TCs. The selected storms include the Severe Cyclonic Storm (SCS) ‘Asani’ (2022), the extremely severe cyclonic storm (ESCS) ‘Mocha’ (2023), and ESCS ‘Biparjoy’ (2023). The initial conditions (ICs) are prepared for the NCUM-R forecasting model by assimilating various observations through the 4DVAR data assimilation (DA) technique. Various thermodynamic and dynamic variables, such as genesis potential parameters (GPP), potential vorticity (PV), relative humidity (RH), landfall position and time errors are examined. The study highlights equatorial moisture’s role in storm development. Moist winds from the equator release latent heat, greatly intensifying storms. The study shows that the track error forecast significantly improved by the NCUM-R model about 7%, 11.2%, 22.8% and 21.9% on 00, 24, 48 and 72 UTC forecasts. The landfall position and time error are relatively less in the regional model. Further, the NCUM-R model’s outputs are validated against India Meteorological Department (IMD) observations and Fifth Generation of ECMWF Atmospheric ReanalysisERA-5, demonstrating a good match in TCs’ pattern and intensity. Various statistical skill scores for rainfall are calculated with respect to satellite-merged estimated precipitation data (i.e. GPM) through the enhanced Model Evaluation Tool (METplus), showing a significant correlation with observed data. The model is suggested to capture low-intensity rainfall well, but its forecast skill decreases as rainfall thresholds increase with the forecast day. The study discerns that the NCUM-R model reasonably well forecasts the storm track, intensity, and structure before and after the genesis of the TCs. © The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2025.