Numerical modelling of the Bay of Bengal Tropical Cyclone Characteristics during 2001-20 and the impact of scatterometer winds

Abstract

The paucity of observational data over the North Indian Ocean (NIO) region poses significant challenges in effectively monitoring and forecasting extreme weather events such as tropical cyclones (or TCs). Since direct observational measurements of dynamic and thermodynamic variables at different atmospheric levels are insufficient to provide a deeper understanding of the interacting forces, a numerical modelling-based approach is needed to address the synoptic scale convective processes. The present study aims to investigate the characteristic features of the Bay of Bengal (BOB) TCs through the Weather Research and Forecasting (WRF) modelling framework that includes three-dimensional variational data assimilation (3DVAR). And intends to focus on various ocean, atmosphere, dynamics, thermodynamic, or physical characteristics to understand the TC characteristic differences pertaining to the distinct intensification, genesis locations, and seasonality. In the process, the impact of the assimilation of scatterometer winds is discussed for 36 BOB TCs during 2010-2020. To conduct the assimilation experiment, the TCs are categorised into three classs, i.e., Cyclonic Storm (CS, 34-47 kts), Sever Cyclonic Storm (SCS, 48-63 kts) that includes both SCS and Very Sever Cyclonic Storms (VSCS, 64-119 kts), and Highly Intensified Cyclonic Strom (HICS) including Extremely Severe Cyclonic Storm (ESCS, 90-119 kts) and Super Cyclonic Storm (SuCS, > 120 kts). Two sets of numerical simulations are conducted for each cyclone, i.e., the control or CTRL experiment without the data assimilation and the second one, considering the 3DVAR (DA simulation). CTRL simulations are initialized with NCEP-FNL and NOAA Sea surface temperature (SST) data sets, and DA simulation considers modified initial conditions prepared through the 3DVAR technique, where scatterometer winds are assimilated into the WRF model. Both simulations utilized the same set of physical parametrizations. The results show improvement in case of DA simulations compared to CTRL for different classes of tropical cyclones during the pre-monsoon, post-monsoon seasons, and sectorial analysis ensured through the Root Mean Square Error (RMSE) of minimum sea level pressure (MSLP) and maximum sustained wind (MSW). It was observed that during the genesis and intensification period, the DA simulation results exhibited more accurate estimates compared to CTRL in predicting MSLP and MSW. The wind shear analysis is done to determine the model's performance during the developing and strengthening stages. RMSE of the predicted wind shear from DA and CTRL run is compared to Indian Monsoon Data Assimilation and Analysis reanalysis (IMDAA), and the results reveal that the DA reduces error in the SCS and HICS classes compared during pre-monsoon season. Overall, DA shows improvement in forecast for MSLP by 12%, 17% and 21% for 24, 36 and 48 h compared to CTRL simulation. Similar improvement in the prediction of MSW is also observed with 53%, 37% and 25% in 24, 36 and 48 h of simulation for DA compared to CTRL simulation for all TCs analysed in a composite mode. These results indicate the potential effectiveness of 3DVAR with scatterometer wind data assimilation in improving the prediction accuracy of TC-related parameters, which can enhance the monitoring and forecasting of extreme weather events in the BOB region.