Impact of Time Step Size on Different Cumulus Parameterization Schemes in the Numerical Simulation of a Heavy Rainfall Event Over Tamil Nadu, India

The frequency of extreme rainfall events in India has increased substantially during the recent decades, causing increased risk of inland flooding. This study investigated the impact and role of four different cumulus parameterization schemes (CPSs) in simulating heavy rainfall events (HRE) over Tamil Nadu State that occurred on 01 December 2015 with five different time step sizes (TSSs) using the Weather Research and Forecasting (WRF) model. Numerical simulations were carried out using a 3-km horizontal grid resolution with relevant forcing parameters from NCEP Global Forecasting System (GFS) datasets. The rainfall data were obtained from the Chennai Doppler Weather Radar station and two automated weather stations (AWS) located at the National Centre for Sustainable Coastal Management (NCSCM 13.01 degrees N, 80.23 degrees E) and Madhavaram AWS (13.2 degrees N, 80.2 degrees E), respectively. The results indicate that simulated HRE using the WRF model is more sensitive to the choice of CPS at different TSS. It is seen that with an increase in TSS, the Kain-Fritsch (KF) scheme over-predicts HRE, while the Grell-Freitas (GF) convection scheme provides reasonably better prediction at 20 s TSS. The results clearly demonstrate that TSS is more sensitive to HRE forecasts in the WRF model, and the corresponding probability of detection (POD), critical success index (CSI) and bias varied from 0 to 81% as TSS increased from 6.67 to 20 s. Also, this study clearly reveals that TSS is very sensitive to the model CPSs in HRE forecasting. Variations attributed to TSS and CPS are reflected not only in rainfall magnitude but also in terms of rainfall location pattern, and depend on changes in simulated distribution of hydrometers, rate of latent heat, vertical velocity and wind flow patterns. Statistical analysis was also carried out to verify model-computed HRE forecasts against the data from two automatic weather stations (AWS). This study indicates that the correlation coefficient (CC) was very good, while the root mean square error (RMSE) was smaller for simulations that used high TSS with simplified Arakawa-Schubert (SAS) and GF CPSs. Numerical simulations that employed KF with high TSS showed a higher intensity in the rainfall forecast.