Determining the sensitive parameters of WRF model for the prediction of tropical cyclones in the Bay of Bengal using Global Sensitivity Analysis and Machine Learning

Abstract

Abstract. The present study focuses on identifying the parameters from the Weather Research and Forecasting (WRF) model that strongly influence the prediction of tropical cyclones over the Bay of Bengal (BoB) region. Three global sensitivity analysis (SA) methods, namely the Morris One-at-A-Time (MOAT), Multivariate Adaptive Regression Splines (MARS), and surrogate-based Sobol' are employed to identify the most sensitive parameters out of 24 tunable parameters corresponding to seven parameterization schemes of the WRF model. Ten tropical cyclones across different categories, such as cyclonic storms, severe cyclonic storms, and very severe cyclonic storms over BoB between 2011 and 2018, are selected in this study. The sensitivity scores of 24 parameters are evaluated for eight meteorological variables. The parameter sensitivity results are consistent across three SA methods for all the variables, and 8 out of the 24 parameters contribute 80 %–90 % to the overall sensitivity scores. It is found that the Sobol' method with Gaussian progress regression as a surrogate model can produce reliable sensitivity results when the available samples exceed 200. The parameters with which the model simulations have the least RMSE values when compared with the observations are considered as the optimal parameters. Comparing observations and model simulations with the default and optimal parameters shows that predictions with the optimal set of parameters yield a 16.74 % improvement in the 10 m wind speed, 3.13 % in surface air temperature, 0.73 % in surface air pressure, and 9.18 % in precipitation predictions compared to the default set of parameters.

The Regional Coupled Suite (RCS-IND1): application of a flexible regional coupled modelling framework to the Indian region at km-scale

“…It will therefore be important to re-examine these and other cyclone cases using revised RAL physics definitions. For example, Baki et al (2021) found simulation of TCs in Bay of…”

WRF model parameter calibration to improve the prediction of tropical cyclones over the Bay of Bengal using Machine Learning-based Multiobjective Optimization

“…The WRF model version 3.9.1 (Skamarock et al 2008) The WRF model represents the unresolved physical processes through seven parameterization schemes: surface layer physics, land surface physics, shortwave radiation, longwave radiation, planetary boundary layer physics, microphysics, and cumulus physics. The parameterization schemes used in the present work are selected based on the studies of Baki et al (2021b), which are: MM5 similarity scheme (Beljaars 1995) for surface layer physics, Unified Noah land surface model (Mukul Tewari et al 2004) for land surface physics, Dudhia shortwave scheme (Dudhia 1989) for shortwave radiation, rapid radiative transfer model (Mlawer et al 1997) for longwave radiation, WRF Single-Moment 6-class (WSM6) scheme (Hong and Lim 2006) for microphysics, Kain-Fritsch (Kain 2004) for cumulus physics, and Yonsei University Scheme (YSU) for planetary boundary layer physics. The parameterization schemes contain a vast number of tunable parameters, and calibrating all of them is an impossible task.…”

“…The parameterization schemes contain a vast number of tunable parameters, and calibrating all of them is an impossible task. Baki et al (2021b) conducted a sensitivity study of 24 parameters from the same physics schemes for tropical cyclones over the Bay of Bengal and reported that a total of eight parameters were found to be most sensitive. Thus, these parameters are selected for calibration to improve the prediction of tropical cyclones over the BoB.…”

“…Thus, there is a need to identify the most sensitive parameters which greatly influence the fundamental meteorological variables. Several researchers have evaluated the sensitivity of the WRF model parameters using qualitative and quantitative methods (Wang et al 2013;Green and Zhang 2014;Quan et al 2016;Di et al 2017;Houle et al 2017;Ji et al 2018;Wang et al 2020;Chinta et al 2021;Baki et al 2021b).…”

“…In view of this, the present study focuses on improving the prediction of tropical cyclones across different intensity categories by employing parameter calibration. The model parameters utilized in the present study are obtained based on a previous investigation conducted by Baki et al (2021b), where three different sensitivity analysis methods, namely Morris One-At-a-Time (MOAT), Multivariate Adaptive Regression Splines (MARS), and Gaussian Process Regression (GPR) based Sobol' were applied to screen the most influential parameters for the simulations of tropical cyclones over the Bay of Bengal region. The selected parameters are tuned in this study by minimizing the error between the model simulations and the observations of precipitation and 2m wind speed for the simulations of ten tropical cyclones by employing the MO-ASMO method.…”