Analysis of updraft velocity in mesoscale convective systems using satellite and WRF model simulations

Updraft vertical velocity is an important dynamical quantity which is strongly related to storm intensity and heavy precipitation. It can be calculated by direct observations, NWP model, and geostationary satellites which can provide the possibility of measuring this quantity with high temporal resolution. This research analyzed updraft velocity based on six derived parameters from INSAT3-D and high temporal and spatial resolution simulations of WRF model in the west and southwest of Iran. The interrelationship among the derived variables was investigated from the immature to mature stages of convective cells in Mesoscale Convective Systems (MCS). Updraft velocity was calculated based on a theoretical framework and real observations. The was a large results discrepancy among the results. This finding was in company with previous studies which concluded that updraft velocity is the resultant of other bulk buoyancy forces and environmental variables. Also, the estimated updraft velocities showed a positive correlation with height. The authors proposed linear regression, as a parametric, and Random Forest (RF), as a non-parametric, machine learning methods for estimation of updraft velocity based on satellite variables. A forward–backward method was applied to reach the best modeling in both methods. In linear regression modeling, the cloud-top cooling rate was the most significant factor, and in the RF, band difference of water vapor, thermal infrared 1, and elevation data had the maximum importance. Results showed that the RF could better estimate updraft velocity.