Estimation of the maximum hail size based оn thermodynamic atmosphere parameters using neural network modeling methods

Abstract

Hailstorms cause significant economic damage, especially in the agricultural sector. However, hail forecasting is performed with insufficient accuracy. Determining hail size, which directly affects the magnitude of losses, is also challenging. This study aims to estimate the maximum hail size based on atmospheric meteorological parameters. To this end, we compiled data on maximum hail size recorded by the Stavropol Militarized Hail Suppression Service, together with the corresponding temperature, humidity, wind direction, and wind speed at standard isobaric levels from a global atmospheric model, used as a substitute for upper-air sounding results. The study was based on neural-network modeling methods, with the diameter of the fallen hail as the dependent variable and atmospheric parameters as the independent variables. The SPSS environment automatically selected a neural model consisting of a single layer with four neurons. According to the results, the errors on the training and test samples were identical, indicating the adequacy of the model. Additional quality assessment criteria, such as prediction and residual plots, also confirmed the adequacy of the model. The key atmospheric parameters influencing the maximum hail size were found to be: the George instability index, the temperature at the convection level, the level at which the temperature difference between the cloud and the surrounding environment reaches its maximum, and the mean moisture deficit in the layer above the condensation level at an altitude of 5 km. It was concluded that the proposed neural-network model for estimating hail size can be effectively applied by services engaged in hail suppression.