ENHANCING HAIL PREDICTION ACCURACY THROUGH RADAR DATA ASSIMILATION IN ASSOCIATION WITH LARGE-SCALE WEATHER PATTERNS: A CASE STUDY OF SURABAYA, FEBRUARY 21, 2022
Nurjanna Joko Trilaksono (a), Muhaji Sahnita Putri (b)

(a) Atmospheric Science Research Group, Faculty of Earth Sciences and Technology, Bandung Institute of Technology, Bandung, Indonesia

(b) Master Program in Earth Science, Faculty of Earth Sciences and Technology, Bandung Institute of Technology, Bandung, Indonesia

Abstract

Hail is one of the extreme weather phenomena that often occurs in Indonesia and has increased in frequency in recent years. Hailstorms can cause widespread damage to crops, property, infrastructure, and can even pose risks to human safety, with larger hailstones presenting a particularly severe threat. Predicting hail events remains a significant challenge, especially due to limitations in observations and detection methods. However, efforts to improve weather prediction continue, including combining observational data with previous short-term forecasts. Improving models by assimilating radar reflectivity and radial velocity data can enhance the initial conditions of moisture and wind speed, potentially affecting the formation of convective clouds leading to rainfall.

This study utilizes the Weather Research and Forecasting Data Assimilation (WRFDA) model with the 3DVar method. The assimilated data include reflectivity and radial velocity data from the BMKG^s C-Band Weather Radar. Hail in Surabaya is influenced by Large-Scale Weather (LSW) patterns. The results of this study demonstrate that radar data assimilation performs better than without assimilation. Model performance was evaluated using the Fraction Skill Score (FSS) and bias methods. The best FSS result was 0.7 when radar reflectivity data assimilation was applied, while the worst result was 0.38 without assimilation. The best bias value was 0.5 with assimilation, compared to 0.2 without assimilation. The evolution of radar reflectivity values between observations and experiments showed consistency in terms of intensity, timing, and area of hail occurrence when radar data assimilation was applied. The distribution of the Maximum Estimated Size of Hail (MESH) values showed better spatial accuracy of the hail area when using radar data assimilation compared to without assimilation. Specifically, the radar data assimilation experiment for hail in Surabaya was able to estimate MESH reaching 10 mm, with observed MESH values being >10 mm. Hail events associated with LSW can enhance predictive accuracy, particularly within operational forecasting.

Keywords: Hail, Weather Radar, Data Assimilation, Weather Research Forecasting Data Assimilation (WRFDA)

Topic: Atmospheric Sciences