Assessing the accuracy of different Z-R relationships for Doppler Weather Radar based rainfall estimation: A comparative study for the Delhi region

Abstract

Z-R relationships are the most used methods for calculating rainfall using radar reflectivity, which provide a relation between the radar reflectivity (Z) to rainfall rate (R). However, different Z-R relationships may yield varying rainfall estimates depending on regional climatic conditions and radar characteristics. This study presents a comparison of various Z-R relationships (Marshall-Palmer (Z = 200R1.6), WSR-88D (Z = 300R1.4), and Rosenfeld tropical (Z = 250R1.2)) for the Delhi radar station for 2019. The study was performed for four seasons (Winter, Pre-Monsoon, Monsoon, and Post-monsoon) as well as for different rainfall intensity (Light, Moderate, and Heavy rain). The accuracy of each relationship was evaluated using statistical variables such as correlation coefficient (R), RMSE, MAE, Bias and NSE. Results indicate significant variability in rainfall estimation across different relationships. The Marshall-Palmer encompasses the best correlation with rain gauge data during the monsoon and post-monsoon, whereas the Rosenfeld Tropical exhibits the strongest correlation for the winter and pre-monsoon. Additionally, Rosenfeld Tropical has a strong correlation for moderate and heavy rainfall intensity, whereas Marshall Palmer offers a satisfactory correlation for light rainfall intensity. However, Marshall-Palmer offers the best performance for the overall dataset with R = 0.623, RMSE of 13.44, and MAE = 10.07, as well as the lowest RMSE for all seasons and rainfall intensity. These findings highlight the significance of selecting a correct Z-R relationship for accurate rainfall estimation in diverse meteorological conditions, and underscore the need for localized calibration of Z-R parameters for enhanced forecasting accuracy in the Delhi region. © 2025 Elsevier Ltd.