Machine Learning-based Model for Enhanced Quantitative Precipitation Estimates (QPE) from Radar Images Extracted Features

• Published in: 8th International Conference on Recent Advances and Innovations in Engineering: Empowering Computing, Analytics, and Engineering Through Digital Innovation, ICRAIE 2023
• Main Author: Osman N.S.; Tahir W.
• Format: Conference paper
• Language: English
• Published: Institute of Electrical and Electronics Engineers Inc. 2023
• Online Access: https://www.scopus.com/inward/record.uri?eid=2-s2.0-85189932511&doi=10.1109%2fICRAIE59459.2023.10468103&partnerID=40&md5=0a02a7b96f6d96690d2057ac8605aa27

Predicting extreme weather occurrences like flash floods has been made possible with the use of models built using high-resolution rainfall data and machine learning techniques. This paper presents a machine learning-based model to enhance radar quantitative precipitation estimation (QPE) derived from extracted radar image characteristics. The machine learning employed in this study is an artificial neural network (ANN) model with Levenberg-Marquardt (LM) algorithm and feedforward function network. The features of rain cells from radar images were extracted using eight characteristics such as rain area and max intensity. The features are then trained with the observed mean areal gauge rainfall at the collocated time and space. The robustness of the ANN-Radar QPE model is assessed using the correlation coefficient, r, and the root mean square error (RMSE). Results show the radar QPE is much improved after integration with the ANN model. The correlation coefficient also indicates a good relationship between ANN-Radar QPE and rain gauge data. For future work, validation of the algorithm and its application will be done by coupling the model with a hydrological or flood model. © 2023 IEEE.