Document Type : Original Article

Author

• Moslem Mohammadi

Payame Noor University (PNU), Tehran, Iran.

Abstract

Abstract :

Accurate precipitation estimation plays a fundamental role in weather forecasting, water resources management, and disaster mitigation. In recent years, machine learning methods have emerged as a powerful tool for improving precipitation estimation. In this paper, artificial neural network regression and support vector machine regression methods are evaluated to map the estimated precipitation from ERA5 data to the precipitation measured at meteorological stations. In this study, various methods, including statistical methods and machine learning, are investigated by combining different features to improve the accuracy of ERA5 data. The study area includes 16 meteorological stations. The results show that the artificial neural network regression method increased the accuracy of rainfall estimation in proportion to the number of input features and, in the best case , led to the achievement of RMSE (2.73 mm for daily), CC (0.71 for daily ), NSE ( 0.50 for daily ), and NRMSE ( 0.05 for daily ).

Keywords

• Precipitation estimation
• neural networks
• support vector machine
• machine learning
• ERA5 data

Main Subjects

• Climate change