Title: Precipitation prediction in Bangladesh using machine learning approaches

Authors: Md. Ariful Islam; Mosa. Tania Alim Shampa

Addresses: Department of Robotics and Mechatronics Engineering, University of Dhaka, Dhaka – 1000, Bangladesh ' Department of Oceanography, University of Dhaka, Dhaka – 1000, Bangladesh

Abstract: In the assessment of different hydrological activities, the prediction of rainfall is essential. As agriculture is critical to survival in Bangladesh, rainfall or precipitation is most important. This study shows how a machine learning approach can be used to make a reliable model for predicting rain. This way, people can know when rain is coming and take the steps they need to protect their crops. Many techniques have been applied so far to predict rainfall. But machine learning algorithms can provide more accuracy in this case. Nine machine learning algorithms have been used to find a good model that can be used to predict rain in Bangladesh. The prediction models were evaluated by dint of evaluation metrics such as coefficient of determination (R²), root mean squared error (RMSE), and mean absolute error (MAE). Among nine algorithms and eight models, the model H including all meteorological exogenous inputs with gradient boosting regressor algorithm led to the best predictions (R² = 0.78, RMSE = 134, MAE = 92) for Sylhet division. The model G excluding wind speed with gradient boosting regressor algorithm shows the best predictions (R² = 0.76, RMSE = 147, MAE = 89) for both Chittagong and Rangpur divisions.

Keywords: rainfall; machine learning algorithms; precipitation; gradient boosting regressor; GBR; Bangladesh.

DOI: 10.1504/IJHST.2024.139395

International Journal of Hydrology Science and Technology, 2024 Vol.18 No.1, pp.23 - 56

Received: 17 Aug 2022
Accepted: 15 Feb 2023
Published online: 02 Jul 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article