Publication: Convolutional Neural Network -Support Vector Machine Model-Gaussian Process Regression: A New Machine Model for Predicting Monthly and Daily Rainfall
Date
2023
Authors
Ehteram M.
Ahmed A.N.
Sheikh Khozani Z.
El-Shafie A.
Journal Title
Journal ISSN
Volume Title
Publisher
Springer Science and Business Media B.V.
Abstract
Rainfall prediction is an important issue in water resource management. Predicting rainfall helps researchers to monitor droughts, surface water and floods. The current study introduces a new deep learning model named convolutional neural network (CONN)- support vector machine (SVM)- Gaussian regression process (GPR) to predict daily and monthly rainfall data in Terengganu River Basin, Malaysia. The CONN-SVM-GRP model can extract the most important features automatically. The main advantage of the new model is to reflect the uncertainty values in the modelling process. The lagged rainfall values were used as the input variables to the models. The proposed CONN-SVM-GRP model successfully decreased the Mean Absolute Error (MAE) of other models by 5.9%-23% at the daily scale and 20%-61% at the monthly scale. The CONN-SVM-GRP model also provided the lowest uncertainty among other models, making it a reliable tool for predicting data points and intervals. Hence, it can be concluded that CONN-SVM-GRP model contributes to the sustainable management of water resources, even when satellite data is unavailable, by using lagged values to predict rainfall. Additionally, the model extracts important features without using preprocessing methods, further improving its efficiency. Overall, the CONN-SVM-GRP model can help researchers predict rainfall, which is essential for monitoring water resources and mitigating the impacts of droughts, floods, and other natural disasters. � 2023, The Author(s), under exclusive licence to Springer Nature B.V.
Description
Keywords
Deep learning model , Machine Learning model , Rainfall pattern , Water resource management , Malaysia , Terengganu , Terengganu Basin , West Malaysia , Convolution , Convolutional neural networks , Deep learning , Disasters , Drought , Floods , Forecasting , Gaussian distribution , Gaussian noise (electronic) , Information management , Learning systems , Rain , Resource allocation , Surface waters , Uncertainty analysis , Water management , Convolutional neural network , Daily rainfall , Deep learning model , Learning models , Machine learning models , Monthly rainfalls , Network support , Rainfall patterns , Support vectors machine , Water resources management , artificial neural network , climate prediction , Gaussian method , machine learning , natural disaster , precipitation assessment , rainfall , support vector machine , water resource , Support vector machines