Daily Solar Radiation Estimation with Gene Expression, Neural Network, and Random Forest Artificial Intelligence Models

Pourrajab, Mohammad

doi:10.22048/wss.2026.585524.1030

Daily Solar Radiation Estimation with Gene Expression, Neural Network, and Random Forest Artificial Intelligence Models

Document Type : Research Article

Author

Mohammad Pourrajab

Departmane of Water Engineering, University of Tabriz, Tabriz, Iran

10.22048/wss.2026.585524.1030

Abstract

Solar radiation plays a key role in a variety of physical, biological, and chemical processes, including snowmelt, evaporation, plant photosynthesis, and crop production. Solar radiation is also required in biological models for assessing forest fire risk, in hydrological simulation models for natural processes, and in environmental, meteorological, and agricultural research, as well as in atmospheric physics. In the present study, using ten-year data from selected meteorological stations across Iran on a daily scale, artificial intelligence models such as gene expression programming (GEP), neural network (ANN), and random forest (RF) were investigated for solar radiation estimation. By introducing different input combinations to different models, the accuracy of each model was evaluated using scatter index (SI), root mean square error (RMSE), and Nash-Sutcliffe efficiency (NS) criteria. The results show that the neural network model with four inputs, having SI of 0.17, RMSE of 3.25, and NS of 0.71, exhibited higher accuracy compared to other models. Additionally, the random forest model demonstrated satisfactory performance in estimating solar radiation.

Keywords

Artificial intelligence

Gene expression

Neural network

Random Forest

Solar radiation

Subjects