Hourly solar irradiance estimation and building-surface solar potential assessment via remote sensing and machine learning for data-scarce mountainous regions

Mountainous regions, as important terrestrial landscapes, possess considerable solar resources due to high altitudes, offering substantial potential for solar energy utilization on building surfaces. However, assessing solar potential on building surfaces in mountainous regions is challenging due to complex terrain shading, dynamic cloud cover, and limited meteorological data. This study develops a novel approach for hourly irradiance estimation and building surface solar potential assessment in mountainous regions by integrating machine learning (ML), MODIS cloud data, Digital Elevation Model, and 3D modeling. A model based on the Gradient Boosting Decision Tree (GBDT) algorithm was developed to estimate global horizontal (GHI), direct normal (DNI), and diffuse horizontal irradiance (DHI). Ablation experiments confirmed the substantial contribution of both cloud and terrain data to model accuracy. Comparative analysis using 5-fold cross-validation demonstrated that the GBDT model outperformed other ML algorithms (MLP, LightGBM, CatBoost, Random Forest), achieving the highest R² values of 0.8448 for GHI, 0.6643 for DNI, and 0.8001 for DHI. By integrating the estimated radiation data with 3D urban modeling tools (ArcGIS, Rhinoceros, Ladybug), the method accurately simulates solar radiation distribution on building surfaces in mountainous regions, demonstrated through a case study in Tai’an Town, Sichuan. The proposed method provides critical theoretical supports for optimizing building-integrated photovoltaic (BIPV) system installations in data-scarce mountainous areas, improving solar resource utilization. • Solves building surface solar assessment issues in data scarce mountains regions. • Solar potential estimation method integrates ML, MODIS, DEM & 3D urban data. • Validation shows significant accuracy improvement in GHI estimation with R 2 of 0.8448. • Provides theoretical basis for solar utilization in mountainous regions.