A satellite based machine learning approach for estimating high resolution daily average air temperature in a megacity in Brazil

This dataset was generated using a Random Forest model as described in the study entitled "A satellite-based machine learning approach to estimate high resolution daily air temperature in São Paulo, Brazil". We developed a generalizable tree-based machine learning approach (Random Forest) to estimate daily mean temperatures at 500 x 500 metres resolution for São Paulo, Brazil. Methods We trained a Random Forest model using open-access remote sensing data, along with derived products, and temperature measurements from 43 ground stations. To prevent overfitting and select relevant features, we employed a forward feature selection algorithm with target-oriented (spatial) cross-validation. Hyperparameter tuning was performed using grid search approach. The model was validated through ten-fold spatial cross-validation and an external hold-out dataset. The model demonstrated strong performance (RMSERF = 0.80; R²RF = 0.95), with slightly reduced accuracy in rural areas (R²rural = 0.91; R²urban = 0.95). Compared to traditional multilinear approaches (RMSEMLR = 1.02; R²MLR = 0.92), the Random Forest model outperformed, likely due to its ability to better capture microclimates and complex relationships between data sources. This 500 x 500 metres daily temperature dataset is the first of its kind in South America, with the São Paulo pipeline and data freely accessible. The approach is adaptable to other regions with appropriate retraining and validation, enabling high-resolution exposure assessments Why is this important? Spatiotemporally resolved ambient temperature data is essential for environmental epidemiological studies, particularly in densely populated and urbanized areas where temperature can vary significantly over short distances. Spatially continuous temporal records at high spatial resolution are, however, often lacking, especially in low- and middle-income countries. This 500 x 500 metres daily temperature dataset is the first of its kind in South America, with the São Paulo pipeline and data freely accessible. The approach is adaptable to other regions with appropriate retraining and validation, enabling high-resolution exposure assessments Metadata: Temporal Resolution Daily, Monthly, Annually, All period Temporal Coverage 2015-2019 Spatial Resolution 500m x 500m Spatial Coverage São Paulo municipality - bounding box (lonmin: -46.9559; latmin: -24.0854; lonmax: -46.2226; latmax: -23.2839, coordinate reference system: WGS84) Format TIFF Dataset upload 12.07.2025