Urban tree planting should consider local characteristics: assessing spatial heterogeneity in canopy cooling effects on land surface temperature using Bayesian spatially varying coefficient models

Introduction Urban trees are essential for mitigating elevated temperatures in cities worldwide, with many municipalities implementing large-scale urban tree planting initiatives. However, the cooling potential of tree canopy coverage is often estimated as a constant value across study areas, despite evidence that temperature reductions depend on local characteristics, including tree traits and urban geometry. Methods We evaluated the ability of Bayesian Spatially Varying Coefficient (SVC) models to capture local variability in the cooling potential of urban trees. The model, implemented in R-INLA, integrated Landsat 8 and 9 Land Surface Temperature (LST) data with aerial LiDAR data. Model performance was assessed using validation metrics obtained through 10-fold spatial cross-validation. Results Although the SVC did not outperform simpler spatio-temporal approaches according to validation metrics, the spatial distribution of local canopy cooling capacity revealed substantial spatial variability. Average estimated values of canopy cooling capacity on LST (defined as the change in LST associated with a 10% increase in tree canopy cover) were −0.28 °C in vacant lands and −0.09 °C in wooded areas. Discussion By providing local estimates, our model underscores how the cooling capacity of tree canopy in built-up environments varies substantially across space. This finding demonstrates the importance of accounting for local environmental characteristics in urban planning and serves as an example of a modeling approach that integrates both local-scale variability in canopy cooling capacity and spatial extent. These results encourage policymakers to adopt context-specific strategies for urban tree planting initiatives rather than applying uniform approaches.