Divergent impact of urban 2D/3D morphology on thermal environment along urban gradients

Optimizing urban morphology is an effective way to alleviate the thermal environment, and it has always been the focus of urban planning. Although previous scholars have explored the impact of urban 2D/3D morphology on Land Surface Temperature (LST), they have rarely revealed differences along urban gradients, thus ignoring the nonlinear relationship between urban morphology and LST. In this study, we use multi-source data such as night-time lights, Landsat 8, land cover, and buildings, and employ dynamic threshold, single-window algorithm, and boosted regression trees methods to reveal the gradient and marginal effects of urban 2D/3D morphology on LST. Results showed that the proportion of built-up land (PCL), digital elevation model (DEM) and normalized difference built-up index (NDBI) have important contribution to LST. Moreover, relative influence of PCL, NDBI and building density (BD) on LST varies significantly along urban gradients, which are +25.87%, +14.85% and − 10.88%, respectively. Interestingly, we found that as the value of each variable increased, its relationship to LST changed. For example, when PCL < ∼49%, it is negatively correlated with LST, so lowering PCL can alleviate thermal environment. Our findings inform the sound practice of urban planning by accounting for marginal effects of urban morphology on thermal environment.