Key attributes of greenspace pattern for heat mitigation vary with urban functional zones

Context Urban green space (UGS) has been widely acknowledged to effectively reduce ambient temperature, and thus improving its cooling effect has become a hot topic in urban planning. Although the impact of UGS spatial pattern on land surface temperature (LST) has been investigated in previous studies, the differences caused by the heterogeneity of human activities within urban areas have not been fully understood. Objectives This study attempted to explore different relationships between UGS spatial pattern and LST among various human activity dominated zones to effectively support urban heat mitigation. Methods Taking the urban area within the third ring road of Changsha City as the study area, the random forest model was applied to evaluate the contribution of UGS spatial pattern on LST and partial dependence curves were generated to identify their nonlinear relationships among different urban functional zones (UFZs). Results In terms of varied dominant influencing factors affecting LST, the area proportion of UGS was the most important factor in residential, industrial, campus, and public zones, which showed a negative impact on the variation of LST. The cohesion between UGS patches was also an important factor in residential zones, and the more aggregated the more conducive to cooling. UGS patch density interpreted the highest importance in commercial zones, which played a significantly negative role when less than a specific threshold. However, the influence of landscape shape complexity on LST was limited in five built-up functional zones. Conclusions The impact of UGS spatial pattern on heat mitigation was quite different in various UFZs. The findings can provide targeted guidance for green space planning in regard to effectively mitigating the urban heat island effect.