Spatialized importance of key factors affecting park cooling intensity based on the park scale

Zhihua Liu,Lizhao Fu,Chaofan Wu,Zhenzhen Zhang,Zhaoyang Zhang,Xingwen Lin,Xuan Li,Yuhong Hu,Haixian Ge

DOI: https://doi.org/10.1016/j.scs.2023.104952

IF: 11.7

2023-09-27

Sustainable Cities and Society

Abstract:Urban parks have been considered as an environmental mitigation strategy for heat island effect, but the spatial contribution of influencing factors affecting Park Cooling Intensity (PCI) is still uncertain. In this study, we selected 814 urban parks in Zhejiang Province, China to locate the key factors and their specific relative importance on the park scale by using the combination of eXtreme Gradient Boosting tree (XGBoost) and Shapley Additive exPlanations algorithm (SHAP). The results are as follows: (1) Most parks in the province had a cooling effect with an average PCI of 1.98°C. (2) Park area was the most important factor for PCI globally, followed by the patch density inside the park, but the water-derived factors except park inner waters were not at the forefront. (3) The significance of the water-related features (BUFwidth, waterLPI etc.) were remarkable for riverine parks indicated by heuristic local SHAP values. This study demonstrates that the interpretable algorithm combined with machine learning method is helpful to reveal the spatially explicit contribution of key factors for urban parks at a finer scale for more scientific management.

energy & fuels,construction & building technology,green & sustainable science & technology