Quantifying the nonlinear and interactive effects of urban form on resilience to extreme precipitation: Evidence from 192 cities of Southern China

Abstract

As extreme precipitation events become frequent due to climate change, understanding how urban form (UF) contribute to urban resilience is critical for developing effective urban resilience strategies. However, there is a lack of quantitative insights into the complex roles that various UF characteristics play in influencing urban resilience to extreme precipitation (REP). To address this issue, this study introduces a “Landscape-Materials-Service” theoretical framework. Utilizing data from 192 cities in Southern China from 2020, the XGBoost-SHAP model is employed to investigate the nonlinear and interactive effects of UF factors on REP. The results indicate that building height ($BH$), road density ($RD$), and the location quotient of living service facilities ($LQ\_Living$) are the dominant factors. Further local analysis shows that each feature exhibits distinct nonlinear effects, with adaptive thresholds observed in shaping resilience performance. Importantly, a positive “triangular interaction” is found among $BH$, $RD$, and $LQ\_Living$. Additionally, blue, green, and gray infrastructures demonstrate both synergistic and antagonistic relationships. This research presents a novel framework for incorporating urban form into resilience planning, highlighting that urban resilience can be enhanced by balancing urban landscapes, physical infrastructure, and service facilities. These insights serve as guidance for urban planners in creating climate-resilient cities.