基于EV-CICR-PCC的城市洪涝灾害韧性指标筛选与评估方法OA

With the increasing frequency and intensity of extreme rainfall events under climate change,urban flood disasters have become a major challenge to sustainable urban development,public safety,and infrastructure security.Improving urban flood resilience is therefore a key objective of resilient city construction and disaster risk management.Despite the proliferation of resilience assessment frameworks in the literature,several critical gaps remain unaddressed.Many indicator systems contain redundant variables,traditional evaluation methods often ignore interactions among indicators,and small-sample datasets make probabilistic risk assessment diffi-cult.These issues may reduce the reliability and practical applicability of resilience evaluations.To address these challenges,an integrated framework is proposed for urban flood disaster resilience assessment that combines indi-cator screening,nonlinear resilience aggregation,and probabilistic risk identification.First,an EV-CICR-PCC in-dicator screening method is developed by integrating the Entropy Variation Coefficient(EV),Cumulative Infor-mation Contribution Rate(CICR),and Pearson Correlation Coefficient(PCC).Entropy and variation coeffi-cients are jointly used to evaluate the information content and discriminative ability of indicators.CICR is then applied to retain indicators with high cumulative information contribution,while PCC analysis removes highly correlated indicators with lower information value.This process effectively reduces redundancy while preserving essential resilience information.Second,the Choquet integral is employed to aggregate the selected indicators.Unlike conventional linear weighting approaches,the Choquet integral can capture nonlinear interactions among indicators,including complementarity and substitutability effects.This feature enables a more realistic represen-tation of the complex relationships among environmental,economic,social,and infrastructure factors that jointly influence urban flood resilience.Third,information diffusion theory is introduced to transform discrete resilience values into continuous probability distributions,allowing exceedance probabilities and risk levels to be estimated under small-sample conditions.The proposed framework is applied to five prefecture-level cities in the Tuojiang River Basin of the upper Yangtze River Basin:Chengdu,Deyang,Luzhou,Neijiang,and Ziyang.Based on statis-tical yearbooks,water resources bulletins,and socio-economic reports from 2016 to 2020,an initial indicator sys-tem consisting of 21 indicators is established across four dimensions:environmental resilience,economic resil-ience,social resilience,and infrastructure resilience.After EV-CICR-PCC screening,10 key indicators are re-tained,including annual average temperature,annual precipitation,heavy-rainfall-day proportion,river network density,per capita disposable income,unemployment rate,illiteracy rate,health institution density,health techni-cians per 10,000 people,and per capita road area.The assessment results reveal significant spatial differences in urban flood resilience.Chengdu exhibits the highest resilience value(0.6264),followed by Deyang(0.4088)and Luzhou(0.4049).Neijiang(0.2858)and Ziyang(0.2319)show relatively low resilience levels.The re-sults indicate a"strong core-weak periphery"pattern within the basin,highlighting the combined influence of economic development,infrastructure capacity,ecological conditions,and social vulnerability.Information diffu-economic development,infrastructure capacity,ecological conditions,and social vulnerability.Information diffu-sion analysis further demonstrates that Neijiang and Ziyang are concentrated in lower-resilience and higher-risk intervals,whereas Chengdu remains within a relatively low-risk category under higher resilience thresholds.Comparisons with traditional linear methods,entropy-weight TOPSIS,and conventional Choquet integral models demonstrate the superiority of the proposed framework.The method achieves better indicator optimization,stron-ger differentiation among cities,and more reliable risk identification,particularly under small-sample conditions.The main contribution of this study is the establishment of a complete methodological framework linking indica-tor refinement,resilience evaluation,and risk assessment.The EV-CICR-PCC method improves the objectivity and efficiency of indicator selection,the Choquet integral captures nonlinear resilience mechanisms,and informa-tion diffusion theory enhances probabilistic risk characterization.The proposed framework provides both method-ological support for urban flood resilience research and practical guidance for risk zoning,infrastructure plan-ning,vulnerability reduction,and resilient urban governance.