街区尺度城市三维建筑形态对地表径流的影响研究OA

Against the backdrop of escalating global climate change and rapid urbanization,enhancing urban flood resilience and mitigating the impacts of extreme precipitation have become critical research topics in urban planning and stormwater management.Buildings are dominant components of the urban underlying surface,and their spatial configurations directly influence local hydrological processes.However,existing studies have largely focused on macro-scale risk assessments or the flood mitigation effectiveness of green space systems,generally treating buildings merely as a single factor that increases runoff.Consequently,there has been a relative lack of systematic quantitative analysis regarding the dual spatial effects,both runoff enhancement and suppression of three-dimensional building forms on surface runoff at the block scale.Taking six typical flood-prone blocks in Shenyang as case studies,this research explores the nonlinear relationships and multidimensional coupling effects between three-dimensional building spatial forms and surface runoff.The study analyzes hydrological processes within the study area based on extreme rainfall scenarios.Specifically,surface runoff data were simulated using a GIS-based Soil Conservation Service Curve Number(SCS-CN)model.To ensure simulation accuracy,the model parameters were localized by calculating comprehensive CN values through an area-weighting method based on block-specific land use,soil properties,and vegetation cover.The hydrological response was then simulated under a 50-year return period design rainstorm scenario,generating the spatial distribution of grid-level surface runoff as the dependent variable for subsequent analyses.To comprehensively quantify the impact of three-dimensional building forms,this study developed a three-dimensional building form index system derived from the intrinsic transmission mechanism of"building form-land surface-surface runoff".These indicators encompass three dimensions:horizontal constraints,vertical regulation,and spatial patterns.After multicollinearity testing and optimization using Pearson's correlation coefficient,seven core predictive indicators were retained.This study aimed to develop a comprehensive analytical framework that integrates linear benchmarking with nonlinear analysis.First,a multivariate linear regression benchmark model was employed to quantitatively analyze the overall linear correlation and influence trends of various indicators on surface runoff.To overcome the"black box"nature of complex models and precisely quantify the threshold effects of dominant variables,the study introduced a nonlinear interpretation framework combining Random Forests(RF)with the Shapley Additive Explanation(SHAP),incorporating feature importance ranking and partial dependency plots for in-depth analysis.The results of the feature analysis indicate that multiple linear regression and the RF-SHAP model can systematically analyze the influence of three-dimensional building morphology on surface runoff.SHAP analysis revealed the relative importance and dual attributes of various three-dimensional building morphological indicators.Spatial Congestion Degree(SCD)and Building Structure Index(BSI)were identified as the primary runoff-enhancing indicators.The next most influential features were the Frontal Area Index(FAI)and Average Height(AH),which serve as key runoff-suppressing indicators.Crucially,SHAP dependency plots generated through nonlinear analysis revealed the multidimensional coupled characteristics of three-dimensional building forms on surface runoff,encompassing"horizontal constraints,vertical regulation,and spatial blocking".For instance,in the horizontal dimension,the impact of spatial density shifts abruptly around a threshold of 30%;exceeding this value causes impermeable patches to infiltrate,leading to a sharp increase in runoff.BSI exhibits a significant runoff-enhancing effect within the 7-20 range.The influence of Building Surface Area Average(BSA)exhibits a clear positive correlation with runoff,particularly becoming more pronounced beyond 1,500 m²,as it triggers a strong local catchment effect causing instantaneous flow rates to far exceed the surface infiltration limit.In the vertical dimension,an average building height between 30 and 60 m produces an optimal flow-suppressing effect by effectively intercepting wind and rain and promoting evaporation.In contrast,when building height exceeds 60 m or windward density surpasses 0.5,building facades force high-altitude rainwater to concentrate at the base,which,combined with near-surface wind resistance suppressing evaporation,actually exacerbates water accumulation.Additionally,regarding spatial patterns,the building Height Variation Coefficient(HVC)exhibits a unique property:low values enhance runoff,whereas high values suppress it.Specifically,positive flow-suppression effects occur when HVC falls within the 0.3-1.0 range,as the staggered arrangement of buildings significantly increases three-dimensional spatial roughness and hydraulic resistance,thereby generating a spatial retention effect.In summary,this study demonstrates the effectiveness of the RF-SHAP framework in analyzing the complex nonlinear effects of three-dimensional building forms on surface runoff.The findings confirm that enhancing the three-dimensional morphological heterogeneity of buildings in both vertical and spatial dimensions effectively reduces the risk of urban flooding caused by horizontal sprawl.These findings provide a solid quantitative foundation for developing coordinated optimization strategies for flood prevention spaces at the block scale under extreme rainfall scenarios(e.g.,mandating SCD of high-density blocks to remain≤30%,subdividing large-scale ground-floor commercial spaces,and vertically expanding low-rise,aging residential communities to 30-60 meters).