城市消防警情的空间异质性及影响因素

Spatial heterogeneity and influencing factors of urban emergency services

城市消防警情主要分为火警和非火警(抢险救援和社会救助),目前对火警的研究相对较多,而对救援救助警情的研究相对较少。该文采用探索性空间数据分析(ESDA),定量研究了城市消防警情的空间差异和集聚程度,确定了影响消防警情的社会因素,构建了多尺度地理加权回归(MGWR)模型,将其应用到城市消防警情的实证研究中,并采用多元线性回归、地理加权回归模型(GWR)进行了对比分析。研究结果表明:火警和救援救助在所研究城市都呈现出一定程度的空间聚集,老城区为火警的热点区域,而救援救助的热点分布相对更广。MGWR模型相较于传统的多元线性回归和经典的GWR模型有着更好的效果,在火警、救援救助和总警情中拟合优度均超过了0.8,且其残差平方和、修正后的Akaike信息准则(AICc)数值最低。通过GWR模型和MGWR模型的带宽对比分析发现,不同社会因素对消防警情的影响存在空间异质性。该研究还发现救援救助与火警在该城市的空间分布及影响因素的空间异质性上具有显著差异。在未来的消防工作中,考虑救援救助对整体警情的影响具有重要意义。

[Objective]Urban emergency services are mainly divided into fire suppression and technical rescues.Relatively more studies have been conducted on fire suppression,whereas relatively fewer studies have focused on technical rescues.Thus,this study aims to map the spatial distribution of fire suppression and technical rescues on a city scale and build their connections quantitatively to the human population and mobility.The findings are expected to help in the planning of urban emergency services to follow the major task change from fire suppression to technical rescues.[Methods]The global spatial autocorrelation of fire suppression,technical rescues,and their totality—whole emergency services—was assessed using global spatial autocorrelation analysis Moran's I,whereas the local components were indexed with local indicators of spatial association and Getis-Ord G*i.The human population and mobility were modeled through the point of interest(POI)and visitor throughput,respectively.Through the stepwise regression method,five broad categories of POI data(14 minor categories of POI)of the highest sensitivities were selected from the available 30 categories of POI.Their associations with fire suppression,technical rescues,and the whole emergency services were established using the multiscale geographically weighted regression(MGWR)model.[Results]Both fire suppression and technical rescues were found to have a certain degree of spatial clustering,but some differences were noted in the spatial distribution of different emergency service types.Old towns were the concentrated hot spots for fire suppression,whereas the formation of additional clusters of technical rescues was extensively distributed;thus,more multiregional linkage and targeted prevention were required for technical rescues than fire suppression.The POI data of residential premises,office premises(such as office buildings,and public administration and public service institutions),industrial premises(such as industrial parks and mines),educational premises(such as schools,scientific research institutions,and training institutions),and commercial premises(such as supermarkets,convenience stores,home appliance stores,digital device stores,beauty salons),and visitor throughput were closely connected to emergency services.The applied MGWR model overtook both the traditional multiple linear regression and conventional GWR,with the goodness of fit R2 exceeding 0.8 for fire suppression,technical rescues,and overall emergency services.The residual sum of squares and the corrected Akaike information criterion(AICc)had the smallest values in the MGWR model.The correlation of the local visitor throughput and POIs(e.g.,residential buildings,offices,and retail stores)to local fire suppressions were approximately spatially uniform.By contrast,the connections of other POIs(e.g.,offices,schools,and industrial parks)to fire suppression and technical rescues varied across the present city domain.[Conclusions]The findings indicate that targeted prevention should be employed in the present city's emergency services according to the local POI distribution and should be steered to technical rescues which have become the main part of the overall emergency services.This study provides an important reference for future emergency preparedness and response,regional prevention work,and location planning of new fire stations.