耦合ER和GWR的福州市生态环境质量的驱动力分析

Analysis of the Driving Forces of Ecological Environment Quality in Fuzhou City Coupled with ER and GWR

城市的生态环境质量受到诸多因素的共同影响,其中以道路建设最为显著,客观分析道路网络及其他因素对生态环境质量的驱动机制,对最大限度减少对生态系统的负面作用具有重要的参考意义。基于道路网络、Landsat系列遥感影像、夜间灯光、数字高程模型、气象和土地利用等多源数据集,在3S技术的支持下,首先采用增量空间自相关和核密度估算(KDE)计算福州市2015年和2020年的道路核密度,利用主成分分析法构建福州市2000、2009和2020年的遥感生态指数(RSEI),在此基础上,分析两者的时空动态变化,接着采用探索性回归(ER)筛选关键影响因子,最后运用地理加权回归模型(GWR)揭示关键影响因子对福州市生态环境质量的驱动机制。结果表明,1)2015年和2020年最佳带宽下KDE的变化范围分别是0-4.090 km∙km^(-2)和0-3.765 km∙km^(-2);高KDE值在2015年主要聚集在福州市区周围和各区县的中心,而至2020年,高KDE值的范围逐渐扩大,呈现向沿海地区蔓延的趋势。2)从时间上看,福州市2000-2020年期间的RSEI呈现先上升后下降的趋势,生态状况整体上相对稳定;从空间上看,生态环境质量好的等级分布在永泰县和闽清县等山区,较差的等级主要分布在市区中心、各区县的中心区、东部沿海区域及沿江两侧。3)对探索性回归筛选出的最佳因子进行拟合,GWR模型拟合效果优于OLS模型。GWR结果表明道路欧氏距离、高程、坡度、林地比例和草地比例与RSEI主要呈正相关关系,道路核密度、夜间灯光、城乡用地比例与RSEI主要呈负相关关系,回归系数的分布呈现明显的空间分异特征。研究结果可为福州市以及其他城市路网规划和生态环境质量提升提供参考依据。

The ecological quality of urban environments is jointly influenced by multiple factors,with road construction being the most significant.Objectively analyzing the driving mechanisms of road networks and other factors on ecological quality is of great importance for minimizing the negative impacts on ecosystems.Based on multi-source datasets including road networks,Landsat series remote sensing images,nighttime lights,digital elevation models,meteorology,and land use,the road density in Fuzhou City in 2015 and 2020 was calculated by using incremental spatial autocorrelation and kernel density estimation(KDE)supported by 3S technology.Then,the remote sensing ecological index(RSEI)for Fuzhou City in 2000,2009,and 2020 was constructed using principal component analysis.On this basis,the spatiotemporal dynamic changes of road density and RSEI were analyzed.Exploratory regression(ER)was then used to identify key influencing factors.Finally,a geographically weighted regression model(GWR)was applied to reveal the driving mechanisms of key influencing factors on the ecological environment quality in Fuzhou City.The results indicate that 1)the range of KDE under the optimal bandwidth in 2015 and 2020 was 0-4.090 km∙km^(-2)and 0-3.765 km∙km^(-2),respectively.In 2015,high KDE values were mainly concentrated around the urban area of Fuzhou and the centers of various districts and counties.However,by 2020,the range of high KDE values had gradually expanded,showing a tendency to spread towards coastal areas.2)From a temporal perspective,the RSEI of Fuzhou City showed an upward trend followed by a downward trend from 2000 to 2020,indicating a relatively stable ecological condition overall.Spatially,areas with good ecological environment quality were mainly distributed in mountainous regions such as Yongtai County and Minqing County.Areas with poorer quality were primarily concentrated in the central urban area,central areas of various districts and counties,eastern coastal regions,and along both sides of the river.3)The GWR model,fitted with the best factors selected through ER,outperformed the OLS model.The GWR results showed that the Euclidean distance,elevation,slope,forest proportion,and grassland proportion were positively correlated with RSEI.Road density,nighttime lights,and urban-rural land ratio were negatively correlated with RSEI.The distribution of regression coefficients exhibited significant spatial variations.These findings can provide meaningful implications for road network planning and improvement of ecological environment quality in Fuzhou and other cities.