基于优化地理探测器的生态环境质量驱动力分析

Eco-environmental Quality Driving Force Detection Using Optimized Geographic Detector

[目的]克服生态环境质量(EQI)演化驱动因子在离散化过程中的随机性和主观性,优化传统地理探测器。[方法]利用相等间隔法、分位数法、自然断点法、几何间隔法和标准差法5种离散方法,对重庆市22项EQI演化驱动因子进行离散化,并结合地理探测器确定各因子的合理离散法与分类数,从而探究各驱动因子的驱动力大小和各驱动因子间交互模式。[结果](1)2005—2020年重庆市EQI处于I、Ⅲ和Ⅳ级的区县数量占比减少0.211,处于Ⅱ和V级的区县数量占比上升0.211,总体呈下降趋势,在空间上呈由西北向东南走向的梯度分布特征。(2)研究中各离散方法适用性排序为:自然断点法>几何间隔法>分位数法>相等间隔法>标准差法。不同因子适用的离散方法各异,应根据因子数据特征择优选取离散方法。(3)识别出影响重庆市EQI演化的关键驱动因子4个(0.37~0.49),主要驱动因子13个(0.14~0.33)、次要驱动因子4个(0.05~0.13)和其他因子1个(0.04),各驱动因子间交互作用均为双重增强或非线性增强。[结论]合理的离散化可以一定程度上克服连续数据离散化过程中的随机性和主观性,从而优化重庆市EQI演化驱动力探测结果。

[Objective]This study arms to overcome the randomness and subjectivity of ecological environment quality(EQI)evolution drivers in the discretization process,and to optimize the traditional geographic detector.[Methods]This study discretized 22 EQI evolutionary driving factors in Chongqing by using five discretization methods,namely equal interval method,quantile method,natural breakpoint method,geometric interval method and standard deviation method.Geographic detector was combined to determine the reasonable discrete method and classification number of each factor,so as to explore the driving force size of each driving factor and the interaction model of each driving factor.[Results](1)From^(2)005 to 2020,the number of districts and counties in grade I,ⅢandⅣdecreased by 0.211,while the number of districts and counties in GradeⅡand V increased by 0.211,showing an overall downward trend and spatial gradient distribution from northwest to southeast.(2)The order of applicability of each discrete method in this study was:natural break point method>geometric interval method>quantile method>equal interval method>standard deviation method.The discretization methods applicable to different factors were different,so the discretization methods should be selected according to the characteristics of factor data.(3)Four key driving factors(0.37~0.49),13 main driving factors(0.14~0.33),four minor driving factors(0.05~0.13)and one other factor(0.04)were identified,and the interaction among all driving factors was doubly enhanced or nonlinear enhanced.[Conclusion]Reasonable discretization can overcome randomness and subjectivity in the process of continuous data discretization to a certain extent,so as to optimize the detection results of EQI evolution driving forces in Chongqing.