2017—2021年湖北省生产建设项目人为扰动区域快速识别和提取

Rapid Identification and Extraction of Anthropogenically Disturbed Regions Resulting from Production and Construction Projects in Hubei Province from 2017 to 2021

[目的]对湖北省2017—2021年生产建设项目人为扰动区域进行逐年识别提取,并对分类结果进行时空特征分析,为后续人为扰动区域提取及水土流失相关研究提供理论支撑和方法参考。[方法]以生产建设项目人为扰动类型丰富的湖北省为例,基于GEE平台调用Sentinel-2时序数据,利用1955个样本数据探究最优分类特征组合及最优分类参数,针对性解决不透水层和耕地易与生产建设项目人为扰动区域混淆的问题,利用随机森林模型对该省2017—2021年生产建设项目人为扰动区域进行逐年识别提取研究。[结果](1)生产建设项目人为扰动区域识别提取的最优特征波段组合为红边波段、绿边波段、蓝边波段、近红外波段以及NDVI,NDWI,NDBI,RRI,dNDVI,对比度和熵;(2)从总体上看,2017—2021年分类的总体精度均高于93.00%,kappa系数均在0.92以上,表明该方法在生产建设项目人为扰动区域提取的问题上可行;(3)2017—2021年湖北省生产建设项目人为扰动地块的总面积呈现“先增后减再增”的变化趋势,在2020年出现了反常的减少。[结论]本文提出的方法在快速识别大尺度、多类型生产建设项目人为扰动区域问题上有较大潜力,生成的高精度、长时序空间数据集可为后续相关工作提供支持。

[Objective] This study attempted to identify and extract anthropogenic disturbance areas resulting from production and construction projects in Hubei Province from 2017 to 2021.Spatiotemporal feature analysis of the classification results was also conducted in order to provide theoretical support and methodological references for the extraction of anthropogenic disturbance areas,and to address soil and water erosion issues.[Methods] The study was conducted in Hubei Province because of its diverse range of anthropogenic disturbance types from production and construction projects.We used the Google Earth Engine(GEE) platform to access Sentinel-2 time-series data.We investigated the optimal combinations of classification features and parameters using a 1 995 dataset.To tackle the issue of differentiating impermeable layers from croplands within anthropogenic disturbance areas caused by production and construction projects,we employed a random forest model for the annual identification and extraction of such areas in Hubei Province from 2017 to 2021.[Results](1) The optimal feature band combination for identifying and extracting anthropogenic disturbance areas from production and construction projects included the red-edge band,green-edge band,blue-edge band,near-infrared band,NDVI,NDWI,NDBI,RRI,dNDVI,contrast,and entropy.(2) Overall,the classification accuracy for the years 2017 to 2021 consistently exceeded 93.00%,with kappa coefficients consistently above 0.92,affirming the method was feasibility for extracting anthropogenic disturbance areas due to production and construction projects.(3) The total area of anthropogenic disturbance land parcels in Hubei Province exhibited a pattern characterized as “increase—decrease—increase” from 2017 to 2021,with an anomalous decrease in 2020.[Conclusion] The proposed method demonstrated substantial potential for the rapid identification of large-scale,diverse anthropogenic disturbance areas resulting from production and construction projects.The resulting high-precision,long-term spatial dataset can provide valuable support for subsequent research endeavors related to this topic.