基于Sentinel-2的江淮分水岭撂荒地提取  

Extraction of Abandoned Land in Jianghuai Watershed Based on Sentinel-2

作　　者：李元庆 宋宏利 刘欢 李伟涛[2] 刘兴宇 LI Yuan-qing;SONG Hong-li;LIU Huan(School of Earth Science and Engineering,Hebei University of Engineering,Handan,Hebei 056038;College of Geographic Information and Tourism,Chuzhou University,Chuzhou,Anhui 239000)

机构地区：[1]河北工程大学地球科学与工程学院,河北邯郸056038 [2]滁州学院地理信息与旅游学院,安徽滁州239000

出　　处：《安徽农业科学》2025年第3期47-52,76,共7页Journal of Anhui Agricultural Sciences

基　　金：河北省自然科学基金项目(D2019402067);安徽省教育厅自然科学重大项目(KJ2021ZD0131)。

摘　　要：[目的]研究江淮分水岭因地形破碎、丘陵起伏、云覆盖严重等原因导致撂荒地提取精度较低问题。[方法]基于GEE云平台,以凤阳县为研究区域,使用Sentinel-2 MSI影像数据,采用随机森林分类算法对凤阳县2017—2023年土地进行分类,获取7年的土地分类数据,并制定撂荒地识别规则,在此基础上提取研究区撂荒地数据。[结果]遥感影像的总体土地分类精度在82.45%~91.37%,符合研究所需精度要求;2018年研究区撂荒地面积最大,为974.52 hm^(2),撂荒率最高,为0.91%;2018—2019年撂荒地面积呈减少趋势,2019—2021年撂荒地面积逐年增加;2018—2023年撂荒地的复耕地面积最大约为678.15 hm^(2),而最小复耕地面积约为78.23 hm^(2),复耕率最高值达到69.59%,复耕率最小值为37.02%。[结论]该研究为评估农田抛荒的风险和生态环境的治理提供有力的数据支撑,也为其他撂荒地提取提供参考。[Objective]To study the problem of low accuracy in extracting abandoned land in the Jianghuai Watershed due to terrain fragmentation,hilly undulations and severe cloud coverage.[Method]Based on the GEE cloud platform,taking Fengyang County as the research area,Sentinel-2 MSI image data was used to classify the land in Fengyang County from 2017 to 2023 using the random forest classification algorithm.Seven years of land classification data were obtained,and abandoned land identification rules were formulated.Based on this,abandoned land data in the study area was extracted.[Result]The overall land classification accuracy of remote sensing images was 82.45%-91.37%,which met the accuracy requirements for research.The maximum areas of abandoned land in the study area was 974.52 hm^(2),with the highest abandonment rate of 0.91%in 2018.The abandoned land area showed a decreasing trend from 2018 to 2019,and increased year by year from 2019 to 2021;the maximum reclamation area of abandoned land from 2018 to 2023 was about 678.15 hm^(2),while the minimum reclamation area was about 78.23 hm^(2),the highest reclamation rate was 69.59%,the minimum reclamation rate was 37.02%.[Conclusion]This study provides strong data support for evaluating the risk of farmland abandonment and ecological environment management,and also provides reference for extracting other abandoned lands.

关 键 词：撂荒地 江淮分水岭 Sentinel-2 随机森林分类算法 凤阳县 

分 类 号：S127[农业科学—农业基础科学] TP79[自动化与计算机技术—检测技术与自动化装置]