摘要
煤炭基地是我国煤炭资源的集中产业地,面临着植被损毁区域大、损毁原因复杂多变的问题。定量分析长时序植被损毁识别方法在煤炭基地的适用性,对当地生态环境质量监管具有十分重要的意义。基于Google Earth Engine云计算平台的Landsat影像数据,从植被损毁识别、不同地表区域适用性、植被损毁时间3个方面,定量对比分析了Land Trendr(LT)和Continuous Change Detection and Classification(CCDC)算法在神东煤炭基地进行长时序(1990-2020年)植被损毁识别中的适用性。研究表明:(1)在植被损毁识别上,LT和CCDC算法总体精度分别为73.6%和84.4%,识别效果较好。(2)LT和CCDC算法都能较好地避免林地区域的错分误差。但LT算法仅能识别露天采场区域部分植被损毁,遗漏误差较大,且基本无法识别到城市扩张所造成的植被损毁,而CCDC算法对这两类区域的识别效果较好。在水体区域,LT算法显著优于CCDC算法。(3)LT和CCDC算法识别损毁时间的误差在1 a内的结果分别占95.7%和90%,损毁时间识别效果很好。总体而言,相较于LT算法,CCDC算法更适用于城市扩张明显、水体面积很少的神东煤炭基地植被损毁识别。上述分析为神东煤炭基地生态环境质量监管提供了数据参考,更为两种算法在其他煤炭基地尺度的进一步应用提供了方法优选借鉴,但是两种算法均存在难以避免的局限性,未来需要研究一种能够准确自适应识别煤炭基地内大范围露天矿群植被损毁事件的新方法。
Coal bases are the concentrated industrial sites of coal resources in China,facing the problems of large areas of vegetation damage and complex and variable causes of destruction.It is of great importance to quantitatively analyze the applicability of the long-time series vegetation damage identification method in coal base to monitor the quality of the local ecological environment.Based on the Landsat images from Google Earth Engine cloud computing platform,this study quantitatively compares the applicability of Land Trendr(LT)and Continuous Change Detection and Classification(CCDC)algorithms for vegetation damage identification in the long-time series(1990—2020)of Shendong coal base in terms of vegetation damage detection,applicability to different surface areas,and vegetation damage time.The study shows that:(1)The overall accuracies of LT and CCDC algorithms are 73.6%and 84.4%,respectively,and the recognition effect is good in vegetation damage detection.(2)Both LT and CCDC algorithms can better avoid the misclassification error in the woodland area.However,the LT algorithm can only identify part of the vegetation damage in the open-pit area,with a large missing error.It is also basically unable to detect the vegetation damage caused by urban expansion,while the CCDC algorithm has a better recognition effect in these two types of areas.In the areas of water body,LT algorithm significantly outperforms CCDC algorithm.(3)The LT and CCDC algorithms have good results in identifying damage time errors within one year,accounting for 95.7%and 90%of the results,respectively.Overall,compared with the LT algorithm,the CCDC algorithm is more suitable for identifying vegetation damage in the Shendong coal base,where urban expansion is obvious and the area of water body is small.The above analysis results provide a data reference for the monitoring of ecological environment quality in Shendong coal base,and also provides a methodological reference for further application of both algorithms in other coal base scales.However,both algorithms have unavoidable limits,and a new method that can accurately and adaptively identify vegetation damage in large-scale open-pit mining groups in coal base needs to be studied in the future.
作者
李军
张艺藂
张彩月
谢慧真
张成业
杜梦豪
王雅颖
LI Jun;ZHANG Yicong;ZHANG Caiyue;XIE Huizhen;ZHANG Chengye;DU Menghao;WANG Yaying(College of Geoscience and Surveying Engineering,China University of Mining and Technology-Beijing,Beijing 100083,China;State Key Laboratory of Coal Resources and Safe Mining,Beijing 100083,China)
出处
《金属矿山》
CAS
北大核心
2023年第1期55-64,共10页
Metal Mine
基金
“十四五”国家重点研发计划项目(编号:2022YFF1303301)
国家自然科学基金项目(编号:42271480,41901291)
中央高校基本科研业务费项目(编号:2022JCCXDC04,2022YQDC08)。
