李恒凯, 雷 军, 杨 柳. 基于Landsat影像的离子稀土矿区植被覆盖度提取及景观格局分析[J]. 农业工程学报, 2016, 32(10): 267-276. DOI: 10.11975/j.issn.1002-6819.2016.10.037
    引用本文: 李恒凯, 雷 军, 杨 柳. 基于Landsat影像的离子稀土矿区植被覆盖度提取及景观格局分析[J]. 农业工程学报, 2016, 32(10): 267-276. DOI: 10.11975/j.issn.1002-6819.2016.10.037
    Li Hengkai, Lei Jun, Yang Liu. Extraction of vegetation coverage and analysis of landscape pattern in rare earth mining area based on Landsat image[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2016, 32(10): 267-276. DOI: 10.11975/j.issn.1002-6819.2016.10.037
    Citation: Li Hengkai, Lei Jun, Yang Liu. Extraction of vegetation coverage and analysis of landscape pattern in rare earth mining area based on Landsat image[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2016, 32(10): 267-276. DOI: 10.11975/j.issn.1002-6819.2016.10.037

    基于Landsat影像的离子稀土矿区植被覆盖度提取及景观格局分析

    Extraction of vegetation coverage and analysis of landscape pattern in rare earth mining area based on Landsat image

    • 摘要: 离子稀土开采导致矿区植被退化和严重的生态环境问题,日益引起人们关注。采用23 a的Landsat系列影像作为数据源,以定南县岭北稀土矿区作为研究案例,对稀土开采扰动下的矿区植被覆盖景观格局变化进行分析。为提高分析的可靠性,以高分遥感影像获取的矿区植被覆盖度作为检验数据,比较森林郁闭度制图模型(forest canopy density mapping model,FCD)、像元二分法模型(dimidiate pixel model,DPM)和线性光谱混合模型(linear spectral mixture model,LSMM)对稀土矿区植被覆盖提取的准确性。结果表明:考虑了稀土矿区特有的裸露尾沙地光谱特征的三端元LSMM,具有更高准确性及普适性;景观格局动态分析表明矿区低植被覆盖和高植被覆盖度破碎化程度较低,集中连片趋势明显;低植被覆盖LPI(largest patch index)在4个年份相对较大,呈波动变化,主要为裸露的稀土矿点,与矿点复垦和扩张有密切关系,且单个矿点造成的植被破坏仍然较大;低植被覆盖度的AI(aggregation index)一直较高,而LPI相对高植被覆盖较小,主要为离子稀土以单个矿点小面积开采为主,矿点多而分散,导致对环境破坏更难治理;1999年后,随着稀土开采的规模化及开采工艺的改进,稀土开采成为矿区景观变化的最主要原因,在矿区总体生态环境改善的前提下,低植被覆盖区域主要集中在矿点及矿点周边,成为具有稀土开采特色的矿点景观分布模式。

       

      Abstract: Mining rare earth ore has brought vegetation degradation, serious ecological and environmental problems in mining area, and it has already received more and more attentions and concerns by the society. By using 23 years Landsat series remote sensing images as data source, and choosing Lingbei rare earth mining area in Dingnan County as study area, in this paper, we analyzed the change of vegetation coverage of landscape pattern under the disturbance of rare earth mining. In order to improve the reliability of analyzing landscape pattern, the vegetation coverage of study area was extracted with four methods of forest canopy density mapping model(FCD), dimidiate pixel model(DPM), three-endmember and four-endmember linear spectral mixture model(LSMM). Taking vegetation coverage of the studied area acquired from the high-spatial-resolution remote sensing image of Pleiades as the checking data, the most suitable method of vegetation coverage in rare earth mining area was selected. And the change of landscape pattern was analyzed with this method. In addition, to prove the universality of this method, vegetation coverage of Heling rare earth mining area in Xunwu County was also extracted with methods of FCD model, DPM and three-endmember LSMM. In the test, we selected randomly 30 samples from a 0.5 meter high-spatial-resolution aerial image as the checking data. The result showed that the correlation coefficients of FCD model, DPM, three-endmember and four-endmember LSMM were 0.900 8, 0.924 7, 0.980 4 and 0.946 5, respectively. The Root Mean Square Error(RMSE) of FCD model, DPM, three-endmember and four-endmember LSMM were 0.341 1, 0.243 4, 0.037 8 and 0.089 3, respectivley. Similar results were obtained from the Heling rare earth mining area in Xunwu County. These two parameters of the correlation coefficients and RMSE could clearly show that the three-endmember LSMM was more suitable and accuracy for the extraction of vegetation coverage due to considering the spectral characteristics of peculiar tail sand in rare earth mining area. The landscape pattern dynamic analysis showed that fragmentation of low-level vegetation coverage and high-level vegetation coverage was at the low level in the mining area, concentrated and connected trend was obvious. And patch density(PD) value of the other three levels′ vegetation coverage were higher, they existed some ecological risks. The Largest Patch Index(LPI) of low-level vegetation coverage was relatively large in four years, and showed the fluctuating change. The area was mainly constituted by the bare of rare occurrences, and it had close relationship between ore occurrences′ reclamation and expanding. When LPI reached 0.543 6 and its area was 1 km2 in 2013, vegetation destruction caused by an individual mine was still larger. When the LPI of high-level vegetation coverage had a more dramatic change, and it reached the maximum of 80.519 0 in 2008, the area of high-level vegetation coverage increased in patches of ways due to people strengthened on ecological governance of rare earth mining area. When around small patches of low-level vegetation coverage gradually transformed into high-level vegetation coverage, high-level vegetation coverage patches became the entire mining matrix patches. The Aggregation Index(AI) of low-level vegetation coverage was consistently high, and its LPI was lower than high-level vegetation coverage. When the ionic rare earth was mined mainly in individual occurrence of small area, and ore occurrences were many and dispersed, this led environmental damage that were more difficult to control. After year 1999, with the scale expansion of mining and its process improvement, rare earth mining became the main cause for mining area′s landscape change. On the premise of overall improvement of ecological environment in mining area, the low-level vegetation coverage area concentrated in the ore occurrences and surrounding ore occurrences become landscape distribution pattern with rare earth mining area features of ore occurrences.

       

    /

    返回文章
    返回