高级检索

留言板

尊敬的读者、作者、审稿人, 关于本刊的投稿、审稿、编辑和出版的任何问题, 您可以本页添加留言。我们将尽快给您答复。谢谢您的支持!

姓名
邮箱
手机号码
标题
留言内容
验证码

多时相高分辨率遥感影像的森林可燃物分类和变化分析

陈冀岱 牛树奎

陈冀岱, 牛树奎. 多时相高分辨率遥感影像的森林可燃物分类和变化分析[J]. 北京林业大学学报, 2018, 40(12): 38-48. doi: 10.13332/j.1000-1522.20180269
引用本文: 陈冀岱, 牛树奎. 多时相高分辨率遥感影像的森林可燃物分类和变化分析[J]. 北京林业大学学报, 2018, 40(12): 38-48. doi: 10.13332/j.1000-1522.20180269
Chen Jidai, Niu Shukui. Classification and change analysis of forest fuels by multi-temporal high resolution remote sensing images[J]. Journal of Beijing Forestry University, 2018, 40(12): 38-48. doi: 10.13332/j.1000-1522.20180269
Citation: Chen Jidai, Niu Shukui. Classification and change analysis of forest fuels by multi-temporal high resolution remote sensing images[J]. Journal of Beijing Forestry University, 2018, 40(12): 38-48. doi: 10.13332/j.1000-1522.20180269

多时相高分辨率遥感影像的森林可燃物分类和变化分析

doi: 10.13332/j.1000-1522.20180269
基金项目: 

国家林业局科技推广项目 2015-04

详细信息
    作者简介:

    陈冀岱。主要研究方向:生态规划与管理。Email:563580663@qq.com   地址:100083   北京市海淀区清华东路35号北京林业大学林学院

    责任作者:

    牛树奎,教授,博士生导师。主要研究方向:生态规划与管理。Email :niushukui@163.com   地址:同上

  • 中图分类号: S762.1

Classification and change analysis of forest fuels by multi-temporal high resolution remote sensing images

  • 摘要: 目的本文针对国内外利用多时相高分辨率遥感影像进行森林可燃物分类研究匮乏的情况,探索高分辨率影像的分类方法,并研究多时相森林可燃物分类结果的差异,以及与海拔、坡度变化的关系。方法以鹫峰林场为研究区,针对鹫峰林场内植被状况及以往研究成果,主要依据植物群落、林型和燃烧特性划分可燃物类别,研究对比不同森林可燃物类型的光谱特征曲线,建立遥感图像与森林可燃物的联系。选用GF-1号5、8、10月的遥感影像为原始数据,利用EnMAP-box中的支持向量机(SVM)算法、随机森林(RF)以及基于CART的决策树分类方法进行森林可燃物分类,将可燃物类别最终划分为:针叶林、阔叶林、针阔混交林、灌木林和非林地5种类别,并分别对其特征进行描述,之后将最优分类方法应用到多时相的遥感影像中,并使用变化检测算法来确定非防火期(5—10月)森林可燃物类型之间土地面积的变化情况。同时,我们将数字高程模型(DEM)分为4类(1类(< 250 m)、2类(250~500 m)、3类((500~750 m)和4类(>750 m)),坡度分3类:缓坡(< 15°)、斜坡(15°~35°)、陡坡(>35°),并使用Jenks方法分别对海拔和坡度每个类别土地面积变化计算百分比,研究随着海拔和坡度变化,森林可燃物面积的变化规律。结果划分的5种森林可燃物类别的光谱特征具有很好区分性,SVM分类最为准确,取得惩罚参数(C)为1 000和核参数(g)为10使得SVM分类模型达到最优,其总体分类精度为91.88%,Kappa系数为0.89,精度相对RF和CART分别提高了2.72%和9.36%。非防火期内(5—11月)森林可燃物类型变化有一定的规律,针叶林、混交林属于中等稳定类别,没有显著变化,分别保持93.74%和94.87%不变。相比之下,阔叶林和灌木林发生了较大变化,分别发生14.64%和13.36%。随着海拔的增加和坡度的变化,森林可燃物类型土地面积也发生了变化,海拔500~750 m和坡度16°~35°的土地上面积变化最大,达到了20%以上。结论多时相高分辨率遥感影像的森林可燃物分类中,基于SVM分类方法能够将可燃物更好地分类,且随着时间、海拔和坡度的变化,森林可燃物面积的变化有一定的规律,5—10月阔叶林和灌木林在海拔500~750 m和坡度16°~35°变化最大。

     

  • 图  1  不同可燃物类型光谱特征曲线对比

    Figure  1.  Comparison in spectrum characteristic curves of different forest fuel types

    图  2  决策树结构

    Figure  2.  Structure of the decision tree

    图  3  参数优化

    Figure  3.  Parameter optimization

    图  4  由DT、RF、SVM分类算法生成的森林可燃分类

    DT.决策树;RF.随机森林;SVM.支持向量机。下同。

    Figure  4.  Forest fuel classification maps generated by DT, RF, SVM classification algorithms

    DT, decision tree; RF, random forest; SVM, support vector machine. The same below.

    图  5  非防火期森林可燃物各类型覆盖范围饼图

    Figure  5.  Pie chart of forest fuels' coverage during non-fireproof period

    图  6  森林可燃物(5—10月)与海拔变化的柱状图

    Figure  6.  Histogram of changes of forest fuels (May to October) with altitude change

    图  7  森林可燃物(5—10月)与坡度变化的柱状图

    Figure  7.  Histogram of changes of forest fuels (May to October) with slope change

    表  1  GF-1原始数据特征

    Table  1.   Characteristics of GF-1 original data

    卫星
    Satellite
    数据标识
    Data record
    日期
    Data
    云量
    Cloud amount/%
    像素分辨率
    Pixel resolution/m
    GF-1PMS2_E116.1_N40.2-PAN22016-05-161.122
    PMS2_E116.1_N40.2-MSS28
    GF-1PMS2_E116.1_N40.2-PAN22016-08-271.262
    PMS2_E116.1_N40.2-MSS28
    GF-1PMS2_E116.1_N40.2-PAN22016-10-292.32
    PMS2_E116.1_N40.2-MSS28
    下载: 导出CSV

    表  2  鹫峰林场海拔和坡度分级

    Table  2.   Altitude and slope grading of Jiufeng Forest Farm

    海拔Altitude1类Category 1(<250 m)2类Category 2(250-500 m)3类Category 3(500-750 m)4类Category 4(>750 m)
    坡度Slope缓坡Gentle slope(<15°)斜坡Slope(15°-35°)陡坡Steep slope(>35°)
    下载: 导出CSV

    表  3  GF-1遥感影像训练样本的类可分离度

    Table  3.   Class separability of training samples of GF-1 remote image

    分离度
    Separability
    针叶林
    Coniferous
    forest
    阔叶林
    Broadleaved
    forest
    混交林Coniferous
    and broadleaved
    mixed forest
    灌木林
    Shrub
    forest
    非林地
    Non-forest
    land
    针叶林Coniferous forest1.521.451.481.96
    阔叶林Broadleaved forest1.521.701.971.98
    混交林Coniferous and broadleaved mixed forest1.451.701.821.93
    灌木林Shrub forest1.481.971.821.96
    非林地Non-forest land1.961.981.931.96
    下载: 导出CSV

    表  4  2016年GF-1遥感图像经过不同分类的精度对比

    Table  4.   Comparison in accuracy of different classifications to the 2016 GF-1 remote sensing mage

    %
    不同分类Different classifersDTRFSVM
    UAPAUAPAUAPA
    针叶林Coniferous forest64.6551.5669.5756.0080.6871.00
    阔叶林Broadleaved forest85.3980.0691.6363.0393.8573.94
    针阔混交林Coniferous and broadleaved mixed forest65.6980.8973.6795.4679.1190.91
    灌木林Shrub forest90.2494.1387.7591.4590.9093.86
    非林地Non-forest land86.8851.5997.3099.4797.3999.64
    注:UA代表用户精度,PA代表生产者精度。Notes:UA stands for user’s accuracy, PA stands for producer’s accuracy.
    下载: 导出CSV

    表  5  森林可燃物2016年5—10月变化率(ACR)

    Table  5.   Quarterly rate of change (ACR) from May to October in 2016

    %
    类别CategoryACR
    5—8月变化量
    Change of May to August
    8—10月变化量
    Change of August to October
    5—10月变化量
    Change of May to October
    针叶林Coniferous forest6.1-5.01.3
    阔叶林Broadleaved forest3.2-11.0-3.0
    针阔混交林Coniferous and broadleaved mixed forest11.6-11.01.1
    灌木林Shrub forest-6.64.0-2.7
    非林地Non-forest land-5.424.04.7
    注:ACR为每个时期的变化率。Note: ACR is the rate of change.
    下载: 导出CSV

    表  6  森林可燃物与海拔的关系

    Table  6.   Relationship between forest fuels and altitude

    hm2
    ha
    类别
    Category
    海拔
    Altitude/m
    针叶林
    Coniferous
    forest
    阔叶林
    Broadleaved
    forest
    针阔混交林
    Coniferous and
    broadleaved
    mixed forest
    灌木林
    Shrub
    forest
    非林地
    Non-forest
    land
    5月May<2508.513.663.424.594.3
    250~50038.617.388.870.320.6
    500~75015.47.973.679.67.6
    >75020.72.8106.275.63.1
    10月October<2508.210.564.221.3103.3
    250~50039.312.490.655.830.1
    500~75015.65.278.960.610.4
    >75025.71.6110.378.49.6
    森林可燃物类型变化
    Category change in forest fuels
    <250-0.3-3.10.8-3.29.0
    250~5000.7-4.91.8-14.59.5
    500~7500.2-2.75.3-19.02.8
    >7505.0-1.24.12.86.5
    下载: 导出CSV

    表  7  森林可燃物与坡度的关系

    Table  7.   Relationship between forest fuels and slope gradient

    hm2
    ha
    类别Category坡度
    Slope gradient/
    (°)
    针叶林
    Coniferous
    forest
    阔叶林
    Broadleaved
    forest
    针阔混交林
    Coniferous and broadleaved
    mixed forest
    灌木林
    Shrub
    forest
    非林地
    Non-forest
    land
    5月May<157.515.638.720.441.8
    16~3558.617.3228.8129.370.6
    >3617.18.764.5100.313.2
    10月October<156.112.428.912.664.0
    16~3559.88.6190.4120.2145.6
    >3618.42.164.889.628.9
    森林可燃物类型变化
    Category change in forest fuels
    <15-1.4-3.2-9.8-7.8-18.8
    16~351.2-8.7-38.4-49.1-55.0
    >361.3-6.60.3-10.7-13.8
    下载: 导出CSV
  • [1] 胡海清, 罗斯生, 罗碧珍, 等.森林可燃物含水率及其预测模型研究进展[J].世界林业研究, 2017, 30(3): 64-69. http://d.old.wanfangdata.com.cn/Periodical/sjlyyj201703012

    Hu H Q, Luo S S, Luo B Z, et al. Forest fuel moisture content and its prediction model[J]. World Forest Research, 2017, 30(3): 64-69. http://d.old.wanfangdata.com.cn/Periodical/sjlyyj201703012
    [2] 吴德慧, 江洪, 徐建辉.森林可燃物的遥感分类研究进展[J].安徽农业科学, 2012, 40(26): 12961-12962. doi: 10.3969/j.issn.0517-6611.2012.26.088

    Wu D H, Jiang H, Xu J H. Advance in the remote sensing classification of forest fuel[J]. Journal of Anhui Agricultural Sciences, 2012, 40(26): 12961-12962. doi: 10.3969/j.issn.0517-6611.2012.26.088
    [3] Jarron L R, Hermosilla T, Coops N C, et al. Differentiation of alternate harvesting practices using annual time series of landsat data[J]. Forests, 2017, 8(1):13. https://www.mdpi.com/1999-4907/8/1/15
    [4] 郭鸿郡, 俞童, 马超, 等.基于遥感技术的森林可燃物类型划分[J].林业科技情报, 2014, 46(2): 1-3. doi: 10.3969/j.issn.1009-3303.2014.02.001

    Guo H J, Yu T, Ma C, et al. The division for the forest combustible type based on the remote sensing technology[J]. Forestry Science and Technology Information, 2014, 46(2): 1-3. doi: 10.3969/j.issn.1009-3303.2014.02.001
    [5] 赵俊卉, 郭广猛, 张慧东, 等.用MODIS数据预估森林可燃物湿度的研究[J].北京林业大学学报, 2006, 28(6):148-150. doi: 10.3321/j.issn:1000-1522.2006.06.027

    Zhao J B, Guo G M, Zhang H D, et al. Estimating fuel moisture content of forests with MODIS data[J].Journal of Beijing Forestry University, 2006, 28(6):148-150. doi: 10.3321/j.issn:1000-1522.2006.06.027
    [6] 张莹, 张晓丽, 李宏志, 等.基于频谱和光谱特征的高光谱地物分类比较[J].北京林业大学学报, 2018, 40(7):1-8. doi: 10.13332/j.1000-1522.20170342

    Zhang Y, Zhang X L, Li H Z, et al. A comparative analysis on hyperspectral land-cover classification based on frequency spectrum and spectral characteristics[J].Journal of Beijing Forestry University, 2018, 40(7):1-8. doi: 10.13332/j.1000-1522.20170342
    [7] 朱江, 胡华全, 范雯琦, 等.基于深度神经网络的高光谱遥感影像分类方法研究[J].装备学院学报, 2017, 28(3): 14-20. doi: 10.3783/j.issn.2095-3828.2017.03.003

    Zhu J, Hu H Q, Fan W Q, et al. Research on hyperspectral remote sensing image classification based on deep neural networks[J]. Journal of Equipment Academy, 2017, 28(3): 14-20. doi: 10.3783/j.issn.2095-3828.2017.03.003
    [8] Pal M, Mather P M. Support vector machines for classification in remote sensing[J]. International Journal of Remote Sensing, 2005, 26(5): 1007-1011. doi: 10.1080/01431160512331314083
    [9] 李大威, 杨风暴, 王肖霞, 等.基于随机森林与D-S证据合成的多源遥感分类研究[J].激光与光电子学进展, 2016(3): 75-82. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=jgygdzxjz201603010

    Li D W, Yang F B, Wang X X, et al. Multisource remote sensing classification based on random forestand adaptive weighted D-S evidence synthesis[J]. Laser & Optoelectronics Progress, 2016(3): 75-82. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=jgygdzxjz201603010
    [10] 侍昊, 李旭文, 牛志春, 等.基于随机森林模型的太湖水生植被遥感信息提取[J].湖泊科学, 2016(3): 635-644. http://d.old.wanfangdata.com.cn/Periodical/hpkx201603020

    Shi H, Li X W, Niu Z C, et al. Remote sensing information extraction of aquatic vegetation in Lake Taihu based on random forest model[J]. Journal of Lake Sciences, 2016(3): 635-644. http://d.old.wanfangdata.com.cn/Periodical/hpkx201603020
    [11] Otukei J R, Blaschke T. Land cover change assessment using decision trees, support vector machines and maximum likelihood classification algorithms[J]. International Journal of Applied Earth Observation and Geoinformation, 2010, 121: 27-31. http://d.old.wanfangdata.com.cn/NSTLQK/NSTL_QKJJ0216787839/
    [12] 杨长坤, 王崇倡, 张鼎凯, 等.基于SVM的高分一号号卫星影像分类[J].测绘与空间地理信息, 2015(9): 142-144. doi: 10.3969/j.issn.1672-5867.2015.09.049

    Yang C K, Wang C C, Zhang D K, et al. Classification of GF-1 satellite image based on SVM[J]. Geomatics & Spatial Information Technology, 2015(9): 142-144. doi: 10.3969/j.issn.1672-5867.2015.09.049
    [13] 刘毅, 杜培军, 郑辉, 等.基于随机森林的国产小卫星遥感影像分类研究[J].测绘科学, 2012, 37(4): 194-196. http://d.old.wanfangdata.com.cn/Periodical/chkx201204063

    Liu Y, Du P J, Zheng H, et al. Classification of China small satellite remote sensing image based on random forests[J]. Science of Surveying and Mapping, 2012, 37(4): 194-196. http://d.old.wanfangdata.com.cn/Periodical/chkx201204063
    [14] 王凯, 牛树奎.基于Rothermel模型的北京鹫峰国家森林公园潜在火行为[J].浙江农林大学学报, 2016, 33(1): 42-50. http://d.old.wanfangdata.com.cn/Periodical/zjlxyxb201601006

    Wang K, Niu S K. Research on the potential fire behavior in Jiufeng National Forest Park of Beijing based on the Rothermel model[J]. Journal of Zhejiang A & F University, 2016, 33(1): 42-50. http://d.old.wanfangdata.com.cn/Periodical/zjlxyxb201601006
    [15] 林海晏, 岳彩荣, 吴晓晖, 等.基于EnMAP-Box的遥感图像分类研究[J].西南林业大学学报(自然科学), 2014, 34(2): 67-71. http://d.old.wanfangdata.com.cn/Periodical/xnlxyxb201402013

    Lin H Y, Yue C R, Wu X J, er al. Remote sensing image classification by EnMAP-Box model[J]. Journal of Southwest Forestry University(Natural Sciences), 2014, 34(2): 67-71. http://d.old.wanfangdata.com.cn/Periodical/xnlxyxb201402013
    [16] 米锋, 黄莉莉, 孙丰军, 等.北京鹫峰国家森林公园生态安全评价[J].林业科学, 2010, 46(11): 52-58. doi: 10.11707/j.1001-7488.20101107

    Mi F, Huang L L, Sun F J, et al. Evaluation of ecological security for the Jiufeng National Forest Park in Beijing[J]. Scientia Silvae Sinicae, 2010, 46(11): 52-58. doi: 10.11707/j.1001-7488.20101107
    [17] 杨闫君, 占玉林, 田庆久, 等.基于GF-1/WFV NDVI时间序列数据的作物分类[J].农业工程学报, 2015, 31(24): 155-161. doi: 10.11975/j.issn.1002-6819.2015.24.024

    Yang Y J, Zhan Y L, Tian Q J, et al. Crop classification based on GF-1/WFV NDVI time series[J]. Transactions of the Chinese Society of Agricultural Engineering, 2015, 31(24): 155-161. doi: 10.11975/j.issn.1002-6819.2015.24.024
    [18] Breiman L. Random forests[J]. Machine Leaning, 2001, 45:5-32. doi: 10.1023/A%3A1010933404324
    [19] Tian F, Yang L, Lü F, et al. Predicting liquid chromatographic retention times of peptides from the Drosophila melanogaster proteome by machine learning approaches[J]. Analytica Chimica Acta, 2009, 644(1-2): 10-16. doi: 10.1016/j.aca.2009.04.010
    [20] Waske B, Van Der Linden S, Oldenburg C, et al. image RF: a user-oriented implementation for remote sensing image analysis with random forests[J]. Environmental ModellinG & Software, 2012, 35: 192-193.
    [21] Kuemmerle R, Steder B, Dornhege C, et al. On measuring the accuracy of SLAM algorithms[J]. Autonomous Robots, 2009, 27(4): 387-407. doi: 10.1007/s10514-009-9155-6
    [22] Manandhar R, Odeh I O A, Ancev T. Improving the accuracy of land use and land cover classification of landsat data using post-classification enhancement[J]. Remote Sensing, 2009, 1(3): 330-344. doi: 10.3390/rs1030330
    [23] Yuan F, Sawaya K E, Loeffelholz B C, et al. Land cover classification and change analysis of the Twin Cities (Minnesota) Metropolitan Area by multitemporal landsat remote sensing[J]. Remote Sensing of Environment, 2005, 98(2-3): 317-328. doi: 10.1016/j.rse.2005.08.006
    [24] Petropoulos G P, Vadrevu K P, Xanthopoulos G, et al. A comparison of spectral angle mapper and artificial neural network classifiers combined with landsat TM imagery analysis for obtaining Burnt Area Mapping[J]. Sensors, 2010, 10(3): 1967-1985. doi: 10.3390/s100301967
    [25] Keshtkar H, Voigt W, Alizadeh E. Land-cover classification and analysis of change using machine-learning classifiers and multi-temporal remote sensing imagery[J]. Arabian Journal of Geosciences, 2017, 10(6): 154-169. doi: 10.1007/s12517-017-2899-y
    [26] 张浪, 朱义, 薛建辉, 等.转型期园林绿化的城市困难立地类型划分研究[J].现代城市研究, 2017(9): 114-118. doi: 10.3969/j.issn.1009-6000.2017.09.016

    Zhang L, Zhu Y, Xue J H, et al. Classification study of challenging urban sites in landscape gardening during urban transition period[J]. Modern Urban Research, 2017(9): 114-118. doi: 10.3969/j.issn.1009-6000.2017.09.016
  • 加载中
图(7) / 表(7)
计量
  • 文章访问数:  694
  • HTML全文浏览量:  199
  • PDF下载量:  39
  • 被引次数: 0
出版历程
  • 收稿日期:  2018-08-21
  • 修回日期:  2018-10-12
  • 刊出日期:  2018-12-01

目录

    /

    返回文章
    返回