Please wait a minute...
img

官方微信

遥感技术与应用  2021, Vol. 36 Issue (4): 847-856    DOI: 10.11873/j.issn.1004-0323.2021.4.0847
遥感应用     
基于Sentinel-2数据的天山山地针叶林识别方法研究
蒋嘉锐1,2(),朱文泉1,2(),乔琨1,2,江源1
1.北京师范大学,地表过程与资源生态国家重点实验室,北京 100875
2.北京师范大学地理科学学部,遥感科学与工程研究院,北京市陆表遥感数据产品工程技术研究中心,北京 100875
An Identification Method for Mountains Coniferous in Tianshan with Sentinel-2 Data
Jiarui Jiang1,2(),Wenquan Zhu1,2(),Kun Qiao1,2,Yuan Jiang1
1.State Key Laboratory of Earth Surface Processes and Resource Ecology,Beijing Normal University,Beijing 100875,China
2.Beijing Engineering Research Center for Global Land Remote Sensing Products,Institute of Remote Sensing Science and Engineering,Faculty of Geographical Science,Beijing Normal University,Beijing 100875,China
 全文: PDF(6554 KB)   HTML
摘要:

阴影是影响山地针叶林遥感识别精度的关键因素。选取天山一块面积约为10 000 km2的区域为案例,基于太阳高度角和方位角差异较大的两期Sentinel-2影像,从遥感数据阴影分布的时相特性、分类特征以及分类器选择三方面进行综合分析,提出了一种适用于天山山地针叶林的遥感综合分类方案。该综合分类方案首先开展阴影识别以及阴影再分类以排除阴影对针叶林识别的影响;然后筛选出了海拔、归一化差值植被指数(NDVI)、红光到近红外波段斜率、蓝光波段、红光波段、短波红外波段和坡度作为区分天山山地针叶林的重要特征;最后比较支持向量机(Support Vector Machine,SVM)、随机森林(Random Forest,RF)和BP神经网络(Back Propagation Neural Network,BPNN)3种分类器的分类效果。结果表明:采用地形校正方法来消除山体阴影的效果不但不明显,反而还会造成过矫正现象,从而影响后续的针叶林识别,但利用太阳高度角和方位角差异较大的两期影像开展阴影识别以及阴影再分类来排除阴影对针叶林识别的影响,可使针叶林的总体精度提高1.3%~3.7%;SVM、RF和BPNN 3种分类器都能取得较好的山地针叶林识别精度,但SVM分类器的分类精度最高,其总体分类精度和Kappa系数分别是93.33%和0.87。该遥感综合分类方案经参数调整之后有望应用于北方干旱半干旱区的其他山地针叶林区域。

关键词: 天山山地针叶林阴影特征选择多光谱遥感    
Abstract:

Shadows are the key factors affecting the identification accuracy of mountains coniferous forests using multi-spectral remote sensing data. Taking Tianshan as the study area, a comprehensive classification scheme was proposed, which comprehensive considered three aspects: the time-phase features of shadow distribution, classification features and classifiers. Firstly, to eliminate the influences of shadows on coniferous forest identification, shadow recognition and shadow reclassification were carried out. Then the altitude, Normalized Difference Vegetation Index (NDVI), spectral slope of red to near-infrared band, blue reflectance band, red reflectance band, short-wave infrared band and slope were selected as the important features for identifying the Tianshan mountain coniferous forest. Finally, three often used classifiers (Support Vector Machine (SVM), Random Forest (RF) and Back Propagation Neural Network (BPNN)) were compared. The results show that the terrain correction method can not effectively eliminate the mountain shadows, and it may cause over-correction, which affects the subsequent identification of coniferous. However, using two-phase images with large differences in solar elevation and azimuth to eliminate the influence of shadows on coniferous forest identification can improve the overall accuracy of coniferous forest by 1.3% to 3.7%; The SVM, RF and BPNN classifier can all achieve better classification accuracy, but the SVM classifier got the highest classification accuracy and Kappa coefficient with a value 93.33% and 0.87, respectively. The proposed remote sensing comprehensive classification scheme is expected to be applied to other mountain coniferous forest areas in the north arid and semi-arid regions after adjusting the parameters.

Key words: Tianshan    Mountains coniferous forest    Shadow    Feature selection    Multispectral remote sensing
收稿日期: 2019-12-30 出版日期: 2021-09-26
ZTFLH:  TP701  
基金资助: 国家自然科学基金重点项目“中国北方山地针叶林生长的时空分异及其对水热条件的响应”(41630750)
通讯作者: 朱文泉     E-mail: jiangjr2019@163.com;zhuwq75@bnu.edu.cn
作者简介: 蒋嘉锐(1995-),男,云南曲靖人,硕士研究生,主要从事植被遥感、植被物候以及土地覆盖分类研究。E?mail:jiangjr2019@163.com
服务  
把本文推荐给朋友
加入引用管理器
E-mail Alert
RSS
作者相关文章  
蒋嘉锐
朱文泉
乔琨
江源

引用本文:

蒋嘉锐,朱文泉,乔琨,江源. 基于Sentinel-2数据的天山山地针叶林识别方法研究[J]. 遥感技术与应用, 2021, 36(4): 847-856.

Jiarui Jiang,Wenquan Zhu,Kun Qiao,Yuan Jiang. An Identification Method for Mountains Coniferous in Tianshan with Sentinel-2 Data. Remote Sensing Technology and Application, 2021, 36(4): 847-856.

链接本文:

http://www.rsta.ac.cn/CN/10.11873/j.issn.1004-0323.2021.4.0847        http://www.rsta.ac.cn/CN/Y2021/V36/I4/847

图1  研究区地理位置(图中影像为2018年9月19日RGB真彩色合成的Sentinel?2影像;使用反地形校正方法[13]突出了山体地形起伏)
图2  研究区训练样本点分布图(图中影像为2018年9月19日RGB真彩色合成的Sentinel?2影像)
图3  技术路线图
图4  不同时相阴影对比图
图5  地形校正前后效果比较
图6  地形校正前后地物光谱曲线(阴影范围代表该类地物反射率的均值±标准差;雪的反射率因在600~1 200 nm太大未完整显示)
序号z-score变量名称序号z-score变量名称
13.03DEM90.94B7
21.54NDVI100.86B8A
31.19斜率(B4~B8)110.78B6
41.06B2120.77B5
51.05B4130.72B8
61.01B11140.67B3
71.01坡度150.36坡向
80.97B9
表1  特征重要性排序表
图7  分类结果局部对比
图8  天山山地针叶林识别结果
SVMRFBPNN
生产者精度%

用户

精度%

生产者精度%

用户

精度%

生产者精度%

用户

精度%

不考虑阴影针叶林88.2290.0184.9689.6685.6287.92
非针叶林92.1890.7692.1988.5090.6388.78
总分类精度%90.4389.9988.40
Kappa系数0.800.770.76
SVMRFBPNN
生产者精度%

用户

精度%

生产者精度%

用户

精度%

生产者精度%

用户

精度%

考虑阴影针叶林93.4691.6788.2491.8493.1588.89
非针叶林93.2394.7193.7590.9194.7991.46
总分类精度%93.3391.3092.17
Kappa系数0.870.820.84
表2  针叶林遥感识别结果的精度评价
图9  不考虑阴影影响下的针叶林识别结果
1 Huang H, Gong P, Clinton N, et al. Reduction of atmospheric and topographic effect on Landsat TM data for forest classification[J]. International Journal of Remote Sensing,2008,29(19):5623-5642.DOI:10.1080/01431160802082148.
doi: 10.1080/01431160802082148
2 Vanonckelen S, Lhermitte S, Van Rompaey A. The effect of atmospheric and topographic correction on pixel-based image composites: improved forest cover detection in mountain environments[J]. International Journal of Applied Earth Observation and Geoinformation, 2015, 35: 320-328. DOI:10.1016/j.jag.2014.10.006.
doi: 10.1016/j.jag.2014.10.006
3 Zhou Jianhua, Zhou Yifan, Guo Xiaohua, et al. Methods of extracting distribution information of plants at urban darken areas and repairing their brightness[J]. Journal of East China Normal University(Natural Science Edition), 2011,2011(6): 1-9.
3 周坚华,周一凡,郭晓华,等. 城镇阴影区植物分布信息提取与亮度修复的方法[J].华东师范大学学报(自然科学版),2011,2011(6): 1-9.
4 Deng Lin, Deng Mingjing, Zhang Lishu. Optimization of shadow detection and compensation method for high-resolution remote sensing images[J]. Remote Sensing Technology and Application, 2015, 30(2): 277-284.
4 邓琳,邓明镜,张力树. 高分辨率遥感影像阴影检测与补偿方法优化[J]. 遥感技术与应用, 2015, 30(2): 277-284.
5 Gao Yongnian, Zhang Wanchang. Simplification and modification of a physical topographic correction algorithm for remotely sensed data[J]. Acta Geodaetica et Cartographica Sinica, 2008, 37(1): 89-94, 120.
5 高永年,张万昌. 遥感影像地形校正物理模型的简化与改进[J]. 测绘学报, 2008, 37(1): 89-94, 120.
6 Lu Lili, Xie Yaowen, Dong Longlong. The comparison of reflectance based on different terrain correction[J]. Remote Sensing Technology and Application, 2017, 32(4): 751-759.
6 吕利利,颉耀文,董龙龙. 基于不同地形校正模型的影像反射率对比分析[J]. 遥感技术与应用,2017,32(4):751-759.
7 Bian Jinhu, Li Ainong, Wang Shaonan, et al. Restoration of information obscured by mountain shadows for Landsat TM images based on MODIS NDVI[J]. Remote Sensing Technology and Application, 2016, 31(1): 12-22.
7 边金虎,李爱农,王少楠,等. 基于MODIS NDVI的Landsat TM影像地形阴影区光谱信息恢复方法研究[J]. 遥感技术与应用, 2016, 31(1): 12-22.
8 Wang Genxu, Liu Guohua, Shen Zehao, et al. Research progress and future perspectives on the landscape ecology of mountainous areas[J]. Acta Ecologica Sinica, 2017, 37(12): 3967-3981.
8 王根绪,刘国华,沈泽昊,等. 山地景观生态学研究进展[J]. 生态学报, 2017, 37(12): 3967-3981.
9 Xie Qiaoyun. Research on leaf area index retrieve methods based on the red edge bands from multi-platform remote sensing data[D]. Beijing: Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences, 2017.
9 谢巧云. 考虑红边特性的多平台遥感数据叶面积指数反演方法研究[D]. 北京:中国科学院遥感与数字地球研究所, 2017.
10 Zhang Weichun, Liu Hongbin, Wu Wei. Classification of land use in low mountain and hilly area based on random forest and Sentinel-2 satellite data: a case study of Lishi town, Jiangjin, Chongqing[J]. Resources and Environment in the Yangtze Basin, 2019, 28(6): 1334-1343.
10 张卫春,刘洪斌,武伟. 基于随机森林和Sentinel-2影像数据的低山丘陵区土地利用分类——以重庆市江津区李市镇为例[J]. 长江流域资源与环境, 2019, 28(6): 1334-1343.
11 Li Meng, Yanyun Nian , Bian Rui, et al. Classification of picea crassifolia and sabina przewalskii based on multi-source remote sensing images[J]. Remote Sensing Technology and Application, 2020, 35(4): 855-863.
11 李萌,年雁云,边瑞,等. 基于多源遥感影像的青海云杉和祁连圆柏分类[J]. 遥感技术与应用, 2020, 35(4): 855-863.
12 Liu Guifeng, Ding Yi, Zang Runguo, et al. Distribution patterns of picea schrenkiana Var. tianschanica population in Tianshan mountains[J]. Chinese Journal of Applied Ecology, 2011, 22(1): 9-13.
12 刘贵峰,丁易,臧润国,等. 天山云杉种群分布格局[J]. 应用生态学报, 2011, 22(1): 9-13.
13 Zhang D, Zhu W, Chen X, et al. A correction technique for false topographic perception of remote sensing images based on an inverse topographic correction technique[J]. International Journal of Digital Earth, 2016, 9(10): 1021-1034. DOI:10.1080/17538947.2016.1187672
doi: 10.1080/17538947.2016.1187672
14 Kim H, Yeom J. Effect of red-edge and texture features for object-based paddy rice crop classification using rapid eye multi-spectral satellite image data[J]. International Journal of Remote Sensing, 2014, 35(19): 7046-7068. DOI:10.1080/01431161.2014.965285.
doi: 10.1080/01431161.2014.965285
15 Schuster C, Förster M, Kleinschmit B. Testing the red ddge channel for improving land-use classifications based on high-resolution multi-spectral satellite data[J]. International Journal of Remote Sensing, 2012, 33(17): 5583-5599. DOI:10.1080/01431161.2012.666812.
doi: 10.1080/01431161.2012.666812
16 Breiman L. Random forests[J]. Machine Learning, 2001, 45(1): 5-32.
17 Gong P, Wang J, Yu L, et al. Finer fesolution observation and monitoring of global land cover: first mapping results with landsat TM and ETM+ data[J]. International Journal of Remote Sensing,2013,34(7):48. DOI:10.1080/01431161. 2012.748992.
doi: 10.1080/01431161. 2012.748992
18 Wen Xiaole, Zhong Ao, Hu Xiujuan. The classification of urban greening tree species based on feature selection of random forest[J]. Journal of Geo-information Science, 2018, 20(12): 1777-1786.
18 温小乐,钟奥,胡秀娟. 基于随机森林特征选择的城市绿化乔木树种分类[J]. 地球信息科学学报, 2018, 20(12): 1777-1786.
19 Lu D, Weng Q. A survey of image classification methods and techniques for improving classification performance[J]. International Journal of Remote Sensing, 2007, 28(5): 823-870. DOI:10.1080/01431160600746456.
doi: 10.1080/01431160600746456
20 Sothe C, De Almeida C M, Schimalski M B, et al. Comparative performance of convolutional neural network, weighted and conventional support vector machine and random forest for classifying tree species using hyperspectral and photogrammetric data[J]. GIScience & Remote Sensing, 2020, 57(3): 369-394. DOI:10.1080/15481603.2020.1712102.
doi: 10.1080/15481603.2020.1712102
21 Noi P T, Kappas M. Comparison of fandom forest, k-nearest neighbor, and support vector machine classifiers for land cover classification using Sentinel-2 imagery[J]. Sensors, 2018, 18(1): 18. DOI:10.3390/s18010018.
doi: 10.3390/s18010018
22 Heydari S S, Mountrakis G. Effect of classifier selection, reference sample size, reference class distribution and scene heterogeneity in per-pixel classification aaccuracy using 26 Landsat sites[J]. Remote Sensing of Environment, 2017, 204: 648-658. DOI:10.1016/j.rse.2017.09.035.
doi: 10.1016/j.rse.2017.09.035
23 van der Linden S, Rabe A, Held M, et al. The enMAP-box—a toolbox and application programming interface for enMAP data processing[J]. Remote Sensing, 2015, 7(9): 11249-11266. DOI:10.3390/rs70911249.
doi: 10.3390/rs70911249
24 Ji Xuan, Chen Yunfang, Luo Xian, et al. Study on the identification method of glacier in mountain shadows based on Landsat 8 OLI image[J]. Spectroscopy and Spectral Analysis. 2018, 38(12): 3857-3863.季漩,陈云芳,罗贤,等. Landsat 8 OLI影像的高原山地阴影区冰川识别方法[J]. 光谱学与光谱分析, 2018, 38(12): 3857-3863.
25 Shahtahmassebi A, Yang N, Wang K, et al. Review of shadow detection and de-shadowing methods in remote sensing[J]. Chinese Geographical Science, 2013, 23(4): 403-420. DOI:10.1007/s11769-013-0613-x.
doi: 10.1007/s11769-013-0613-x
26 Purevdorj T, Tateishi R, Ishiyama T, et al. Relationships between percent vegetation cover and vegetation indices[J]. International Journal of Remote Sensing, 1998, 19(18): 3519-3535. DOI:10.1080/014311698213795.
doi: 10.1080/014311698213795
27 Forkuor G, Dimobe K, Serme I, et al. Landsat-8 vs. Sentinel-2: dxamining the added value of Sentinel-2's red-edge bands to land-use and land-cover mapping in Burkina Faso[J]. GIScience & Remote Sensing, 2017, 55(4): 331-354. DOI:10.1080/15481603.2017.1370169.
doi: 10.1080/15481603.2017.1370169
28 Sothe C, De Almeida C M, Liesenberg V, et al. Evaluating Sentinel-2 and Landsat 8 data to map sucessional forest stages in a subtropical forest in Southern Brazil[J]. Remote Sensing, 2017, 9(8): 838. DOI:10.3390/rs9080838.
doi: 10.3390/rs9080838
29 Liu Yi, Du Peijun, Zheng Hui, 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.
29 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.
30 Chen Tao, Niu Ruiqing, Li Pingxiang, et al. An artificial neural network method for vegetation cover retrieval with "Beijing-1" microsatellite data[J]. Remote Sensing Technology and Application, 2010, 25(1): 24-30.
30 陈涛,牛瑞卿,李平湘,等. 基于人工神经网络的植被覆盖遥感反演方法研究[J]. 遥感技术与应用, 2010, 25(1): 24-30.
31 Chen Bingmei, Fan Xiaoping, Zhou Zhiming, et al. The principle and prospect of Support Vector Machine [J]. Manufacturing Automation, 2010, 32(14): 136-138.
31 陈冰梅,樊晓平,周志明,等. 支持向量机原理及展望[J]. 制造业自动化, 2010, 32(14): 136-138.
32 Liu Tianfu, Chen Xuehong, Dong Qi, et al. Application of deep learning in globeLand 30-2010 product refinement[J]. Remote Sensing Technology and Application, 2019, 34(4): 685-693.
32 刘天福,陈学泓,董琪,等. 深度学习在GlobeLand30-2010产品分类精度优化中应用研究[J]. 遥感技术与应用, 2019, 34(4): 685-693.
33 Simonyan K, Zisserman A. Very deep convolutional networks for large-scale image recognition[J]. ArXiv E-Prints, 2014: 1409-1556.
[1] 刘鹤,顾玲嘉,任瑞治. 基于无人机遥感技术的森林参数获取研究进展[J]. 遥感技术与应用, 2021, 36(3): 489-501.
[2] 廖鸿燕,周小成,黄洪宇. 基于无人机遥感技术的台风灾害倒伏绿化树木检测[J]. 遥感技术与应用, 2021, 36(3): 533-543.
[3] 李强,冯德俊,瑚敏君,伍燚垚,杨历辉. 集成特征分量的高分二号影像阴影检测[J]. 遥感技术与应用, 2019, 34(6): 1252-1260.
[4] 张少伟,惠刚盈,韩宗涛,孙珊珊,田昕. 基于光学多光谱与SAR遥感特征快速优化的大区域森林地上生物量估测[J]. 遥感技术与应用, 2019, 34(5): 925-938.
[5] 丁哲,汪小钦,邬群勇. 遥感影像空间分辨率对城市建筑物高度估算精度的影响[J]. 遥感技术与应用, 2018, 33(3): 418-427.
[6] 赖日文,池毓锋,张泽均. 基于亮度恢复模型的Landsat 8数据山区阴影去除[J]. 遥感技术与应用, 2018, 33(3): 563-572.
[7] 刘慧珺,苏红军,赵-波. 基于改进萤火虫算法的高光谱遥感多特征优化方法[J]. 遥感技术与应用, 2018, 33(1): 110-118.
[8] 吴迪,史文中,高利鹏,张华,何鹏飞. 一种改进的基于上下文信息的多源数据融合目标提取方法[J]. 遥感技术与应用, 2018, 33(1): 128-135.
[9] 田峰,陈冬花,黄新利,李虎,姚国慧. 基于形态学阴影指数的高分二号影像建筑物高度估计#br#[J]. 遥感技术与应用, 2017, 32(5): 844-850.
[10] 唐志光,王建,王欣,彭焕华,梁继. 近15年天山地区积雪时空变化遥感研究[J]. 遥感技术与应用, 2017, 32(3): 556-563.
[11] 李炳亚,潘剑君,夏超,陈昕,隋传嘉. 基于空间位置关系的山地湖泊水体提取方法研究[J]. 遥感技术与应用, 2016, 31(5): 983-993.
[12] 边金虎,李爱农,王少楠,赵伟,雷光斌. 基于MODIS NDVI的Landsat TM影像地形阴影区光谱信息恢复方法研究[J]. 遥感技术与应用, 2016, 31(1): 12-22.
[13] 姜爱辉,陈富龙,刘国林,代杰. ENVISAT ASAR相干目标的分类识别探测试验与分析—以世界文化遗产都江堰为例[J]. 遥感技术与应用, 2015, 30(5): 842-848.
[14] 邓琳,邓明镜,张力树. 高分辨率遥感影像阴影检测与补偿方法优化[J]. 遥感技术与应用, 2015, 30(2): 277-284.
[15] 陈阳,范建容,文学虎,曹伟超,王蕾. 基于时空数据融合模型的TM影像云去除方法研究[J]. 遥感技术与应用, 2015, 30(2): 312-320.