Analysis of the potential of GF-6 WFV data for forest and non-forest land identification

doi:10.13474/j.cnki.11-2246.2020.0239

Abstract

Abstract: In order to study the recognition contribution of the first GaoFen (GF-6) with red edge band in China on forest and non-forest land, this paper selectes Huangshan city, Anhui province, as a research area, and adopt feature optimization (RFE). The method combined with random forest (RF) has carried out research on the identification potential of forest and non-forest land. First it builds a sample database based on field surveys, Google Earth images, and forest land "one map" sample data. Then, it classifies based on characteristics such as DEM, multitemporal spectral features, vegetation index, red edge index, and compares the accuracy of different models and the importance of different variable. The results show that the red edge information of GF-6 is more important for the identification of forest and non-forest land. The introduction of red edge information can improve the overall classification accuracy by 2%. Compared with single-phase data, the use of data can improve the classification accuracy of forest types by 2.93%~4.1%. The single-phase classification results are the best in June, the second in September, and the worst in December. The preferred features can effectively reduce data input Dimensions (46 to 15), and achieve the highest classification accuracy. It does not sacrifice accuracy while ensuring a reduction in the amount of computing data and clarifying the contribution of different variables, which has strong application significance.

Key words: GF-6, forest and non-forest land, RFE, random forest, feature importance

CLC Number:

P237

LIU Daichao, LI Xiaosong, LI Xiangchen, YANG Guangbin, YANG Junting. Analysis of the potential of GF-6 WFV data for forest and non-forest land identification[J]. Bulletin of Surveying and Mapping, 2020, 0(8): 5-12,17.

References

[1] WULDER M A, BATER C W, COOPS N C, et al. The role of LiDAR in sustainable forest management[J]. Forestry Chronicle,2008, 84(6):807-826.
[2] GROOT A, CORTINI F, WULDER M A. Crown-fibre attribute relationships for enhanced forest inventory:progress and prospects[J]. The Forestry Chronicle, 2015,91(3):266-279.
[3] THOMPSON I D, MAHER S C, ROUILLARD D P, et al. Accuracy of forest inventory mapping:some implications for boreal forest management[J]. Forest Ecology and Management, 2007, 252(1):208-221.
[4] WHITE J C, COOPS N C, WULDER M A, et al. Remote sensing technologies for enhancing forest inventories:a review[J]. Canadian Journal of Remote Sensing, 2016, 42(5):619-641.
[5] TOKOLA T, SARKEALA J, LINDEN M V D. Use of topographic correction in Landsat TM-based forest interpretation in Nepal[J]. International Journal of Remote Sensing, 2001, 22(4):551-563.
[6] JIA K, LIANG S, ZHANG L, et al. Forest cover class-ification using Landsat ETM+ data and time series MODIS NDVI data[J]. International Journal of Applied Earth Observation and Geoinformation, 2014, 33:32-38.
[7] HILL R A, WILSON A K, GEORGE M, et al. Mapping tree species in temperate deciduous woodland using time-series multi-spectral data[J]. Applied Vegetation Science, 2010, 13(1):86-99.
[8] TRUONG V T, HOANG T T, CAO D P, et al. JAXA annual forest cover maps for vietnam during 2015-2018 using ALOS-2/PALSAR-2 and auxiliary data[J]. Remote Sensing, 2019, 11(20):2412.
[9] ISUHUAYLAS L A V, HIRATA Y, SANTOS L V, et al. Natural forest mapping in the Andes (Peru):a comparison of the performance of machine-learning algorithms[J]. Remote Sensing, 2018, 10(5):782.
[10] XIE Z, CHEN Y, LU D, et al. Classification of land cover, forest, and tree species classes with ZiYuan-3 multispectral and stereo data[J]. Remote Sensing, 2019, 11(2):164.
[11] 张兆鹏. 基于多源遥感数据的林地类型精细识别与变化监测研究[D].西安:西安科技大学,2018.
[12] GRABSKA E, HOSTERT P, PFLUGMACHER D, et al. Forest stand species mapping using the Sentinel-2 time series[J]. Remote Sensing, 2019, 11(10):1197.
[13] HOŠCIŁO A, LEWANDOWSKA A. Mapping forest type and tree species on a regional scale using multi-temporal Sentinel-2 data[J]. Remote Sensing, 2019, 11(8):929.
[14] PASQUARELLA V J, HOLDEN C E, WOODCOCK C E. Improved mapping of forest type using spectral-temporal Landsat features[J]. Remote Sensing of Environment, 2018, 210(1):193-207.
[15] KIM M, MADDEN M, WARNER T A. Forest type mapping using object-specific texture measures from multispectral ikonos imagery[J]. Photogrammetric Engineering & Remote Sensing, 2009(11):819-829.
[16] PAX-LENNEY M, WOODCOCK C E, MACOMBER S A, et al. Forest mapping with a generalized classifier and Landsat TM data[J]. Remote Sensing of Environment, 2001, 77(3):241-250.
[17] JIA M, WANG Z, WANG C, et al. A new vegetation index to detect periodically submerged mangrove forest using single-tide Sentinel-2 imagery[J]. Remote Sensing, 2019, 11(17):2043.
[18] MARTIN M E, NEWMAN S D, ABER J D, et al. Determining forest species composition using high spectral resolution remote sensing data[J]. Remote Sensing of Environment, 1998, 65(3):249-254.
[19] 郑利娟. 基于高分一/六号卫星影像特征的农作物分类研究[D].北京:中国科学院大学(中国科学院遥感与数字地球研究所),2017.
[20] PERSSON M, LINDBERG E, REESE H. Tree species classification with multi-temporal Sentinel-2 data[J]. Remote Sensing, 2018, 10(11):1794.
[21] 郑凡.黄山市森林资源现状及变化分析[J]. 安徽林业科技, 2015,41(3):54-57.
[22] ESA. Introducing Sentinel-2[EB/OL].[2020-01-05]. http://www.esa.int/Our_Activities/Observing_the_Earth/Copernicus/Sentinel-2/Introducing_Sentinel-2.
[23] 孙家抦.遥感原理及应用[M].武汉:武汉大学出版社, 2003.
[24] 杨静学,王云鹏,杨勇.基于高程或气溶胶厚度与6S模型校正参数回归方程的遥感图像大气校正模型[J].遥感技术与应用,2009,24(3):331-340.
[25] 唐守正. 林业发展要重视森林经营[N]. 中国绿色时报,2012-12-17(004).
[26] 周学武,贺东北,柯善新,等.森林资源二类调查成果衔接林地"一张图"等成果问题探讨[J].中南林业调查规划,2016,35(4):5-8,13.
[27] 张沁雨,李哲,夏朝宗,等.高分六号遥感卫星新增波段下的树种分类精度分析[J].地球信息科学学报,2019,21(10):1619-1628.
[28] CLEVERS J G P W, GITELSON A A. Remote estimation of crop and grass chlorophyll and nitrogen content using red-edge bands on Sentinel-2 and -3[J]. International Journal of Applied Earth Observation & Geoinformation,2013,23:344-351.
[29] 杨三红, 李晋敏. 基于DEM的森林空间分布格局的研究[J]. 山西林业科技, 2008(2):15-17,24.
[30] FRANKLIN S E. Satellite remote sensing of forest type and landcover in the subalpine forest region, Kananaskis Valley, Alberta[J]. Geocarto International, 1992, 7(4):25-35.
[31] 曾宏达. 基于DEM和地统计的森林资源空间格局分析——以武夷山山区为例[J]. 地球信息科学学报, 2005, 7(2):82-88.
[32] CARLSON T N, RIPLEY D A. On the relation between NDVI, fractional vegetation cover, and leaf area index[J]. Remote Sensing of Environment, 1997, 62(3):241-252.
[33] PEARSON R L, MILLER D L. Remote mapping of standing crop biomass for estimation of the productivity of the shortgrass prairie[J]. Remote Sensing of Environment, 1972,4(2):1357-1381.
[34] HUETE A, DIDAN K, MIURA T, et al. Overview of the radiometric and biophysical performance of the MODIS vegetation indices[J]. Remote Sensing of Environment, 2002, 83(1-2):195-213.
[35] MATSUSHITA B, YANG W, CHEN J, et al. Sensitivity of the enhanced vegetation index (EVI) and normalized difference vegetation index (NDVI) to topographic effects:a case study in high-density cypress forest[J]. Sensors, 2007, 7(11):2636-2651.
[36] GITELSON A, MERZLYAK M N. Spectral reflectance changes associated with autumn senescence of aesculus hippocastanum L. and acer platanoides L. leaves. spectral features and relation to chlorophyll estimation[J]. Journal of Plant Physiology, 1994, 143(3):286-292.
[37] ERDLE K, MISTELE B, SCHMIDHALTER U. Comparison of active and passive spectral sensors in discriminating biomass parameters and nitrogen status in wheat cultivars[J]. Field Crops Research,2011, 124(1):74-84.
[38] BREIMAN L. Random forests[J]. Machine Learning, 2001, 45:5-32.
[39] GISLASON P O, BENEDIKTSSON J A, SVEINSSON J R. Random forests for land cover classification[J]. Pattern Recognittion Letters, 2004, 27:294-300.
[40] GHIMIRE B, ROGAN J, RODRÍGUEZ G V, et al. An evaluation of bagging, boosting, and random forests for land-cover classification in Cape Cod, Massachusetts, USA[J]. GIScience & Remote Sensing, 2012, 49(5):623-643.
[41] RÄSÄNEN A, RUSANEN A, KUITUNEN M, et al. What makes segmentation good? a case study in boreal forest habitat mapping[J]. International Journal of Remote Sensing, 2013, 34(23):8603-8627.
[42] SUN J, LAI Z. Airborne LiDAR feature selection for urban classification using random forests[J]. Geomatics and Information Science of Wuhan University, 2014, 39(11):1310-1313.
[43] ATKINSON J T, ISMAIL R, ROBERTSON M. Mapping bugweed (solanum mauritianum) infestations in pinus patula plantations using hyperspectral imagery and support vector machines[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2014, 7(1):17-28.
[44] 樊东东, 李强子, 王红岩, 等. 通过训练样本采样处理改善小宗作物遥感识别精度[J]. 遥感学报, 2019,23(4):730-742
[45] YANG C, LIU S, BRUZZONE L, et al. A feature-metric-based affinity propagation technique for feature selection in hyperspectral image classification[J]. IEEE Geoscience and Remote Sensing Letters, 2013, 10(5):1152-1156.
[46] 何云,黄翀,李贺,等.基于Sentinel-2A影像特征优选的随机森林土地覆盖分类[J].资源科学,2019,41(5):992-1001.
[47] IMMITZER M, ATZBERGER C, KOUKAL T. Tree species classification with random forest using very high spatial resolution 8-band WorldView-2 satellite data[J]. Remote Sensing, 2012, 4(9):2661-2693.
[48] CONGALTON R G. Considerations and techniques for assessing the accuracy of remotely sensed data[C]//Proceedings of the 12th Canadian Symposium on Remote Sensing Geoscience and Remote Sensing Symposium, Vancouver:IEEE, 1989.
[49] 赵英时. 遥感应用分析原理与方法[M]. 北京:科学出版社, 2003.