基于Radarsat-2全极化数据的多种雷达植被指数差异分析
|
摘要
[目的]雷达植被指数(Radar Vegetation Index,RVI)作为评价雷达影像植被分布与生长状态的重要指标,对植被生长动态监测具有重要意义。然而,不同算法的雷达植被指数对于同一地物类型的表征往往存在一定的差异。文章通过对比分析3种常用RVI在多种类型地物上的差异,为其在SAR影像特征提取、分类、识别等应用提供指导性意见。[方法]实验基于武汉市Radarsat-2全极化数据,结合Google earth历史影像和实地调研数据,选取林地、灌丛、草地、耕地、水生植被、建筑、道路、裸地、湖泊、河流10种典型地物样本,从样本折线图分布、类内标准差等方面,对分别通过H/A/alpha分解、Freeman分解和后向散射系数计算得到的3种常用雷达植被指数Van_RVI、Freeman_RVI和Kim_RVI进行了测算分析。[结果] 3种雷达植被指数有着相似的折线图走势,对植被的监测能力良好,但对于不同地物的敏感性稍有差异:Freeman_RVI对林地等高密度植被区域敏感程度较高; Van_RVI对耕地与林地、灌木与林地具有一定的区分性; Kim_RVI对水体与建筑的敏感程度较高。[结论] Freeman_RVI对高密度植被识别能力最好,可用于林地提取、森林制图; Van_RVI对植被与非植被的区分能力最好,适用于植被提取; Kim_RVI数据预处理计算速度最快,但提取精度不高,可用于应急制图。
Radar vegetation index( RVI) is one of the most-widely used measurements to monitor and evaluate the vegetation distribution and vegetation growth status of Synthetic Aperture Radar( SAR) images. It is of great significance to the dynamic monitoring of vegetation growth. However,the RVIs calculated from different algorithms have some differences in the evaluation of the same land cover type. This research mainly analyzes the differences of three commonly used RVIs on the performance of various types of land cover,and provides guidance for its application in SAR image feature extraction,image classification and recognition. The experimental data were collected from Wuhan city,including Radarsat-2 full-polarization data,Google earth historical image and ground truth,and the ten typical samples of forest land,shrub,grassland,cultivated land,aquatic vegetation,buildings,roads,bare land,lakes and rivers were selected. The Van_RVI derived from H/A/alpha polarimetric decomposition,Freeman_RVI derived from Freeman-Durden decomposition and Kim _ RVI calculated from the backscattering coefficients were compared and analyzed from histogram distribution and the standard deviation. The result shows that three kinds of RVI have similar histogram trend,and have the capacity to monitor vegetation effectively. However,their sensitivities to different land covers are slight difference. Freeman_RVI has good sensitivity to high-density vegetation such as forestland; Van_RVI can distinguish cultivated land from forestland,or shrubs from forest land. Kim_RVI is very sensitive to water and buildings. In conclusion,Freeman_RVI can be used for forest extraction and forest mapping for its best ability to recognize high-density vegetation. Van_RVI can be used for vegetation extraction,due to the ability to distinguish between vegetation and non-vegetation features. Kim_RVI cannot effectively distinguish different land cover types,but it can be used to calculate for vegetation index emergency mapping because of its simple preprocessing flow.
Radar vegetation index( RVI) is one of the most-widely used measurements to monitor and evaluate the vegetation distribution and vegetation growth status of Synthetic Aperture Radar( SAR) images. It is of great significance to the dynamic monitoring of vegetation growth. However,the RVIs calculated from different algorithms have some differences in the evaluation of the same land cover type. This research mainly analyzes the differences of three commonly used RVIs on the performance of various types of land cover,and provides guidance for its application in SAR image feature extraction,image classification and recognition. The experimental data were collected from Wuhan city,including Radarsat-2 full-polarization data,Google earth historical image and ground truth,and the ten typical samples of forest land,shrub,grassland,cultivated land,aquatic vegetation,buildings,roads,bare land,lakes and rivers were selected. The Van_RVI derived from H/A/alpha polarimetric decomposition,Freeman_RVI derived from Freeman-Durden decomposition and Kim _ RVI calculated from the backscattering coefficients were compared and analyzed from histogram distribution and the standard deviation. The result shows that three kinds of RVI have similar histogram trend,and have the capacity to monitor vegetation effectively. However,their sensitivities to different land covers are slight difference. Freeman_RVI has good sensitivity to high-density vegetation such as forestland; Van_RVI can distinguish cultivated land from forestland,or shrubs from forest land. Kim_RVI is very sensitive to water and buildings. In conclusion,Freeman_RVI can be used for forest extraction and forest mapping for its best ability to recognize high-density vegetation. Van_RVI can be used for vegetation extraction,due to the ability to distinguish between vegetation and non-vegetation features. Kim_RVI cannot effectively distinguish different land cover types,but it can be used to calculate for vegetation index emergency mapping because of its simple preprocessing flow.
引文
[1]康耀江.植被指数在草地遥感中的应用初探.湖南农业科学,2011(4):39-41.
[2] Gascon M,Cirach M,David Martínez,et al. Normalized Difference Vegetation Index(NDVI)as a marker of surrounding greenness in epidemiological studies:the case of Barcelona city. Urban Forestry&Urban Greening,2016,19:88-94.
[3] Dusseux P,Corpetti T,Hubert-Moy L,et al. combined Use of Multi-Temporal Optical and Radar Satellite Images for Grassland Monitoring.Remote Sensing,2014,6(7):6163-6182.
[4] Huang Y,Walker J P,Gao Y,et al. Estimation of Vegetation Water Content From the Radar Vegetation Index at L-band. IEEE Transactions on Geoscience&Remote Sensing,2016,54(2):981-989.
[5] Ling F,Li Z,Chen E,et al. comparison of ALOS PALSAR RVi and landsat TM NDVI for forest area mapping//Asian-pacific Conference on Synthetic Aperture Radar. IEEE,2010.
[6]许涛,廖静娟,沈国状,等.基于高分一号与Radarsat-2的鄱阳湖湿地植被叶面积指数反演.红外与毫米波学报,2016,35(3)
[7]沈国状,廖静娟. SAR数据湿地植被生物量反演方法研究进展.遥感信息,2016,31(3):1-8.
[8]岳继博,杨贵军,冯海宽.基于随机森林算法的冬小麦生物量遥感估算模型对比.农业工程学报,2016,32(18)
[9] Freeman A,Durden S L. A three-component scattering model for polarimetric SAR data. IEEE Transactions on Geoscience and Remote Sensing,2002,36(3):963-973.
[10]王庆,曾琪明,廖静娟.基于极化分解的极化特征参数提取与应用.国土资源遥感,2012,24(3):103-110.
[11]何海燕,凌飞龙,汪小钦,等.基于雷达植被指数的水土流失区植被覆盖度估测.国土资源遥感,2015,27(4):165-170.
[12]马腾,王耀强,李瑞平,等.基于微波遥感极化目标分解的土地覆盖/土地利用分类.农业工程学报,2015,31(2):259-265.
[13]杨浩,杨贵军,顾晓鹤,等.小麦倒伏的雷达极化特征及其遥感监测.农业工程学报,2014(7)
[14]基于改进水云模型和Radarsat-2数据的农田土壤含水量估算.农业工程学报,2016(22)
[15]郝莹莹,罗小波,仲波,等.基于植被分区的中国植被类型分类方法.遥感技术与应用,2017(2)
[16] Zyl,Jakob J. van.“An overview of the analysis of multi-frequency polarimetric SAR data.”6th European Conference on Synthetic Aperture Radar(EUSAR 2006),2006.
[17] Kim Y,Van Zyl J J. A Time-Series Approach to Estimate Soil Moisture Using Polarimetric Radar Data. IEEE Transactions on Geoscience and Remote Sensing,2009,47(8):2519-2527.
[2] Gascon M,Cirach M,David Martínez,et al. Normalized Difference Vegetation Index(NDVI)as a marker of surrounding greenness in epidemiological studies:the case of Barcelona city. Urban Forestry&Urban Greening,2016,19:88-94.
[3] Dusseux P,Corpetti T,Hubert-Moy L,et al. combined Use of Multi-Temporal Optical and Radar Satellite Images for Grassland Monitoring.Remote Sensing,2014,6(7):6163-6182.
[4] Huang Y,Walker J P,Gao Y,et al. Estimation of Vegetation Water Content From the Radar Vegetation Index at L-band. IEEE Transactions on Geoscience&Remote Sensing,2016,54(2):981-989.
[5] Ling F,Li Z,Chen E,et al. comparison of ALOS PALSAR RVi and landsat TM NDVI for forest area mapping//Asian-pacific Conference on Synthetic Aperture Radar. IEEE,2010.
[6]许涛,廖静娟,沈国状,等.基于高分一号与Radarsat-2的鄱阳湖湿地植被叶面积指数反演.红外与毫米波学报,2016,35(3)
[7]沈国状,廖静娟. SAR数据湿地植被生物量反演方法研究进展.遥感信息,2016,31(3):1-8.
[8]岳继博,杨贵军,冯海宽.基于随机森林算法的冬小麦生物量遥感估算模型对比.农业工程学报,2016,32(18)
[9] Freeman A,Durden S L. A three-component scattering model for polarimetric SAR data. IEEE Transactions on Geoscience and Remote Sensing,2002,36(3):963-973.
[10]王庆,曾琪明,廖静娟.基于极化分解的极化特征参数提取与应用.国土资源遥感,2012,24(3):103-110.
[11]何海燕,凌飞龙,汪小钦,等.基于雷达植被指数的水土流失区植被覆盖度估测.国土资源遥感,2015,27(4):165-170.
[12]马腾,王耀强,李瑞平,等.基于微波遥感极化目标分解的土地覆盖/土地利用分类.农业工程学报,2015,31(2):259-265.
[13]杨浩,杨贵军,顾晓鹤,等.小麦倒伏的雷达极化特征及其遥感监测.农业工程学报,2014(7)
[14]基于改进水云模型和Radarsat-2数据的农田土壤含水量估算.农业工程学报,2016(22)
[15]郝莹莹,罗小波,仲波,等.基于植被分区的中国植被类型分类方法.遥感技术与应用,2017(2)
[16] Zyl,Jakob J. van.“An overview of the analysis of multi-frequency polarimetric SAR data.”6th European Conference on Synthetic Aperture Radar(EUSAR 2006),2006.
[17] Kim Y,Van Zyl J J. A Time-Series Approach to Estimate Soil Moisture Using Polarimetric Radar Data. IEEE Transactions on Geoscience and Remote Sensing,2009,47(8):2519-2527.