不同等级土壤遥感分类的尺度匹配性探讨
|
摘要
采用Landsat5 TM遥感影像为数据源,以湖北省江汉平原之潜江市为试验区,探讨同一遥感信息源下对不同等级的土壤分类的尺度匹配性.以Landsat5 TM遥感影像和两个级别的土壤类型图(土壤亚类及土属)为基础数据,集成主成分分析、归一化植被指数等图像处理技术,提取多种影像特征建立土壤分类特征数据集;采用最大似然监督分类方法对潜江市不同级别的土壤分别进行遥感分类;并利用混淆矩阵方法对分类结果分别进行精度验证.结果表明:土属的总体分类精度较高,达到92.79%,Kappa系数为0.919 5;土壤亚类相对较低,总体分类精度只有84.71%,Kappa系数为0.820 1.可见土壤遥感分类具有显著的尺度适宜性特征,在两个级别的土壤分类实验中,Landsat5 TM更适宜土壤的最基层的土属类型划分.在土壤遥感分类时,应首先探讨土壤类型等级与遥感影像的尺度匹配性.
The aim of this study is to test the suitability of different levels of soil remote sensing classification based on Landsat5 TM remote sensing in Qianjiang city,Hubei Province. The authors employed the Landsat5 TM remote sensing image and the two levels of soil type maps( soil subcategories map and soil genus map) as basic data sources. The TM image was processed to extract classification features by using a variety of image processing techniques,which included such means as principal component analysis,tasseled cap transformation and normalized differential vegetation index. Then the authors incorporated all classification features into a dataset,and used maximum likelihood classifier of supervision to classify the above two levels of soil in Qianjiang City,respectively. And the accuracy of the two levels soil classification results was verified by using the confusion matrixes method. The results suggest that soil classification based on Landsat5 Tremote sensing image,soil genus is the highest,and the overall classification accuracy can reach 92. 79%,the Kappa coefficient was 0. 919 5. The overall classification accuracy of soil subcategories is the second highest,the overall classification accuracy was 84. 71%,the Kappa coefficient was 0. 820 1. The main reason is that soil genus contained less other soil types in the soil samples at the same spatial scale and the sample pixelluminance value is highly correlated with the pixel brightness value in the TM remote sensing image,which leads to the higher overall classification accuracy. It can be seen that Landsat5 TM is the more suitable for soil genus classification in these two grades of soil classification,and the soil remote sensing classification may have significant scale suitability characteristics. Therefore,the scale matching of soil types and remote sensing image is the most important factor in soil remote sensing classification. This study can provide a reference for selecting different suitable remote sensing data sources for different soil classification studies at different levels.
The aim of this study is to test the suitability of different levels of soil remote sensing classification based on Landsat5 TM remote sensing in Qianjiang city,Hubei Province. The authors employed the Landsat5 TM remote sensing image and the two levels of soil type maps( soil subcategories map and soil genus map) as basic data sources. The TM image was processed to extract classification features by using a variety of image processing techniques,which included such means as principal component analysis,tasseled cap transformation and normalized differential vegetation index. Then the authors incorporated all classification features into a dataset,and used maximum likelihood classifier of supervision to classify the above two levels of soil in Qianjiang City,respectively. And the accuracy of the two levels soil classification results was verified by using the confusion matrixes method. The results suggest that soil classification based on Landsat5 Tremote sensing image,soil genus is the highest,and the overall classification accuracy can reach 92. 79%,the Kappa coefficient was 0. 919 5. The overall classification accuracy of soil subcategories is the second highest,the overall classification accuracy was 84. 71%,the Kappa coefficient was 0. 820 1. The main reason is that soil genus contained less other soil types in the soil samples at the same spatial scale and the sample pixelluminance value is highly correlated with the pixel brightness value in the TM remote sensing image,which leads to the higher overall classification accuracy. It can be seen that Landsat5 TM is the more suitable for soil genus classification in these two grades of soil classification,and the soil remote sensing classification may have significant scale suitability characteristics. Therefore,the scale matching of soil types and remote sensing image is the most important factor in soil remote sensing classification. This study can provide a reference for selecting different suitable remote sensing data sources for different soil classification studies at different levels.
引文
[1]赵其国.我国土壤调查制图及土壤分类工作的回顾与展望[J].土壤,1992,24(6):281-284.
[2]张甘霖,史学正,龚子同.中国土壤地理学发展的回顾与展望[J].土壤学报,2008,45(5):792-801.
[3]Amani M,Parsian S,Mir Mazloumi S M,et al.Two new soil moisture indices based on the NIR-red triangle space of Landsat-8 data[J].Int J Appl Earth Obs,2016,50:176-186.
[4]Fan X,Weng Y,Tao J.Towards decadal soil salinity mapping using Landsat time series data[J].Int J Appl Earth Obs,2016,52:32-41.
[5]史舟,李艳,程街亮.水稻土重金属空间分布的随机模拟和不确定评价[J].环境科学,2007,28(1):209-214.
[6]Sridhar B B M,Vincent R K,Witter J D,et al.Mapping the total phosphorus concentration of biosolid amended surface soils using Landsat TM data[J].Sci Total Environ,2009,407(8):2894-2899.
[7]李建龙,蒋平,刘培君,等.利用遥感光谱法进行农田土壤水分遥感动态监测[J].生态学报,2003,23(8):1498-1504.
[8]Wang C,Qi J,Moran S,et al.Soil moisture estimation in a semiarid rangeland using ERS-2 and TM imagery[J].Remote Sens Environ,2004,90(2):178-189.
[9]杨涛,宫辉力,李小娟,等.土壤水分遥感监测研究进展[J].生态学报,2010,30(22):6264-6277.
[10]Filion R,Bernier M,Paniconi C,et al.Remote sensing for mapping soil moisture and drainage potential in semi-arid regions:applications to the Campidano plain of Sardinia,Italy[J].Sci Total Environ,2016,543(B):862-876.
[11]李燕丽,潘贤章,王昌昆,等.广西中南部耕地土壤有机质和全氮变化的遥感监测[J].生态学报,2014,34(18):5283-5291.
[12]章文龙,曾从盛,高灯州,等.闽江河口湿地土壤全磷高光谱遥感估算[J].生态学报,2015,35(24):8085-8093.
[13]Gao J,Pan G,Jiang X.Land-use induced changes in topsoil organic carbon stock of paddy fields using MODIS and TM/ETM analysis:a case study of wujiang county,china[J].Journal of Environmental Sciences,2008,20(7):852-858.
[14]El Harti A,Lhissou R,Chokmani K,et al.Spatiotemporal monitoring of soil salinization in irrigated Tadla plain(Morocco)using satellite spectral indices[J].Int J Appl Earth Obs,2016,50:64-73.
[15]王深法,蒋亨显,王人潮.浙江省石灰土光谱特征及其自动识别分类技术研究[J].土壤学报,1994,31(3):312-321.
[16]沙晋明,李小梅.基于遥感信息的哈夫曼优化树在山地土壤资源调查中的应用——以浙江省龙游县为例[J].山地学报,2002,20(2):223-227.
[17]刘娟,蔡演军,王瑾.青海湖流域土壤遥感分类[J].国土资源遥感,2014,26(1):57-62.
[18]亢庆,张增祥,赵晓丽.基于MODIS产品的区域土壤遥感分类研究——以新疆为例[J].遥感技术与应用,2007,22(6):690-695.
[19]王人潮,吴嘉平,王深法,等.SPOT图像的土壤解译与制图效果研究[J].浙江农林大学学报,1991,17(4):341-346.
[20]王宏志,潘方杰,周勇,等.异质景观条件下江汉平原土壤的空间分异[J].生态学报,2016,36(18):5682-5690.
[21]赵英时.遥感应用分析原理与方法[M].北京:科学出版社,2003.
[2]张甘霖,史学正,龚子同.中国土壤地理学发展的回顾与展望[J].土壤学报,2008,45(5):792-801.
[3]Amani M,Parsian S,Mir Mazloumi S M,et al.Two new soil moisture indices based on the NIR-red triangle space of Landsat-8 data[J].Int J Appl Earth Obs,2016,50:176-186.
[4]Fan X,Weng Y,Tao J.Towards decadal soil salinity mapping using Landsat time series data[J].Int J Appl Earth Obs,2016,52:32-41.
[5]史舟,李艳,程街亮.水稻土重金属空间分布的随机模拟和不确定评价[J].环境科学,2007,28(1):209-214.
[6]Sridhar B B M,Vincent R K,Witter J D,et al.Mapping the total phosphorus concentration of biosolid amended surface soils using Landsat TM data[J].Sci Total Environ,2009,407(8):2894-2899.
[7]李建龙,蒋平,刘培君,等.利用遥感光谱法进行农田土壤水分遥感动态监测[J].生态学报,2003,23(8):1498-1504.
[8]Wang C,Qi J,Moran S,et al.Soil moisture estimation in a semiarid rangeland using ERS-2 and TM imagery[J].Remote Sens Environ,2004,90(2):178-189.
[9]杨涛,宫辉力,李小娟,等.土壤水分遥感监测研究进展[J].生态学报,2010,30(22):6264-6277.
[10]Filion R,Bernier M,Paniconi C,et al.Remote sensing for mapping soil moisture and drainage potential in semi-arid regions:applications to the Campidano plain of Sardinia,Italy[J].Sci Total Environ,2016,543(B):862-876.
[11]李燕丽,潘贤章,王昌昆,等.广西中南部耕地土壤有机质和全氮变化的遥感监测[J].生态学报,2014,34(18):5283-5291.
[12]章文龙,曾从盛,高灯州,等.闽江河口湿地土壤全磷高光谱遥感估算[J].生态学报,2015,35(24):8085-8093.
[13]Gao J,Pan G,Jiang X.Land-use induced changes in topsoil organic carbon stock of paddy fields using MODIS and TM/ETM analysis:a case study of wujiang county,china[J].Journal of Environmental Sciences,2008,20(7):852-858.
[14]El Harti A,Lhissou R,Chokmani K,et al.Spatiotemporal monitoring of soil salinization in irrigated Tadla plain(Morocco)using satellite spectral indices[J].Int J Appl Earth Obs,2016,50:64-73.
[15]王深法,蒋亨显,王人潮.浙江省石灰土光谱特征及其自动识别分类技术研究[J].土壤学报,1994,31(3):312-321.
[16]沙晋明,李小梅.基于遥感信息的哈夫曼优化树在山地土壤资源调查中的应用——以浙江省龙游县为例[J].山地学报,2002,20(2):223-227.
[17]刘娟,蔡演军,王瑾.青海湖流域土壤遥感分类[J].国土资源遥感,2014,26(1):57-62.
[18]亢庆,张增祥,赵晓丽.基于MODIS产品的区域土壤遥感分类研究——以新疆为例[J].遥感技术与应用,2007,22(6):690-695.
[19]王人潮,吴嘉平,王深法,等.SPOT图像的土壤解译与制图效果研究[J].浙江农林大学学报,1991,17(4):341-346.
[20]王宏志,潘方杰,周勇,等.异质景观条件下江汉平原土壤的空间分异[J].生态学报,2016,36(18):5682-5690.
[21]赵英时.遥感应用分析原理与方法[M].北京:科学出版社,2003.