艾比湖湿地自然保护区土壤盐分多光谱遥感反演模型
|
摘要
土壤盐分是衡量土壤质量的要素,也是作物生长发育的基本条件。因此,迫切地需要一种可以快速了解土壤盐分含量(SSC)的方法。针对艾比湖湿地自然保护区,基于Landsat8 OLI多光谱遥感影像,以该研究区36个土壤表层样品的盐分含量为数据源,选择相关性较好的多光谱遥感指数分析研究区土壤盐分分布状况,并将其分别与实测SSC构建线性、对数、二次函数模型,进而优选精度最高的模型来反演该研究区SSC。结果表明:(1)在多光谱遥感指数中,与SSC相关性最高的是增强型植被指数(EVI),其相关性范围为(-0.70～-0.67);其次是传统型植被指数(TVI),其范围为(-0.58～-0.46);土壤盐分指数(SI)与SSC的相关性最低,其范围为(-0.45～0.16),其中SI3和SI4与SSC均没有相关性。(2)将实测土壤盐分值所反演的分布图与EVI对比分析,发现在西北、正南方向的艾比湖湖边周围和东北方向盐池桥的SSC均较高,其EVI的值较低,说明通过该研究区实测土壤盐分值所反演的盐分分布图与EVI的空间分布结果较为一致,表明EVI对该地区土壤盐分具有一定的敏感性,能较好地反演SSC的空间分布;(3)分别将三种EVI与实测SSC建模分析比较,发现SSC与增强型比值植被指数(ERVI)所构建的二次函数模型最好;其验证集的决定系数(R~2)为0.92,均方根误差(RMSE)为2.48,相对分析误差(RPD)为2.09,模型精度较高、稳定性较为可靠,相比之下,说明ERVI对该湿地自然保护区土壤盐分有更高的敏感性,可以用来预测该区域SSC,从而进行空间反演。在TVI中加入Landsat8多光谱遥感影像的b6和b7波段,得到EVI,以此来反演SSC是可行的,且比传统可见光和近红外波段所构建的植被指数反演效果更好。因此该研究不仅可以为遥感反演提供理论参考,而且对该地区SSC的定量估算和动态监测具有重要的意义,也可作为其他区域SSC预测反演的备选方案。
Soil salinity is an important factor for measuring soil quality, and it is also a basic condition for the growth of crops. Therefore, it is urgent to find a method that can understand soil salt content quickly. This paper is based on the Landsat8 OLI multispectral remote sensing image for the Ebinur Lake Wetland Nature Reserve, and we use the salt content of 36 soil surface samples in the study area as the data source, and choose several multispectral remote sensing indices which have the superior correlation with soil salinity to analyze the soil salinity distribution in the study area. The linear, logarithmic and quadratic function models were constructed with the measured soil salinity, and optimum inversion model of soil salt content was selected. The result shows that:(1)Among these multispectral remote sensing indices, the enhanced vegetation indices show the closest correlation with soil salinity, and the correlation coefficient range is between-0.67 and-0.70. The second is the traditional vegetation indices, and the correlation coefficient range is between-0.46 and-0.58. The correlation of soil salt index is the farthest t,and its range is between 0.16 and-0.45, and there is no correlation between SI3, SI4 and soil salt content.(2)Comparing and analyzing the salt distribution map inverted by measured soil salinity values and the spatial distribution of Enhanced Vegetation indices, we found that the soil salt content around the Ebinur Lake of the northwest and south direction and tne Yan Chi Bridge in the northeast is higher, but the enhanced vegetation indices are lower. The result shows that the salt distribution map inverted by measured soil salinity values is consistent with the spatial distribution of Enhanced Vegetation indices. It indicates that the enhanced vegetation indices have a higher sensitivity to soil salinity,which can better reverse the spatial distribution of soil salinity in the study area.(3)From the comparison and analysis of those models, which builds the three enhanced vegetation indices and measured soil salt content respectively.We found that the enhanced ratio vegetation index is the best choice to construct the quadratic function model. The determination coefficient of its validation set(R~2) is 0.92, and the root mean square error(RMSE) is 2.48, and the relative analysis error(RPD) is 2.09. The data show that this model is more accurate and reliable. In summary, ERVI is more sensitive to soil salinity and predict the soil salinity content, while is more suitable for inversion of soil salinity in this study area. Therefore, the study indicates that it is feasible to invert the soil salinity by the enhanced vegetation index constructed by the b6 and b7 band of Landsat8 multipectral remote sensing imagery. And its inversion effect is better than that of traditional visible light band. Therefore, this study not only provides a theoretical reference for remote sensing inversion, but also has important implications for the quantitative estimation and dynamic monitoring of soil salinity for the study area. Otherwise, it can be used as an alternative offer for prediction of soil salt content in other regions.
Soil salinity is an important factor for measuring soil quality, and it is also a basic condition for the growth of crops. Therefore, it is urgent to find a method that can understand soil salt content quickly. This paper is based on the Landsat8 OLI multispectral remote sensing image for the Ebinur Lake Wetland Nature Reserve, and we use the salt content of 36 soil surface samples in the study area as the data source, and choose several multispectral remote sensing indices which have the superior correlation with soil salinity to analyze the soil salinity distribution in the study area. The linear, logarithmic and quadratic function models were constructed with the measured soil salinity, and optimum inversion model of soil salt content was selected. The result shows that:(1)Among these multispectral remote sensing indices, the enhanced vegetation indices show the closest correlation with soil salinity, and the correlation coefficient range is between-0.67 and-0.70. The second is the traditional vegetation indices, and the correlation coefficient range is between-0.46 and-0.58. The correlation of soil salt index is the farthest t,and its range is between 0.16 and-0.45, and there is no correlation between SI3, SI4 and soil salt content.(2)Comparing and analyzing the salt distribution map inverted by measured soil salinity values and the spatial distribution of Enhanced Vegetation indices, we found that the soil salt content around the Ebinur Lake of the northwest and south direction and tne Yan Chi Bridge in the northeast is higher, but the enhanced vegetation indices are lower. The result shows that the salt distribution map inverted by measured soil salinity values is consistent with the spatial distribution of Enhanced Vegetation indices. It indicates that the enhanced vegetation indices have a higher sensitivity to soil salinity,which can better reverse the spatial distribution of soil salinity in the study area.(3)From the comparison and analysis of those models, which builds the three enhanced vegetation indices and measured soil salt content respectively.We found that the enhanced ratio vegetation index is the best choice to construct the quadratic function model. The determination coefficient of its validation set(R~2) is 0.92, and the root mean square error(RMSE) is 2.48, and the relative analysis error(RPD) is 2.09. The data show that this model is more accurate and reliable. In summary, ERVI is more sensitive to soil salinity and predict the soil salinity content, while is more suitable for inversion of soil salinity in this study area. Therefore, the study indicates that it is feasible to invert the soil salinity by the enhanced vegetation index constructed by the b6 and b7 band of Landsat8 multipectral remote sensing imagery. And its inversion effect is better than that of traditional visible light band. Therefore, this study not only provides a theoretical reference for remote sensing inversion, but also has important implications for the quantitative estimation and dynamic monitoring of soil salinity for the study area. Otherwise, it can be used as an alternative offer for prediction of soil salt content in other regions.
引文
[1] Netternicht G I, Zinck J A. Remote Sensing of Environment, 2003, 85(1): 1.
[2] WANG Fei, DING Jian-li, WEI Yang, et al(王飞, 丁建丽, 魏阳, 等). Acta Ecologica Sinica(生态学报), 2017, 37(15): 5007.
[3] LUO Chong, WANG Ling, PENG Li, et al(罗冲, 王玲, 彭丽, 等). Journal of Shihezi University·Natural Science(石河子大学学报·自然科学版), 2016, 34(6): 779.
[4] Alexakis D D, Daliakopoulos I N, Panagea I S, et al. Geocarto International, 2016, 33(4): 321.
[5] Gorji T, Sertel E, Tanik A. Ecological Indicators, 2017, 74: 384.
[6] Gorji T, Sertel E, Tanki A, et al. Geoderma, 2011, 162: 1.
[7] SU Xiang-ming, LIU Zhi-hui, WEI Tian-feng, et al(苏向明, 刘志辉, 魏天锋, 等). Research of Soil and Water Conservation(水土保持研究), 2016, 23(3): 252.
[8] PENG Xiang, HU Dan, ZENG Wen-zhi, et al(彭翔, 胡丹, 曾文治, 等). Transactions of the Chinese Society of Agricultural Engineering(农业工程学报), 2016, 32(11): 167.
[9] LU Guang, HAN Mei, WANG Min, et al(路广, 韩美, 王敏, 等). Ecology and Environmental Sciences(生态环境学报), 2017, 26(3): 422.
[10] WANG Fei, DING Jian-li, WU Man-chun(王飞, 丁建丽, 伍漫春). Transactions of the Chinese Society of Agricultural Engineering(农业工程学报), 2010, 26(8): 168.
[11] Steig E J, Ding Q, Battisti D S, et al. Annals of Glaciology, 2012, 53(60): 19.
[12] LIU Ya, PAN Xian-zhang, WANG Chang-kun, et al(刘娅, 潘贤章, 王昌昆, 等). Acta Pedologica Sinica(土壤学报), 2012, 49(4): 824.
[13] LAI Ning, LI Xin-guo, LIANG Dong(赖宁, 李新国, 梁东). Journal of Arid Land Resources and Environment(干旱区资源与环境), 2015, 29(2): 151.
[14] Weng Y L, Gong P, Zhu Z L. Pedosphere, 2010, 20(3): 378.
[15] ZHANG Tian-you, WANG Ling, WANG Hui, et al(张添佑, 王玲, 王辉, 等). Acta Ecologica Sinica(生态学报), 2017, 37(9): 3009.
[16] WANG Duo-duo, JIA Wen-xiao, WANG Zhi-bao, et al(王多多, 贾文晓, 王志保, 等). Journal of Agricultural Science and Technology(中国农业科技导报), 2018, 20(3): 55.
[2] WANG Fei, DING Jian-li, WEI Yang, et al(王飞, 丁建丽, 魏阳, 等). Acta Ecologica Sinica(生态学报), 2017, 37(15): 5007.
[3] LUO Chong, WANG Ling, PENG Li, et al(罗冲, 王玲, 彭丽, 等). Journal of Shihezi University·Natural Science(石河子大学学报·自然科学版), 2016, 34(6): 779.
[4] Alexakis D D, Daliakopoulos I N, Panagea I S, et al. Geocarto International, 2016, 33(4): 321.
[5] Gorji T, Sertel E, Tanik A. Ecological Indicators, 2017, 74: 384.
[6] Gorji T, Sertel E, Tanki A, et al. Geoderma, 2011, 162: 1.
[7] SU Xiang-ming, LIU Zhi-hui, WEI Tian-feng, et al(苏向明, 刘志辉, 魏天锋, 等). Research of Soil and Water Conservation(水土保持研究), 2016, 23(3): 252.
[8] PENG Xiang, HU Dan, ZENG Wen-zhi, et al(彭翔, 胡丹, 曾文治, 等). Transactions of the Chinese Society of Agricultural Engineering(农业工程学报), 2016, 32(11): 167.
[9] LU Guang, HAN Mei, WANG Min, et al(路广, 韩美, 王敏, 等). Ecology and Environmental Sciences(生态环境学报), 2017, 26(3): 422.
[10] WANG Fei, DING Jian-li, WU Man-chun(王飞, 丁建丽, 伍漫春). Transactions of the Chinese Society of Agricultural Engineering(农业工程学报), 2010, 26(8): 168.
[11] Steig E J, Ding Q, Battisti D S, et al. Annals of Glaciology, 2012, 53(60): 19.
[12] LIU Ya, PAN Xian-zhang, WANG Chang-kun, et al(刘娅, 潘贤章, 王昌昆, 等). Acta Pedologica Sinica(土壤学报), 2012, 49(4): 824.
[13] LAI Ning, LI Xin-guo, LIANG Dong(赖宁, 李新国, 梁东). Journal of Arid Land Resources and Environment(干旱区资源与环境), 2015, 29(2): 151.
[14] Weng Y L, Gong P, Zhu Z L. Pedosphere, 2010, 20(3): 378.
[15] ZHANG Tian-you, WANG Ling, WANG Hui, et al(张添佑, 王玲, 王辉, 等). Acta Ecologica Sinica(生态学报), 2017, 37(9): 3009.
[16] WANG Duo-duo, JIA Wen-xiao, WANG Zhi-bao, et al(王多多, 贾文晓, 王志保, 等). Journal of Agricultural Science and Technology(中国农业科技导报), 2018, 20(3): 55.