基于Sentinel-1A与Landsat 8数据的北黑高速沿线地表土壤水分遥感反演方法研究

发布时间：2018-08-24 20:58

【摘要】：土壤含水量是影响水文和气候变化的基本参数,研究土壤含水量分布,对气候变化、水资源分布、农作物估产等有着重要的现实意义和科学价值。本文以2015年6月21日的Sentinel-1A(哨兵1号)双极化合成孔径雷达影像为基础,结合同时段辅助光学影像Landsat 8,对北安—黑河高速沿线地区不同植被覆盖程度下复杂地表土壤含水量进行反演研究,探讨不同极化组合方式在不同土地利用方式下的土壤水分含量反演结果。结果表明:VH极化及VH与辅助变量NDVI(Normalized Difference Vegetation Index)组合反演精度分别为52.1%和53.6%,整体效果并不理想。VV极化(VV Polarization)图像和双极化VV/VH(VH Polarization)组合在裸露和低植被地区反演更具有优势,其精度分别为75.4%和59.5%,而在高植被覆盖度地区并不适用。VH极化反演结果中耕地土壤含水量比实际值低9.37%,VV极化在低植被区域土壤含水量比实际值低10.45%,在灌木及耕地地区VV/VH反演结果精度比单极化及其组合反演结果低,最高精度模型的反演是VV结合NDVI。VV与辅助变量NDVI结合能综合反映复杂地表环境下土壤含水量,其精度达84%,标准误差RMSE为2.07,比VV极化反演精度提高8.8%,RMSE比VV极化降低2.704。VV与辅助变量NDVI组合方式在中等植被覆盖地区土壤含水量反演更具有优势,并能够更好地发挥哨兵1号C波段合成孔径雷达在土壤水分研究中的潜力与有效性。
[Abstract]:Soil water content is the basic parameter that affects hydrology and climate change. The study of soil water content distribution has important practical significance and scientific value to climate change, water resources distribution, crop yield estimation and so on. This paper is based on the Sentinel-1A (Sentinel 1) dual-polarization synthetic aperture radar image of June 21, 2015. With the aid of optical image Landsat 8 at the same time, the soil moisture content of complex surface in different vegetation coverage areas along Bei'an-Heihe high speed is studied. The inversion results of soil moisture content under different land use patterns were discussed in this paper. The results show that the inversion accuracy of the combination of VH and NDVI (Normalized Difference Vegetation Index) is 52.1% and 53.6%, respectively. The whole effect is not ideal. The combination of VV polarization (VV Polarization) image and dual-polarization VV/VH (VH Polarization) has more advantages in bare and low-vegetation areas. The precision is 75.4% and 59.5%, respectively. However, in the high vegetation coverage area, the soil moisture content of cultivated land is 9.37% lower than the actual value, and 10.45% lower than the actual value in the low vegetation area, and 10.45% in shrub and cultivated land. The accuracy of the VV/VH inversion results in the region is lower than that of the single-polarization and its combination inversion results. The inversion of the highest precision model is that VV combined with NDVI.VV and auxiliary variable NDVI can comprehensively reflect soil water content in complex surface environment. The precision is 84 and the standard error RMSE is 2.07, which is more accurate than that of VV polarization inversion, which is better than the combination of VV polarization reduction 2.704.VV and auxiliary variable NDVI in the inversion of soil water content in middle vegetation covered area. The potential and effectiveness of Sentinel 1 C-band synthetic Aperture Radar (SAR) in soil moisture research can be better developed.
【作者单位】：东北林业大学森林作业与环境研究中心;东北林业大学土木工程学院;四川农业大学资源学院;
【分类号】：S127;S152.7

【相似文献】