基于高分三号卫星合成孔径雷达数据的农田土壤水分反演

Retrieval of soil moisture based on Gaofen-3(GF-3)satellite synthetic aperture radar data over agricultural fields

土壤水分是农作物生长的基本条件,本研究基于高分三号卫星C波段合成孔径雷达数据,提出新的土壤水分反演算法,并获取区域尺度8 m空间分辨率的农田区土壤水分。首先,通过PROSAIL模型、实测植被冠层含水量、Landsat-8光学数据优选光学植被水分指数,计算水云模型参数并获得土壤直接后向散射系数;其次,利用高级积分方程模型模拟雷达后向散射影响机制,采用雷达影像高低入射角特性计算地表组合粗糙度;最后,利用高分三号卫星同极化雷达数据反演农田区土壤水分,并基于实测数据开展精度验证。结果表明:土壤水分反演值与野外实测值具有良好一致性,垂直极化下反演精度更高,其决定系数为0.5956,均方根误差为0.0415 m3/m3。本研究成果可为我国自主研发的高分三号卫星获取高分辨率土壤水分信息提供算法参考。

Soil moisture is the basic condition for crop growth.A new retrieval algorithm for soil moisture was proposed based on C-band synthetic aperture radar(SAR)data from Gaofen-3(GF-3)satellite,and soil moisture of agricultural fields with a regional scale spatial resolution of 8 m was obtained.First,the algorithm selected the optical vegetation water index based on PROSAIL model,measured vegetation canopy water content and Landsat-8 optical data.The parameters of water cloud model were calculated,and soil direct backscattering coefficients were obtained.Second,the radar backscattering influence mechanism was simulated using an advanced integral equation model(AIEM),and the combined roughness of soil surface was calculated based on the characteristics of radar data at high and low incidence angles.Finally,soil moisture was retrieved using co-polarization radar data from GF-3 satellite over agricultural fields,and this was verified with measured data.The results showed that there was a high consistency between the measured soil moisture and estimated soil moisture,and vertical-vertical(VV)polarization exhibited higher retrieval accuracy,with a determination coefficient of 0.5956 and a root mean square error of 0.0415 m3/m3.The results can provide algorithmic references for the GF-3 satellite to obtain high-resolution soil moisture information.