利用GPS卫星信号信噪比的土壤湿度反演方法

Retrieval method of soil moisture by GPS SNR

为了进一步研究裸土区域表层土壤湿度的全球卫星导航系统反射信号干涉测量(GNSS-IR)遥感,提出1种利用GPS卫星信号信噪比数据的土壤湿度反演方法:建立振幅、相位、频率观测量与土壤湿度间的多元线性回归模型;并对振幅观测量取对数作非线性变换,以提升性能。实验结果表明:多元回归模型预测结果对比原位数据,平均均方根误差(RMSE)为2.43%;非线性变换后平均均方根误差为1.23%,比线性变换前降低49.27%。

In order to further study on GNSS-IR remote sensing of surface soil moisture in bare soil area,the paper proposed a retrieval method of soil moisture using GPS SNR:a multiple linear regression model between amplitude,phase,frequency observations and soil moisture was established;and the logarithm of the amplitude observation was taken as a nonlinear transformation to improve the performance.Experimental result showed that through comparing the prediction results of multiple regression models with the in-situ data,the average RMSE would be 2.43%;while the average RMSE after nonlinear transformation would be 1.23%,which reduced by 49.27% after linear transformation.