机载X-波段双天线InSAR数据森林树高估测方法

短波长的干涉合成孔径雷达(InSAR)适用于数字表面模型(DSM)提取,但难以提取准确的林下地相位,在缺乏高精度数字高程模型(DEM)的森林区域,短波长InSAR数据估测树高的能力受到限制。针对这一问题,采用机载X-波段单极化(HH)双天线InSAR数据开展了森林树高估测方法研究。双天线InSAR可以忽略时间去相干的影响,并且X-波段波长较短,入射角较大(中心入射角45.77°),地表对干涉去相干的贡献可以忽略,因此可将干涉复相干作为体去相干,对体去相干模型中的结构函数进行勒让德展开,截取第0阶展开式得到了基于相干幅度的森林树高估测模型,利用均匀选取的LiDAR冠层高度模型(CHM)检验样本对估测结果进行严格的精度评价,并与差分法的树高估测结果进行对比。精度评价结果显示:相干幅度法与差分法都得到了较高的估测精度,两者的R~2、RMSE、总精度分别为0.81、0.86;1.20m、0.97m;86.4%、88.7%。研究结果表明:相干幅度与森林树高具有负相关关系,适用于估测树高,基于单极化相干幅度的估测模型也可以得到较高的估测精度,与差分法的估测结果相比,虽然估测精度略有降低,但此方法具有两方面的优势:一方面,估测结果不需要实测样地数据标定,对于没有实测样地数据的森林区域亦能进行高精度的树高估测;另一方面,相干幅度法不需要高精度的DEM,具有更强的实用性。

Forest Height Estimation from Airborne X-band Single\|pass InSAR Data

The ground topography of forestry areas is difficult to derive based on short wavelength InSAR data for it is more applicable to estimate Digital Surface Model (DSM).The tree height estimation based on short wavelength InSAR data is restricted without high precision Digital Elevation Model (DEM).To address the problem,The X\|band HH polarization single\|pass InSAR data was used for tree height estimation.The X\|band single\|pass InSAR data is characterized by non\|temporal decorrelation,and the surface return would be ignored.Thus the volume decorrelation equals to the interferometric coherence.The Fourier\|Legendre polynomials can be used to describe the structure function of volume decorrelation term,and the zeroth\|order Legendre polynomial yields a SINC function developped as the tree height estimation model,namely coherent amplitude method.The model precision was validated with LiDAR CHM data compared with the method based on phase difference.The results showed that the estimation precision of both methods were high,for the R2 were 0.81,0.86,respectively,the RMSE were 1.20m,0.97m and the overall accuracy were 86.4%,88.7%.It can be concluded that there is a remarkable negative correlation between coherent amplitude and tree height,and the single\|polarization coherent amplitude can be used to obtain a satisfactory estimation result based on coherent amplitude method.Though the estimation precision of coherent amplitude method was slightly lower compared with the method based on phase difference,there are two advantages:firstly,the estimation result of this method don’t need to be calibrated,so it can be used for forest areas without true data.Secondly,it is more practical for the high precision DEM is not required.