Three-stage inversion process for polarimetric SAR interferometry
Three-stage inversion process for polarimetric SAR interferometry
复制标题
DOI：
10.1049/ip-rsn:20030449
复制
发表时间：
2003-06-01
期刊：
影响因子：
--
通讯作者：
Papathanassiou, KP
中科院分区：
文献类型：
--
作者：
Cloude, SR;Papathanassiou, KP
作者提供了一种新的几何方法来反演两层相干散射模型，该模型广泛用于偏振干涉SAR数据的解释。最近的一些出版物表明，通过在多个极化通道中使用干涉图，可以估计植被高度、下伏地面地形和平均消光。此外，这可以用单个频率传感器实现，而不需要单独的参考DEM，其他先验信息或使用数据特定的回归公式。作者首先回顾了这种方法的细节，然后开发了一个三阶段的反演程序来说明参数估计所涉及的步骤。然后，他们考虑了反演中几个可能的误差来源。他们特别关注了垂直树结构的影响和时间去相关对反演精度的影响。结果表明，前者主要在消光估计上导致误差，而后者不改变模型结构，但减少了可用参数集，增加了参数估计的方差。最后，将新算法应用于随机冠层的矢量相干SAR模拟数据。
The authors provide a new geometrical approach for the inversion of a two-layer coherent scattering model, widely used for the interpretation of polarimetric interferometric SAR data. It has been shown in several recent publications that, by using interferograms in multiple polarisation channels, estimation of vegetation height, underlying ground topography and mean extinction is possible. Furthermore, this can be achieved with a single frequency sensor without the need for a separate reference DEM, other a priori information or the use of data-specific regression formulas. The authors first review the details of this approach and then develop a three-stage inversion procedure to illustrate the steps involved in parameter estimation. They then consider several possible sources of error in the inversion. In particular, they concentrate on the effects of vertical tree structure and on the effects of temporal decorrelation on inversion accuracy. It is shown that the former leads to errors, mainly in the extinction estimation, while the latter does not change the model structure but reduces the available parameter set and increases the variance of the parameter estimates. Finally, the new algorithm is applied to simulated vector coherent SAR data for a random canopy.