多维度SAR数据地形效应校正方法研究

Multi-dimensional is the current development trend of SAR technology. In the application of qualitative and quantitative remote sensing, the method of combining multi-dimensional SAR features will effectively improve the accuracy of the application. However, we must first overcome the effects of topographic relief on the characteristics of multi-dimensional SAR. As the traditional terrain radiation correction methods are insufficient to support the terrain correction of multi-dimensional SAR data, this project will study and develop terrain effect correction methods for multi-dimensional SAR data. We will further enrich and perfect SAR terrain effect correction algorithms in the polarization and interference dimensions. For polarimetric SAR data, the new terrain correction method will be more automated and the whole process will be heteromorphic. For interferometric SAR data, the new method will be developed based on the high-order Legendre volume decorrelation model and generalized algebraic difference approach. Based on this, combined with the forest canopy height inversion method (RVoG model or complex coherence differential), the synergetic terrain correction algorithm will be developed for multi-dimensional SAR data. Through the research of this project, it can promote the application of multi-dimensional SAR technology in different industries and promote the rapid development of SAR technology, which has important theoretical and practical significance.

多维度是目前SAR技术的发展趋势，在定性和定量遥感的应用中联合多维度SAR特征将有效提高应用效果。但是，必须要克服地形起伏对于多维度SAR特征的影响。鉴于传统的地形辐射校正方法已不足以支撑多维度SAR数据的地形校正，本项目将研究和发展面向多维度SAR数据的地形效应校正方法。首先，进一步丰富和完善极化和干涉维度的SAR地形效应校正方法，提出高度自适应化且全流程异态化的极化SAR地形效应校正算法，发展基于高阶勒让德体散射失相干模型和广义代数差分方法的InSAR相干性的地形效应校正算法；然后，在此基础上结合RVoG模型或复相干差分等森林冠层高度反演算法，发展多维度SAR协同的地形效应校正方法。通过本项目的研究，可以促进多维度SAR技术在不同行业中的应用，推动SAR技术的快速发展，具有重要的理论和实践意义。

多维度是目前SAR成像技术的发展趋势，在定性和定量遥感的应用中联合多维度SAR特征将有效提高应用效果。但是，必须要克服地形起伏对于多维度SAR特征的影响。本研究针对多维度SAR地形辐射校正的关键问题开展研究，取得的主要研究成果如下：（1）发展了面向极化SAR监督分类的三阶段地形辐射校正算法，解决了极化SAR地形效应校正中角度效应校正无法应用于分类的难题，实验结果表明，新方法可有效的去除极化SAR中的AVE效应影响，而且极化SAR的分类总精度可提升9%左右，对于山区的森林类型分类而言，分类精度可提升20%左右；（2）创新了基于空隙穿透假设条件的干涉SAR多层模型，该相干性理论模型更符合短波长InSAR的空隙穿透特点和森林的异质性结构特征，为发展更精准的干涉SAR相干性地形效应校正算法奠定了理论基础；（3）研究了基于P-波段干涉SAR的林下地形提取算法，对比分析了基于RVoG模型和子孔径分解方法的效果，实现了高精度林下地形提取，为多维度SAR协同的地形效应校正奠定了理论基础。