联合邻域像素观测信息的单基线PolInSAR森林高度反演方法

张兵; 赵薇; 朱建军; 宋伟东; 任东风; 朱洪波; 胡忠超

doi:10.13203/j.whugis20240498

联合邻域像素观测信息的单基线PolInSAR森林高度反演方法

1 辽宁工程技术大学测绘与地理科学学院, 辽宁阜新, 123000;
2 中南大学地球科学与信息物理学院, 湖南长沙, 410013;
3 阜新生态集团数字科技研究院有限公司, 辽宁阜新, 123000

基金项目:

国家自然科学基金(42204031)

中国博士后科学基金(2022MD723791)

辽宁省教育厅基本科研项目(LJ212410147015,LJ212410147065)

2024年度辽宁省自然基金联合基金计划项目(20240324,20240315)

详细信息

作者简介:
张兵，博士，副教授，研究方向为极化干涉SAR参数反演、复数测量平差理论。zhangbing@lntu.edu.cn

通讯作者:
朱建军，博士，教授。zjj@csu.edu.cn

计量
- 文章访问数: 2
- HTML全文浏览量: 0
- PDF下载量: 2
出版历程

A Single-Baseline PolInSAR Forest Height Inversion Method with Joint Neighborhood Pixel Observation Information

1 School of Surveying, Mapping and Geoscience, Liaoning Technical University, Fuxin 123000, China;
2 School of Geoscience and Information Physics, Central South University, Changsha 410013, China;
3 Fuxin Ecological Group Digital Science and Technology Research Institute Co, Fuxin 123000, China

摘要

摘要: 目前极化合成孔径雷达干涉测量技术（polarimetric interferometric synthetic aperture radar,PolInSAR）在森林高度反演领域应用最为广泛的模型为随机地体二层散射模型（random volume over ground model,RVoG），但模型未知参数较多，基于单基线PolInSAR数据难以进行参数解算。鉴于此，提出了联合单基线邻域像素PolInSAR观测信息的森林高度反演方法，基于传统RVoG模型，假设在邻域像素内森林高度及消光系数保持不变，地体幅度比参数随像素改变而改变。方法可以理解为以牺牲部分空间分辨率换取丰富的观测信息，解决单基线配置下基于RVoG模型的参数反演必须借助先验信息的问题。通过采用覆盖Krycklan针叶林实验区以及Mabounie热带雨林实验区的PolInSAR数据对提出的联合邻域像素单基线森林高度反演算法进行了实验验证，结果表明，方法相对传统解算策略在针叶林实验区均方根误差分别减小25%及19%、在热带雨林实验区均方根误差分别减小23%及15%。该方法适用于森林高度反演。
- 随机地体二层散射模型 /
- 极化合成孔径雷达干涉测量技术 /
- 森林高度反演算法 /
- 参数解算 /
- 空间分辨率
Abstract: Objectives: The most widely used model for Polarimetric Interferometric Synthetic Aperture Radar (PolInSAR) in forest height inversion is the random volume over ground (RVoG) model. However, there are many unknown parameters in the model, and it is difficult to solve the parameters based on single baseline PolInSAR data. Methods: In view of this, a forest height inversion method that combines the PolInSAR observation information of single-baseline neighboring pixels is proposed, based on the traditional RVoG model, assuming that the forest height and extinction coefficient remain unchanged within the neighboring pixels, and that the ground-body magnitude ratio parameter changes with pixel changes. The method can be understood as sacrificing part of the spatial resolution for rich observation information, which solves the problem that the parameter inversion based on the RVoG model in the single-baseline configuration has to rely on a priori information. Results: The proposed joint-neighborhood pixel single-baseline forest height inversion algorithm is experimentally validated using PolInSAR data covering the Krycklan coniferous forest experimental area and the Mabounie rainforest experimental area, and the results show that, compared with the traditional solving strategy, the root mean square error of the proposed method are reduced by 25% and 19% for the coniferous forest experimental area, and by 23% and 15% for the rainforest experimental area. in the coniferous forest and 23% and 15% in the tropical rainforest, respectively. Conclusions: Therefore, the single-baseline PolInSAR forest height inversion algorithm with joint domain pixels can obtain higher accuracy compared to the traditional solution strategy and is more suitable for forest height inversion.
- random volume over ground model /
- polarimetric interferometric synthetic aperture radar /
- forest height inversion algorithm /
- parameter solving /
- spatial resolution