多源遥感数据协同反演植被叶面积指数研究

With the developments of the terrestrial observation satellite projects, the extraction method of leaf area index (LAI) based on the multi-source remote sensing data has become a trend. Currently, the LAI inversion algorithms based on the multi-source remote sensing data have two mainly types of method, one way is fusing the existing LAI products, and another way is synthesizing the LAI inversions based on single sensor remote sensing data during in a certain period. However, there is not a LAI inversion algorithm really making use of the rich information provided by multi-source remote sensing data. First, this project makes sure the synthetical theory and method based on the multi-source remote sensing data, quantitatively analyzes the information increments and the errors introduced by the collaborative multi-source remote sensing data, and constructs the multi-source and multi-angle datasets. Then, this project wants to clarify the response mechanism between the vertical structures and the spectral characteristics of vegetation canopy, and develop the canopy reflectance model which consider the vegetation vertical structures. After that, based on the prior knowledge and the information response mechanism between the multi-angle observations and the vegetation structures, this project aims to establish a LAI inversion method based on the multi-source remote sensing data. At last, the accuracy of LAI inversion from multi-source remote sensing data evaluates by the transformed ground-based LAI measurements to pixel-scales. This project is expected to increase the accuracy of LAI extraction by using the multi-angle information provided by multi-source remote sensing data.

随着陆地观测卫星计划的发展，基于多源遥感数据的叶面积指数(leaf area index, LAI)反演已经成为一个发展趋势。但目前的多源LAI反演算法主要利用已有LAI产品进行融合、或者在一个合成周期内对单一传感器遥感数据反演结果进行合成，并没有从真正意义上协同使用多传感器数据提供的丰富信息。本项目明确多源遥感数据协同理论与方法，定量分析多源遥感数据协同引入的信息增量和误差，构建多源多角度数据集；明确植被冠层垂直结构与波谱特征的响应机制，构建考虑植被垂直结构的冠层反射率模型；联合先验知识，以多角度观测的植被结构信息响应机制为指导，建立多源遥感数据协同反演LAI方法；利用转换到像元尺度的地面LAI测量真值对反演结果进行精度评价。本项目有望通过利用多源遥感数据提供的多角度信息增量，提高地表植被LAI参数提取精度和时空连续性。

项目利用多源遥感数据联合优势即借助多传感器提供的多角度观测信息增强地表参数信息量，但在增加多角度信息的同时也会引入观测误差，因此开展了多传感器观测角度分布及信息量评价，以平衡增加观测角度引入的观测误差提高反演精度，研究结果显示在5天外的净信息量趋于稳定。同时考虑到多角度观测可以获取垂直结构信息，基于考虑冠层垂直结构的冠层反射率模型分析冠层植被垂直分层贡献率，研究结果显示浓密植被冠层在郁闭度高的情况下，底层植被贡献小于角度效应影响。多源遥感数据协同反演关键问题在于如何有效利用多角度提供的信息量有效约束可接受解空间维度，因此开展了多源数据多角度查找表反演方法和最优多角度反演方法研究，利用地面实测数据验证结果显示多源数据LAI估算精度较单一角度观测的反演结果相对误差降低10%以上。项目研究相关成果已发表论文6篇，其中SCI论文4篇、EI论文2篇。