无并发协同数据支持下的农田复杂下垫面遥感反演叶面积指数尺度转换

This project aims at scaling transformation of leaf area index (LAI) inversed from remote sensing data for farmland complex underlying surface, studying on extraction methods of sub-pixel information without the support of concurrent high spatial resolution image, and reaches the goal of developing LAI scale transformation model for landscape pattern with multi-type spatial heterogeneity. Based on spectral linear unmixing analysis and temporal-spatial fusion, with crop phonological information, the high-accuracy reappearance of sub-pixel spatial heterogeneity could be achieved without concurrent required data at high spatial resolution. Specifically for the heterogeneity of farmland, caused by the cover type changes and the density changes within the same crop, the contextural parameters of the mixed pixel and fractal theory would be used to quantitatively describe the scaling effect caused by the two kinds of spatial heterogeneity, respectively. Consequently, the scaling transformation model of inversed LAI for heterogeneous farmland could be built. Based on this research, it is expected to solve the major scientific problem about scaling transformation and correction of land surface parameters inversed by remote sensing data, and also lay the foundation for the improvement of satellite remote sensing product practicability and its validation.

本项目以农田复杂下垫面遥感反演叶面积指数尺度转换为目标，研究无并发高空间分辨率协同数据支持下的亚像元信息提取方法，发展面向多类型空间异质性景观格局的叶面积指数尺度转换模型。项目借助线性混合像元分解、遥感影像时空融合等技术，引入作物物候信息，实现基于非并发协同数据支持下的亚像元空间异质性高精度复演；针对农田下垫面类间异质性（端元组分差异）和类内异质性（作物长势差异）同时存在的复杂景观格局，分别利用像元结构参数与分形理论定量化描述两类空间异质性引入的尺度效应，发展面向农田复杂下垫面的遥感反演叶面积指数尺度转换方法。通过本项目的研究，解决遥感反演地表特征参数尺度转换与校正面临的主要科学问题，为提高卫星遥感产品的实用性及其真实性检验奠定基础。

我国南方农业田块破碎化程度高，混合像元普遍存在，导致表征作物生长状况的叶面积指数（LAI）遥感反演尺度效应明显。项目以农田复杂景观格局下遥感反演作物LAI尺度转换为目标，开展的工作与取得的成果如下：基于物候和地表空间结构信息，分别发展了基于物候和地表空间信息的时空融合改进算法，提高了高时空分辨率遥感影像的重构精度，实现了无高分影像支持的亚像元空间异质性信息高精度提取；针对农田景观格局下类内与类间异质性同时存在的情况，构建了基于分形理论与结构参数的作物LAI遥感反演尺度校正模型，削弱了由尺度效应引起的遥感反演误差；针对应用广泛但存在明显高（或低）估的MODIS LAI产品，构建了一套同时考虑空间尺度效应、成像参数差异以及反演模型差异的MODIS LAI产品校正方法，并开展了面向真实农田复杂下垫面场景的MODIS LAI产品真实性检验的应用。项目成果为遥感卫星产品的普适性研究、真实性检验、误差校正提供新的方法和思路。