基于多源数据的山区公里级叶面积指数反演及验证

Complex terrain accounts for a quarter of the world's land area. Improving the accuracy of leaf area index (LAI) estimation is of great significance to climate change and ecological assessment. Kilometer-scale satellite remote sensing has become the main method of global LAI mapping because of its high-frequency monitoring capability. However, the kilometer-scale pixels in mountainous areas include both the radiation transfer processes between slopes and within the canopy, which makes "terrain-vegetation information coupling" a major difficulty in LAI retrieval and validation. Existing algorithms are usually based on the assumption of flat surface or single-slope surface, which are not suitable for kilometer-scale pixels. Moreover, LAI products did not consider topographic effects leading to huge uncertainties. To this end, this project intends to carry out the following research. Firstly, build a comprehensive radiative transfer model in mountain areas to depict the complete radiative transfer process and decouple terrain-vegetation information. Secondly, combine a digital elevation model (DEM) and high-resolution images to provide terrain and sub-pixel information and use look-up tables to achieve LAI high-precision retrieval. Finally, use novel terrain miniaturization and computer simulation methods to validate the proposed algorithm. This study provides a theoretical basis and practical method for retrieval and validation of LAI in mountainous areas, helps to improve the accuracy of global LAI estimation, and has important scientific significance for remote sensing modeling and parameter retrieval in mountainous areas.

山区约占世界陆地面积的四分之一，提高山区叶面积指数(LAI)估算精度对气候变化和生态评估等研究意义重大。公里级分辨率卫星遥感因其高频监测能力而成为全球LAI制图的主要方式。然而，山区公里级像元同时包括坡面间和冠层内两层次的辐射传输过程，使得地形-植被信息耦合成为LAI反演与验证的主要困难。现有算法通常基于平坦地表或单坡面假设，不适合公里级像元尺度反演；常用LAI产品也未对地形影响做特殊处理，存在不确定性。为此，本项目开展如下研究：首先，构建山地综合辐射传输模型刻画像元内完整辐射传输过程，解耦地形-植被信息；其次，联合数字高程模型和高分辨率影像分别提供亚像元地形和植被分布信息，借助查找表实现LAI高精度反演；最后，利用地面实测、地形缩微和场景模拟对算法进行真实性检验。本研究为反演和验证山区LAI提供理论依据和实践方法，有助于提高全球LAI估算精度，对山区遥感建模及参数反演具有重要意义。

山区约占我国陆表的三分之二，利用遥感手段对山地生态系统叶面积指数（Leaf Area Index, LAI）进行反演及大范围长时序监测，对研究陆表碳循环、实现“双碳”目标具有重要意义。然而，山区具有地形复杂及冠层离散等异质性特征，为LAI的准确反演带来极大挑战。现有模型与算法对山地像元中异质性信息的刻画不完善，是目前制约山地LAI反演精度提高的主要障碍。对此，本项目针对山区地表LAI反演，开展了如下研究：（1）构建了能够准确刻画复杂地形及离散冠层的双向反射率模型，提高了山区冠层双向反射率建模精度，为解耦地形-植被信息提供了模型基础；（2）分析评估了得到广泛应用的MODIS&VIIRS LAI产品的质量及时间序列稳定性，掌握了大尺度LAI产品的时空特征，为提高山区LAI产品生产的稳定性提供了借鉴；（3）分析研究了MODIS LAI反演算法及其他地表参量生产算法的特点及性能表现，为山区LAI反演算法的开发、优化和验证工作提供了参考。本项目所取得的研究成果，有助于更好地了解复杂地形对冠层双向反射率及地表参量反演的影响，对于其他地表参量卫星产品的算法开发与产品生产也具有一定的参考与借鉴意义。