基于多源数据和GORT模型的多尺度森林LAI垂直分布反演研究

Leaf area index(LAI) and vertical foliage profile are important biophysical variables in terrestrial ecosystems, and It plays an important role in forest photosynthesis, respiration, transpiration, rainfall interception, energy exchange and other ecological processes , and it has a significant influence on the exchange of energy and material between the surface and the atmosphere. There are obvious stratification characteristics in the vertical direction, and the contribution of the canopy and the understory vegetation to the material and energy cycle of the forest ecosystem is quite different. However, the internal structural differences are not considered in most of the current inversion methods of forest parameters, which affects the accuracy of the forest LAI estimation. In this study, in order to accurately monitor the vertical foliage profile of forest LAI, the canopy is divided into two layers based on the original GORT model. Considering the multiple scattering among different layers, we build a two-layer GORT model. Based on the two-layer GORT model, multi-source remote sensing data (multi angle, multi spectral MODIS and MISR data and LiDAR full waveform data) were used to estimate the vertical foliage profile of multi-scale forest LAI. Forest canopy LAI and understory vegetation LAI were obtained respectively, providing important data support for accurately estimating material and energy balance in forest area, analyzing succession stage of woodland and biodiversity, and improving forest resource management.

叶面积指数（LAI）是研究陆地生态系统及过程的关键参数，森林LAI是反映森林光合、呼吸、蒸腾、降水截留、能量交换等生态过程的重要因子，显著影响地表与大气间的能量和物质交换，进行森林LAI反演研究意义重大。森林在垂直方向上有明显的分层特征，林上冠层与林下植被对森林生态系统的物质和能量循环的贡献有较大差异。而当前大多数森林参数反演方法不考虑其内部的结构差异，影响了森林LAI反演的精度。本研究为了能够准确地监测森林LAI的垂直分布，在原GORT模型基础上，考虑森林的分层特性及各层间的多次散射，构建双层GORT模型；基于该模型，利用多源遥感数据（多角度、多光谱MODIS和MISR数据及LiDAR全波形数据）反演多尺度森林LAI垂直分布，得到林冠LAI和林下LAI，为精确估算森林区域的物质与能量平衡、分析林地演替、保护生物多样性、提高森林资源及生态管理水平提供重要的数据支撑。

叶面积指数（LAI）是研究陆地生态系统及过程的关键参数，森林LAI是反映森林光合、呼吸、蒸腾、降水截留、能量交换等生态过程的重要因子，进行森林LAI反演研究意义重大。森林在垂直方向上有明显的分层特征，林上冠层与林下植被对森林生态系统的物质和能量循环的贡献有较大差异。针对目前大多数森林参数反演中未考虑其内部的结构差异影响了森林LAI反演的精度，本项目对以下几方面进行了研究，并取得了初步成果：.1）考虑了森林的分层特性及各层的间隙率模型，在GORT模型基础上，构建了双层GORT模型，提高模型的模拟精度；.2）基于改进后的雷达方程和双层GORT模型，利用星载和机载 LiDAR全波形数据、点云数据等，分别反演得到激光雷达的森林冠层LAI的垂直分布，确定林冠和林下植被的分布边界，进而积分得到林冠和林下植被的LAI分布。.3）协同多角度、多光谱MODIS数据及星载/机载LiDAR等数据，利用改进后的双层GORT模型，基于TensorFlow环境，采用MLP多层感知机（全连接）神经网络反演大尺度的森林冠层LAI和林下植被LAI分布。.多尺度森林林冠LAI和林下植被LAI为精确估算森林区域的物质与能量平衡、分析林地演替、保护生物多样性、提高森林资源及生态管理水平提供重要的数据支撑。