基于冠层结构光谱不变性理论的农作物光合色素含量估算研究

Foliar pigments content is an indicator of crop stress and photosynthesis ability. Imaging spectroscopy is an important technology to detect crop pigments content in a large area. However, canopy reflectance is affected by multiple factors such as atmosphere condition, solar illumination condition and canopy structure, which raises the difficulty and uncertainty for retrieval of foliar pigments content from imaging spectroscopy data. It is crucial to remove the canopy structure effects for foliar pigments content retrieval. Recently, spectral invariant theory provides an alternative approach to solve this problem. In this project, we will acquire imaging spectroscopy data and in-situ measurements for three species (winter wheat, turnip rape and corn) with contrary canopy structures. The aim of this study is to develop robust model for foliar pigments estimation based on spectral invariant theory using radiative transfer model, vegetation indices and machine learning algorithms. First of all, we will clarify the mechanism for foliar pigment estimation based on canopy scattering coefficient. Secondly, we will develop foliar pigments content sensitive vegetation indices based on canopy scattering coefficient. Finally, we will develop models for foliar pigments content estimation combing canopy scattering coefficient and machine learning algorithms. This project will enhance the ability for foliar pigments content estimation via imagining spectroscopy data and further assistant production decision in precision agriculture.

叶片光合色素含量是农作物环境胁迫与光合作用能力的指示器。成像光谱技术是大尺度探测光合色素含量的重要手段。冠层光谱反射率受到大气、光照条件、冠层结构等多种因素的影响，增加了探测叶片光合色素的难度和不确定性。去除冠层结构影响是解决这一问题的关键因素之一。近来，光谱不变性理论的提出为去除冠层结构影响提供了新思路。为此，本项目以冬小麦、油菜、夏玉米三种冠层结构差异显著的典型农作物为研究对象，采用成像光谱数据和实测农作物生理生化参数，以光谱不变性理论为基础，结合辐射传输模型、植被指数和机器学习算法，探索光合色素含量估算方法。并拟对如下问题开展研究：(1)冠层散射系数估算叶片光合色素机理研究。(2)基于冠层散射系数的叶片光合色素敏感植被指数研究。(3)基于冠层散射系数与机器学习算法的光合色素估算模型研究。本项目将极大提高成像光谱技术估算农作物叶片光合色素的能力，为农业精细化管理和生产决策发挥巨大作用。

项目的主要研究内容是开发适用于多种不同冠层结构类型的农作物叶绿素含量遥感估算方法。通过实测多种农作物冠层光谱反射率，叶绿素含量和冠层结构参数与植被冠层辐射传输模型结合开展了相关的研究工作，包括：开展基于多种机器学习算法估算农作物冠层结构参数的方法，为开发适用于不同冠层结构类型农作物叶绿素含量遥感估算提供先验知识；构建了新的叶绿素含量敏感而冠层结构不敏感光谱指数，为采用植被指数大尺度估算多种农作物叶绿素含量提供了依据；提出了冠层反射率与冠层散射系数相结合的农作物叶绿素含量遥感反演新方法，为遥感估算多种类型冠层结构农作物叶绿素含量提供了新思路。本项目投稿和撰写SCI论文3篇，培养研究生3名。