Extending leaf optical property models for improved retreval of surface and biochemical characteristics

鉴于加拿大作为粮食生产国的作用，以及气候变化对全球粮食安全构成的相关威胁，开发对环境压力具有更高耐受性的新作物是我们独特的挑战。 在加拿大和世界各地，公共资金的主要投资正在用于作物开发和植物科学的高通量数据采集产品和服务。这些“植物表型分析”系统使用了一些仪器，但一般都是围绕近距离高分辨率图像和光谱图像建立的。然而，所需的模型，以可靠地确定基础的植物生物化学的基础上，这些数据源尚未开发和验证在亚冠层尺度。没有这些模式，这些投资在我们全球粮食安全方面的全部价值就无法实现。在萨斯喀彻温大学的一个主要的植物表型倡议的背景下工作，这项研究计划的长期目标是通过改善植物和植物冠层光谱图像的详细生物物理和生物化学信息的提取来实现投资的价值。 为了支持研究计划的长期目标，我们将通过对模型参数进行敏感性和相关性分析来比较现有的叶片光学特性模型。这将有助于管理模型复杂性，同时改善参数检索。我们将致力于将叶片表面特性的模型相结合的定向散射，偏振和表面测量技术参数化叶片表面特性。通过这些努力，我们将扩展目前的叶片光学特性模型，包括额外的色素和表面性状。 通过发展描述叶片光学特性的高保真模型， 将使植物生理学和育种研究充分利用 在光谱成像基础设施方面的主要公共和私人投资。 这些模型将促进关键性状的高通量测量 与粮食作物的耐旱性、耐热性和耐寒性有关， 作物育种工作。这项工作还将缓解目前缺乏 用于遥感和环境领域叶模型开发的数据集 全球监测。

Given Canada's role as a food producing nation, and related threats posed by climate change to global food security, developing new crops with improved tolerances to environmental stresses is a challenge we are uniquely positioned to address. In Canada, and around the world, major investments of public funds are being made into high-throughput data acquisition products and services for crop development and plant science in general. These “plant phenotyping” systems employ a number of instruments but are generally built around close-range, high-resolution imagery and spectral imagery. However, the models required to reliably determine the underlying plant biochemistry based on those data sources have yet to be developed and validated at the sub-canopy scale. Without those models the full value of these investments in our global food security will not be realized. Working within the context of a major plant phenotyping initiative based at the University of Saskatchewan the long-term objective of this research program is to realize the value of investments made by improving the extraction of detailed biophysical and biochemical information from spectral imagery of plants and plant canopies. In support of the long-term objectives of the research program, we will compare existing leaf optical property models by performing a sensitivity and correlation analysis of model parameters. This will inform efforts to manage model complexity while improving parameter retrievals. We will work toward incorporating leaf surface characteristics into the models by combining directional scattering, polarization, and surface measurement techniques to parameterize leaf surface characteristics. Informed by these efforts we will extend current leaf optical property models to include additional pigments and surface traits. By developing hi-fidelity models describing leaf optical properties we will allow plant physiology and breeding research to make full use of major public and private investments in spectral imaging infrastructure. These models will facilitate high-throughput measurement of key traits linked to drought, heat, and cold tolerance in food crops, aiding in crop breeding efforts. This work will also alleviate the current lack of datasets for leaf model development in remote sensing and environmental monitoring world-wide.