小麦矿物粉尘胁迫的高光谱响应机理与诊断方法研究

The dust pollution is getting increasingly severe, and it has caused serious damage to many kinds of crops, such as wheat, rice and corn. However, the traditional detecting methods are unsuitable for the analysis of crop dust pollution stress in the macro-scale, and it urgently needs a new approach to solve this problem. Accordingly, based on the analysis of mineral dust pollution stress response mechanism of wheat leaf and canopy spectra, a method to diagnose crops dust pollution stress by hyper-spectral remote sensing will be opted in this project. Firstly, the wheat hyper-spectral database will be created through the simulation experiments of mineral dust pollution stress, and using it as the data source, the mechanism of mineral dust stress to wheat spectral response will be clarified through analysis of the physiological changes, the spectral response characteristics and it’s physiological significance under mineral dust stress. Secondly, based on the previous step, the characteristic information which can reflect mineral dust pollution stress will be extracted by various methods such as wavelet analysis, fractal algorithm and so on. And using the characteristic information, different hyper-spectral indices, which enable us to identify whether the dust stress of wheat occurred or not, and to determine the dust stress intensities and types, will be designed. Then, the optimal indices, which are sensitive to mineral dust pollution stress effect but insensitive to interference factors, will be screen out by sensitivity analysis. Finally, using the optimal spectral indices, the wheat mineral dust stress diagnosis model will be built, evaluated and optimized by comparative analysis. This project might reveal the hyper-spectral response mechanism of dust pollution stress, and provide a new idea and method to monitor dust pollution stress for wheat and other crops.

日益加剧的粉尘污染已对小麦等农作物造成严重危害，而传统方法又难以在宏观尺度上评估其对作物的影响。本项目以矿物粉尘为例，通过分析其污染胁迫下小麦叶片与冠层光谱的响应机理，以探索小麦等农作物粉尘胁迫的高光谱诊断方法。首先通过田间模拟胁迫试验建立小麦光谱数据库，分析矿物粉尘胁迫下小麦的生理变化与光谱响应特征，以及小麦光谱响应所表征的生理学意义，从而明确矿物粉尘污染对小麦胁迫效应的高光谱响应机理；在此基础上，探索可实现矿物粉尘胁迫特征光谱高效提取的相关方法，构建可进行粉尘胁迫识别、胁迫强度分级以及胁迫类型分类鉴别的光谱指数，并通过敏感性分析，择取其中既可敏感反映胁迫信息，又对干扰因素具有抗性的部分；最后利用优选的指数建立小麦矿物粉尘胁迫诊断模型，并对模型进行评估与优化。研究成果将揭示粉尘胁迫的高光谱响应机理，为小麦等农作物粉尘胁迫的监测提供新的思路与方法。

日益加剧的粉尘污染已对小麦等农作物造成严重危害，而传统方法又难以在宏观尺度上评估其对作物的影响，本项目通过分析粉尘胁迫下小麦叶片与冠层光谱的响应机理，初步建立了小麦等农作物粉尘胁迫的高光谱诊断方法体系。主要研究成果如下：（1）通过模拟与田间试验相结合的方法，分别获取了物理模型模拟光谱数据集与粉尘胁迫实验真实光谱数据集，并建立了光谱数据处理与光谱指数优化的方法体系。（2）基于上述数据分析了粉尘胁迫强度、小麦光谱响应特征以及小麦各理化参量之间内在联系：首先利用各类高光谱指数实现了小麦叶面积指数、叶绿素含量、氮含量以及含水量的高精度估测，为揭示粉尘胁迫对小麦生理状况的影响过程及其光谱响应的机理奠定了基础；随后对粉尘胁迫下小麦叶片理化参量进行了分析，得知在相同生长期，粉尘胁迫下的小麦叶片氮含量、叶绿素含量与含水量等比正常小麦低；最后通过光谱波形分析得知小麦受胁迫时，其反射光谱曲线会出现蓝光与红光吸收减弱，绿峰被削平，红边斜率降低等现象。（3）在实验条件下，建立了小麦粉尘胁迫强度的高光谱分级评价模型。通过田间试验，利用不同胁迫水平下(严重胁迫、中等胁迫、轻度胁迫与正常生长）的小麦叶片光谱数据集建立了分级评价模型。结果表明，模型对不同胁迫强度下小麦样本的预测精度可达87.5％，即在实验条件下利用高光谱进行小麦粉尘胁迫强度的诊断是可行的。（4）在非实验条件下，构建了不同植物粉尘胁迫识别光谱数据集。利用光谱特征分析，光谱角匹配与综合分析等方法对受到不同胁迫强度的样本进行了匹配分析，结果表明各方法对不同胁迫等级样本的识别率在70%以上，说明在非实验条件下，利用高光谱进行植被胁迫强度的分级评价也是可行的。研究成果初步揭示粉尘胁迫的高光谱响应机理，可为小麦等植被粉尘胁迫的监测提供新的思路与方法。