基于冠层光辐射异质性的小麦氮素诊断机理研究

The nondestructive sensing of wheat nitrogen status is critical for recommending top-dressing N amount, so as to achieve higher yield and high-efficiency utilization of agricultural resource efficiency. In view of the practical problems of lower precision in monitoring N affected by canopy spatial distribution, this research intends to explore the field wheat anisotropy and sensing mechanism of plant nitrogen diagnosis by comprehensive utilization of three-dimensional permeability of multi-angle reflection spectrum and transmission spectrum through the vegetation canopy. The field experiments were conducted with different plant type, planting density and nitrogen fertilizer level, and time-course measurements were taken on canopy spectral reflectance, plant light distribution, nitrogen vertical gradient by layered or directed testing method. The study investigated temporal and spatial dynamic characteristics of nitrogen components, canopy multi-angle reflectance spectra and canopy transmission spectral energy with fusion of spectral analysis, crops cultivation and plant nutrition under different cultivation measures. The project explain covariant relationship of multi-angle reflectance, transmission spectral energy and plant N vertical distribution, further extract suitable spectral parameters to indicate plant nitrogen composition and its abundance level. Plant nitrogen diagnosis models are established based on new fusion technique of the reflected light direction and transmitted light spatial variability. The indirect approach of early distinguishing wheat N deficiency with multi-source remote sensing were developed, which provided the theory basis and the key technology for monitoring and precisely evaluating wheat growth characteristics. The results of this project not only can enrich crop nondestructive monitoring theory, and broaden early N diagnosis method, which improve the level and efficiency of crop N management to promote fertilizer efficiency, yield loss inhibition and environmental protection.

本研究针对当前作物氮素遥感监测精度受冠层空间性的影响和限制这一亟待解决的实际问题，综合利用多角度反射探测冠层深度的优势，透射光谱穿透植被冠层的特点，探讨田间小麦冠层光辐射异质性及氮素诊断的感知机理。依托不同株型、种植密度和氮肥水平组合的小麦田间试验，采用多角度反射光谱、网格化冠层光分布和分层氮梯度测试方法，将定量分析光谱学、植物营养学与小麦栽培生理学知识相结合，研究株型、密度和氮肥等栽培因子对植株氮素组分、多角度反射光谱和冠层透射能谱的影响，解析反射光谱的方向性、冠层光衰减的空间性、氮素垂直分布的梯度性规律及三者间耦联机制，提取适宜表征小麦氮素组分及丰缺状况的专属能谱参数及其生物物理学关系，构建基于反射光方向性和透射光空间性相融合的小麦植株氮素诊断模型及感知技术。研究结果不仅丰富作物生长无损监测理论，而且拓宽氮素遥感监测和早期诊断方法与技术，提高作物氮肥精确管理水平，实现肥料减施增效环保。

作物氮素的快速监测与精确诊断是国内外农业遥感的一个重要研究热点。本项目依托不同品种、地点、水分、种植密度和氮肥水平组合的小麦田间试验，采用多角度反射光谱、网格化冠层光分布和叶层氮梯度测试方法，基于植物营养学、小麦栽培生理学知识原理采用定量分析光谱学、模拟模型、计算数学等方法确立了小麦氮素监测模型及氮素精确诊断方法。.（1）各器官临界氮浓度与生物量间均符合幂函数关系（N=aDW-b），当生物量相同时，灌水处理临界氮浓度高于不灌水处理。双Logistic模型可以很好地拟合LAI的动态变化，单Logistic模型则能很好地拟合地上部氮素积累量（AGNU）、地上部生物量（AGDW）和叶片氮含量（LNC）变化轨迹。.（2）基于冠层反射光谱随入射角及观测角的变化，构建并筛选对角度变化反应敏感的植被指数。新建植被指数WANI对叶片氮含量方程拟合效果明显改善，在天顶角0°到-30°内可建立不同栽培条件统一监测模型。利用红边特征及面积算法构建的新型植被指数mRPA在-20°至10°观测角度范围内可以建立统一的氮积累量监测模型。.（3）通过分析不同冠层空间位点的光截获量与主要生长指标间相关性，PAR1-3与LAI相关性在各个生育时期均较好，可以用统一的线性方程监测LAI变化；叶面积氮指数与PAR1-3相关性好于LNC和LAI。利用R/FR1-9可以较好指示叶面积指数、植株氮含量、植株含水量和植株生物量的动态变化，监测精度较高。.（4）基于FA-BPNN的叶片氮含量反演精度在不同观测角度下均高于常规光谱参数。在分析生物量与氮含量机理关系基础上，构建了多个植被指数协同、且对氮素营养指数NNI敏感的复合植被指数(PPR, PSNDb, SAVI)，其对NNI拟合精度达到0.78。支持向量机比传统机器学习方法运算速度快，对NNI具有良好的预测性和适用性，训练集和测试集R2高于0.86。. 本项目构建了基于反射光方向性和透射光空间性的小麦植株氮素诊断模型及感知技术，不仅丰富作物生长无损监测理论，而且拓宽氮素监测及诊断方法，提高作物氮肥精确管理水平，实现肥料减施增效环保。

内容获取失败，请点击重试