基于叶片异质性辐射传输模型的油菜氮素时空动态监测机理与方法研究

Monitoring temporal-spatial dynamics of crop nitrogen has important economic and ecological significance for conducting precise fertilization, improving high yield and quality, and reduce environmental pollution. The current crop nitrogen monitoring mainly relies on the empirical / semi-empirical statistical modeling based on the spectral information acquired from the crop canopy or single leaf, and lacks fundamental understanding of leaf nitrogen to the spectral response. In addition, the heterogeneity in leaf optical, biochemical, and structural properties will also affect the robustness and the accuracy of leaf nitrogen diagnosis. This research was aimed to develop a leaf-heterogeneous radiative transfer model and solve an inverse problem to quantify nitrogen status in oilseed rape leaves. Leaf-nitrogen-vertical-distribution characterization factor, leaf anisotropy factor and bidirectional reflectance distribution function (BRDF) are introduced to PROSPECT-5 model, and PROSPECT-Ht model is developed with the time extension function. Multi-step inversion strategy and intelligent optimization algorithms were used to estimate the nitrogen content from PROSPECT-Ht model. The nitrogen dynamics monitoring method is finally developed based on the PROSPECT-Ht model coupled with three-dimensional spectral imaging. The results provide a fundamental understanding of the nitrogen dynamics as well as the leaf spectral response. This research would not only improve the robustness and accuracy of leaf nitrogen monitoring but also provide a new way to understand the nitrogen accumulation, loss and transfer in oilseed rape.

作物氮素时空动态监测对实现精准施肥、促进作物优质高产和减少环境污染具有重要的经济和生态意义。目前作物氮素监测主要以基于作物冠层或单一叶片波谱信息的经验/半经验模型方法实现氮素诊断，存在氮素波谱响应机理不明和时空异质性解析困难等问题，致使氮素监测的普适性和准确性难以保证。本项目拟以油菜为研究对象，将叶片氮素垂直立体表征因子、叶片各向异性因子以及二向性反射分布函数引入辐射传输模型，通过时间扩展，构建叶片氮素时空异质性辐射传输模型；采用多阶段反演策略和智能优化算法建立基于叶片异质性辐射传输模型的氮素定量反演方法；探明油菜不同生育期和叶层氮素转运分配和波谱响应规律，揭示叶片氮素时空异质性监测机理，建立全生育期油菜氮素时空动态监测的三维立体图谱方法，并进行大田试验验证与评估。研究结果有助于提高氮素监测的普适性和准确性，也为更全面地揭示油菜氮素积累、损失、运转规律提供新思路。

氮素是作物生长发育中不可缺少的重要营养元素，实施油菜氮素的时空动态监测可以更全面地揭示氮素积累和损失规律，为油菜精准科学施肥提供理论方法和技术支撑。遥感光谱技术作为一种无损快速的分析手段，越来越多的应用于作物的农学参数监测与反演。而现有的方法无法满足实际生产管理中对氮素诊断的普适性和准确性要求，因此如何利用遥感光谱信息实现作物氮素准确定量解析是本领域的一个研究热点。本项目探明了不同氮素处理水平下油菜叶片时空异质性对氮素的波谱响应机理，阐明了叶绿素荧光信号与植株氮素营养状况之间的动态关系，结果表明缺氮会降低苗期油菜叶片的光系统I和II性能，通过最大光化学效率等荧光参数可对氮素胁迫进行诊断。突破了基于辐射传输机理模型的叶片生化参数反演技术，叶绿素含量的反演误差低于15%。分析了不同光谱分辨率、光谱噪声、生长时期和氮素处理水平对叶片生化参数反演的影响，构建了基于光谱指数的多尺度叶片生化参数评估方法，相对误差低于14%。为了进一步提升氮素监测的准确性，建立基于TOF相机的作物三维形态结构重建方法，提出了基于TOF相机的植物三维点云优化处理方法，有效点占比均值达92.05%，实现了不同生长时期的油菜三维形态结构的准确评估，为后续与图谱数据融合提供了三维信息。研究了基于深度学习的近端遥感图像校正方法，构建了面向油菜冠层生理参数的三维立体图谱解析方法，解析精度达96%，建立了稳健的油菜氮素营养诊断模型。研究成果对指导精准施肥、促进作物优质高产和减少环境污染具有重要经济和生态意义。