基于全生育期氮素动态利用率的冬小麦高光谱遥感估产

Crop yield estimation using remote sensing is one of the challenges in the frontier of hyperspectral application. Taking winter wheat as the research object, using hyperspectral remote sensing technology as the foundation, from the point of view of the contribution of nitrogen utilization to the final yield of winter wheat, this project focuses on the following contents:.(1) During the annual important growth stages of winter wheat, extracting the general characteristic spectral information that can comprehensively reflect nitrogen content in soil and plant to establish an integrated monitoring model for soil and plant nitrogen content detection. Then a new index of nitrogen dynamic usage efficiency (NDUE) in important growth period of winter wheat was constructed based on hyperspectral information..(2) Combined with mechanism of plant photosynthesis, verify the physiological mechanism of nitrogen absorption and utilization in winter wheat in different growth stages and explain the change characteristics of NDUE with time going. Furthermore, explore the contribution weights of plant nitrogen absorption and utilization to the final yield..(3) Using the obtained contribution weights to transform the nitrogen characteristic spectral information, we can obtain the integrated spectral index, which can reflect the grain yield information of winter wheat. Based on the spectrum, the winter wheat final yield estimation models were established. Then, the NDUE and yield estimation method was carried out using the ground-observation and UAV remote sensing data.

农作物产量遥感估测是高光谱应用研究前沿领域的难点。本项目以冬小麦为研究对象，以高光谱遥感技术为基础，从冬小麦氮素利用对最终产量的贡献角度出发，重点开展以下研究内容：.（1）.在冬小麦年度重要生育期内，挖掘表征背景土壤与冬小麦作物氮素含量的特征光谱信息，建立小麦土壤-作物氮素含量一体化监测模型，构建基于高光谱信息的冬小麦重要生育期氮素动态利用效率新指标；.（2）.结合作物光合作用发生机理，明确冬小麦不同生育期氮素吸收利用的生理机制，解析氮素动态利用率随生育期的推进而产生的变化特征，探究冬小麦不同生育期作物氮素吸收、利用对最终产量的贡献权重；.（3）.利用所获得的贡献权重对不同时期所获的氮素通用特征光谱信息进行变换,获得可以反映冬小麦产量的综合光谱指数。以此为基础，构建基于综合光谱指数的冬小麦产量估算模型。随后，开展基于地面观测、无人机遥感数据的冬小麦氮素动态利用率与产量估算应用示范。

农作物产量估测是遥感应用研究前沿领域的重点和难点。本项目以冬小麦为研究对象，开展基于近地-无人机-卫星多尺度遥感平台的冬小麦养分动态监测与产量高精度预报研究。首先，利用改进优化策略的灰狼算法提取冠层光谱中可有效表征土壤、冬小麦植株氮素含量的特征光谱信息，利用特征分解与深度融合算法提出可表征冬小麦-土壤氮素含量的特征基矩阵，进而构建适用于不同遥感尺度的农田土壤-作物氮素含量一体化精准监测模型。解析作物生长过程中氮素营养分配以及光合作用动态变化，结合绝对目标改进的分析层次过程理论，量化不同时期冬小麦作物生长对最终产量形成的贡献权重，获得可以综合反映冬小麦产量的有效加权光谱信息，分别构建了基于机器学习算法以及时序预报分析法的冬小麦产量高精度预测模型。此外，针对冬小麦作物叶绿素、叶面积指数等作物生长过程关键参数，构建了基于特征选择与机器学习相结合的高精度反演算法，实现了冬小麦关键长势参数的高精度、通用反演。本项目实现了近地-无人机-卫星等多遥感平台数据尺度转换、模型平移等关键技术突破，实现冬小麦营养长势、籽粒产量的区域化监测与估算。