综合前期光谱和上茬作物时序遥感数据的冬小麦播期监测方法研究

There is a strong demand for monitoring sowing date of winter wheat as early as possible in the early growing stages. It is later to some extent to get sowing date by ordinary vegetation phenology methods using remotely sensed data involving the whole growth period after the harvest,however the estimating accuracy based on the spectra at wheat early growing stage of the season is constrained because of the weak vegetation spectral signal. This proposal will develop a novel model to monitor sowing date by integrated information from spectral characteristics of wheat at early stage and time-series data of crops in previous season ,based on the spectra difference of winter wheat at early stage caused by the different sowing date, and on the different time series patterns from the previous crops to the seeding and to early stage of wheat in order to improve the accuracy and time, and provide new ideas and theoretical basis for the sowing dates monitoring. Our studies including: (1) Using spectra from plot experiment and simulation data of the crop growth and radiative transfer coupled models, mechanism of winter wheat canopy spectra responsive difference will be analyzed to find the optimal time window and the best spectrum. (2) Based on the time series features from the previous crops to early stage of wheat, the new remote sensing index will be created for sowing date monitoring. (3) The new sowing date monitoring model will be developed by integrated information from spectral characteristics of wheat at early stage of the season and time series remotely sensed data of crops in previous season. (4)The data of plots sampling and sowing date results at regional scale from other method with different remote sensing data sources will be used to validate the sowing date results from our studies,then optimize model and analyze uncertainty.

在冬小麦生长早期实现播期遥感监测有强烈需求。常规植被物候方法因需全生育期遥感数据参与引起冬小麦播期监测时间滞后；仅利用冬小麦生长前期光谱进行播期监测的方法虽时间提前，但因植被光谱信号较弱导致精度不高。项目根据冬小麦生长前期不同播期冠层光谱响应差异机制，以及上茬作物成熟—冬小麦生长前期时序遥感数据变化特征，提出综合利用冬小麦生长前期光谱和上茬作物时序遥感数据构建播期监测新模型，为播期遥感监测提供新思路和理论依据，提前监测时间和提高精度。研究内容包括：(1)基于大田实验数据以及耦合作物模型和辐射传输模型的模拟数据，研究不同播期冬小麦生长前期光谱响应差异机制，优选监测时相和光谱；(2)构建基于上茬作物时序数据表征冬小麦播期的新型遥感光谱指数；(3)研究集成前期光谱和上茬作物时序数据的冬小麦播期遥感监测综合模型；(4)对播期遥感监测模型进行典型地块抽样验证、区域全覆盖交叉验证，分析不确定性并优化。

本项目针对当前利用全生育期遥感数据基于植被物候遥感监测方法拟合播期时间滞后、仅利用冬小麦生长前期光谱监测播期因弱植被光谱信号引起的监测精度不高的问题，在北京市昌平区小汤山国家精准农业示范基地和河北省冬小麦种植区开展实验，深入挖掘冬小麦生长前期和上茬作物生长后期的多时相高光谱、多光谱数据与实测基础参数、播期间的关系等，并通过改进的ESTARFM算法生成河北省高时空分辨率遥感数据集，从地面实测高光谱、无人机多光谱、卫星多光谱三个层面对上茬作物及冬小麦生于前期的时序特征、光谱特征进行分析，实现了北京小实验区和河北大实验区的冬小麦播期遥感监测。主要从以下4个方面开展研究：(1)不同播期冬小麦生长前期冠层光谱差异机制和变化规律。(2)利用上茬作物时序遥感数据表征冬小麦不同播期的新型遥感光谱指数。(3)综合前期光谱和上茬作物时序遥感数据的冬小麦播期遥感监测模型。(4)冬小麦播期遥感监测结果验证和方法优化。研究发现：基于上茬作物和冬小麦生长前期时序遥感数据构建的物候曲线可实现在冬小麦越冬期实现播期及时准确提取；每年11月中下旬-12月中下旬是河北地区冬小麦播期遥感监测的合适窗口；利用红波段、近红外波段，尤其是辅助红边波段可实现冬小麦生长前期低植被信号下播期准确反演；基于多元线性回归模型和包络线模型可实现播期遥感监测模型的构建。研究产生了综合前期光谱和上茬作物时序遥感数据的适宜冬小麦播期监测的新型遥感光谱指数、冬小麦播期遥感监测模型、河北省2017年冬小麦种植区播期遥感监测专题图、高时空分辨率遥感影像系列图集等一系列研究成果，发表了SCI、EI、核心期刊等期刊论文10篇、专利2个、软著2个。该项目研究成果不仅可为冬小麦播期遥感监测、稀疏植被遥感监测等提供研究思路方法借鉴，还可为农业部门冬小麦管理提供信息支持。目前研究成果已被河北省农业技术推广总站所采用，为该区域冬小麦播期快速监测和科学管理提供了有力支撑。