高分辨率反射率时间序列重建算法研究

Research on High-Resolution/Reflectivity Time-Series Reconstruction Algorithm

该文主要围绕着高分辨率植被反射率时间序列重建过程中所遇到的数据噪声多、数据缺失量大等问题,提出基于气象数据和物候信息的高分辨率反射率时间序列重建算法(HTRA)。HTRA基于重建预测模型构建、时序数据去噪算法研究、重建策略建立为主要内容,以神经网络和多源数据耦合为主要技术,利用真实值进行重建。以华北平原的一个区域为研究区,重建得到该区域农田的红光反射率和近红外反射率2016—2019年时空连续数据。通过对重建结果进行定性的分析验证可以得到,HTRA算法具备很强的抗噪、抗缺失能力,重建出的反射率能够很好的重建出植被本身的物候特征。

This paper focusing on the problems encountered in the reconstruction of high-resolution vegetation reflectance time series,such as excessive data noise and large amount of data missing,a high-resolution/reflectivity time-series reconstruction algorithm(HTRA)based on meteorological data and phenological information is proposed.HTRA is based on reconstruction prediction model construction,temporal data de-noising algorithm research,reconstruction strategy establishment as the main content,neural network and multi-source data coupling as the main technology,and uses the real value for reconstruction.Taking an area of the North China Plain as the study area,the spatial-temporal continuous data of the red light reflectance and nearinfrared reflectance of the farmland in the area from 2016 to 2019 were reconstructed.Through qualitative analysis and verification of the reconstruction results,it can be concluded that the HTRA algorithm has strong anti-noise and anti-deletion ability,and the reconstructed reflectivity can well reconstruct the phenological characteristics of the vegetation itself.