Research Of Harmonic Analysis Extension Method On Rmotely Sensed Data Temporal Filter And Reconstruction

As the continuously innovation of satellite and sensor technology, more and more remotely sensed observations have been archived. As the important input, the remotely sensed time series data has been widly used in regional and global environmental change research. With the deep development of the research, there is an increasing demand of the spatio-temporal continuity and intergrity of remotely sensed time series data. However, the remotely sensed time series data are inevitably influenced by the observation conditations and the instrument failure during the data acquisition process causing a lot of information missing, which severely hiders the further applicaton of these data. How to reconstruct the spatio-temporal continuious and complete remotely sensed time series data under the current observational condition have facilitated the development of the temporal filter and temporal reconstruction technology. This paper will propose new temporal filter and temporal reconstruction methods to overcome shortages of the existing methods focusing on improving the spatio-temporal continuity and intergrity of the remotely sensed time series data.The main work is summarized as follows:(1) Focusing on improving the spatio-temporal continuity and intergrity of the remotely sensed time series data, this paper analyzes the problems existing in the remotely sensed time series data and introduces the study of the restoration approaches of remotely sensed images. And it also summarizes their blemishes and shortages in real applications to bring forward the research goal.(2) Proposing the moving weigthed harmonic analysis NDVI temporal filter method. In order to overcome the blemish of the HANTS method in over fitting and over smoothing, this paper introcduces the moving support domain to the HANTS method. In each of the moving support domain, the weight of each point was calculated using cubic spline method. These weights are used to control the impact of the points in reconstruction process. In addition, this paper desingend a four-step processing flow based on the characteristics of the NDVI to make the reconstructed results approaching the upper envelop of NDVI, which could currectly obtain the real growth change tendency of the vegetation. The experiments have shown that this method could identify the noising points correctly and make the reconstructed results approaching the upper envelop of the NDVI time series. Especially, this method could correctly estimate the NDVI values in vegetation dormant period and it performs well in dealing with the successive fluctuation stations in many cases.(3) Proposing the harmonic analysis and poisson equation synergistic land surface reflectance temporal reconstruction method. In consideration of some defects of the existing methods that they cannot effectively reconstruct daily remotely sensed time series data, a new method is proposed in this paper aimed to overcome these defects. Different from the temporal filter methods, this method is aimed to fill the gaps rather than denoising using the temporal information. The main idea is firstly to calculate the initial values using the multi year weighted average, and then temporal interpolating the gaps remained after step 1 using HANTS with true value constraint. Finally, a seamless post-processing is carried out using the poisson image editing. The experiments have indicated that this method could reconstruct daily spatio-temporal continuous and complete surface reflectance products, which also keep the spectral completeness.(4) Proposing the physical constraint considered land surface temperature temporal reconstruction method. In consideration of the impact of the cloud to the land surface temprature, our method is designed to couple the related physical quantities to reconstruct more real land surface temperature products. This method is firstly to reconstruct the high quality land surface temperature products under clear sky conditions using the temporal information of the time series data. And then reconstruct the information under the cloud through building the relationships between the related physical quantities and the original data. The experiments have shown that this method can not only promote the quality of the low quality pixels under clear sky conditations, but also correctly reconstruct the pixel values under cloud.