Research On Construction Method Of High Spatiotemporal Resolution NDVI On Complex Surface

Due to the universality of surface heterogeneity,monitoring of ecological and environmental changes at the global and regional scales urgently requires remote sensing data with high spatial and temporal resolution as support.However,due to the limitations of remote sensing satellite technology,it is difficult for the existing remote sensing data to meet the requirements of high temporal and spatial resolution over long time periods.Especially in the fog-covered area,the available high spatial resolution data(such as Landsat)can only be obtained at individual observation moments each year.In order to alleviate the contradiction between the spatial and temporal scales of remote sensing data,scholars at home and abroad have proposed spatiotemporal data fusion method to synthesize Remote sensing data with a high spatiotemporal resolution.However,the existing spatiotemporal fusion algorithms are usually highly hypothetical and their applications on complex land surfaces with strong spatial heterogeneity or diverse cropping systems are greatly limited.At the same time,most of the algorithms are based on research areas with simple surface coverage or large base-pair data,and there are few studies on complex surface with limited available data and diverse planting systems.Based on the RTSM algorithm,this paper selects the Guanzhong area,Shaanxi province,covered by cloud and fog as the research area,and focuses on the research of the construction method of high spatiotemporal NDVI of the double-season crops under complex surface with limited available data and diverse planting systems.In this paper,an adaptive fitting curve is introduced to improve the curve fitting method of RTSM algorithm,and the RTSM algorithm is extended to the application of complex subsurface with diverse cropping systems.The main works are as follows:(1)An adaptive fitting curve was introduced to improve the curve fitting method of RTSM algorithm.The applicability of the RTSM algorithm is thus extended to complex land surface with diverse cropping systems.The improved RTSM algorithm is more suitable for high spatial and temporal NDVI construction of complex surface.According to the real growth and change rules of different types of ground objects,the Hants curve fitting model can adaptively and effectively fit its growth and change curves by selecting reasonable relevant parameters.The fitted curve accurately represented the crop growth rules under different cropping systems.(2)Different research areas were selected with respects to small data volume,general data volume and large area scale to verify the accuracy of the improved RTSM algorithm.The RTSM algorithm was also compare with the ESTARFM(Enhanced Spatial and Temporal Adaptive Reflectance Fusion Model,ESTARFM)algorithm.The experimental results show that the improved RTSM algorithm is more suitable for double-season crop areas under complex surface.The determination coefficient of predicted image and real image generated by the improved RTSM algorithm is higher than that of ESTARFM algorithm.Meanwhile,the curve of NDVI time series generated by the fusion results is smooth and continuous,which can better reflect the real growth trend of two-season crops under the complex surface.(3)Different numbers and distributions of base pair data sets were selected and used for spatiotemporal data fusion.By comparing the fusion results of the improved RTSM algorithm and ESTARFM algorithm based on different base pair data conditions,the sensitivity of the two algorithms to the base pair data set was evaluated.Through spatial comparison,scatter plot comparison and time trend comparison,it is found that when there are few base pair data,the inclusion of peak period data can greatly improve the accuracy of the ESTARFM algorithm results.The improved RTSM algorithm is less sensitive to base pair data sets.With a large number of base pair data,the two algorithms are less affected by the peak period data.At the same time,increasing the number of base pairs can improve the fusion accuracy of the two algorithms.In general,the predicted images generated by the improved RTSM algorithm are more correlated with the real images based on the same base pair data set.(4)In order to ensure the high accuracy of the fusion results and the vegetation time series curve,,the sensitivity of the improved RTSM and ESTARFM algorithms to base and data was studied.The distribution and quantity of the base pair data should be considered comprehensively when using the spatiotemporal fusion algorithm.The improved RTSM algorithm and ESTARFM algorithm should select as many observation data as possible as the base pair data.When there is peak period data in the observation data and the ESTARFM algorithm is used for spatiotemporal data fusion,the peak period data should be selected as the base pair data as much as possible to make the predicted image closer to the real image and the time series curve generated is more in line with the true growth trend of vegetation.