期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2006
卷号：XXXVI Part 7
出版社：Copernicus Publications
摘要：Long time series of satellite data provide key input for geographically explicit monitoring and modelling of the environmental condition in the context of land use and land cover change (LUCC) analysis. Within the present study, the quality of vegetation index (VI) time series data was analysed under the assumption that a post-processing filtering algorithm considerably reduces noise and enhances the usability for land cover degradation monitoring purposes. As a prerequisite for data analysis, a filtering workflow for VI time series data was developed that accounts for disturbances in the satellite data time series which are related to atmospheric effects. The filter algorithm includes linear interpolation of cloudy pixels, long-term change trend calculation, iterative filtering and fitting of the new time series based on the Savitzky-Golay filter algorithm. Missing or uncertain values are replaced by long-term trend values and fitted using a polynomial function fitting method The time series filtering results in a new data set of enhanced temporal trajectories of VI values. The filter method was applied to a four year time series of MODIS vegetation index data (MOD13Q1). The results indicate the high potential of the Savitzky-Golay based filtering workflow to minimize noise in lower frequencies while conserving natural peaks of VI values that are related to phenological and seasonal phenomena in vegetation physiology. As a consequence, the filtered time series data provide improved input for the assessment of climate related versus human induced land degradation in semi-humid regions
