In this thesis, we have described satellite data with emphasis on Sentinel-2 satellites. We showed definition of time series and methods for their collection over internet. We compared cloud mask algorithm, used and developed for sentinel-hub portal, with other commonly used cloud mask algorithms. We gave short description of Saviztky-Golay, LOESS and Whittaker-Eilers signal smoothing algorithms with NDVI, EVI and EVI2 vegetation indices.
In the second part of the thesis, we provide a simplified way for getting and storing generalised raster statistical data in Python programming language and Spatialite database. We compared two series smoothing methods concerning input smoothing parameters. Similarly, we compared time series of three smoothed vegetation indices. In the end, we provided method for building comparable vectors and demonstrated our program on simple SVM classification model.
For this thesis, we written program in Python, which is freely available online and simplifies work with Sentinel-2 time series.