In the following, the most important steps of the processing workflow
of {mapme.vegetation} are described; the figure guides the
reader through these descriptions. The overall goal here is to download
the matching Sentinel-2 tiles a spatio-temporal extent of interest and
process the downloaded data to derive a dense time series of surface
reflectance bands and vegetation indices. On the other end of this
process, there are optional additional analysis steps such as the
calculation of pixel-wise trends or the extraction of zonal
statistics.
- download only relevant bands for a given spatio-temporal extent of
Sentinel-2 L2A data
- prepare a custom cloud mask based on the sen2cor Scene Cover
Classification (SCL)
- calculate indices and extract surface reflectance bands and
aggregate values to a regular sparse timeseries
- apply an interpolation and smoothing strategy to fill in missing
data, reduce noise, and create a dense timeseires
- optionally, calculate pixel-wise trends through the time
dimension
- optionally, extract zonal statistics of areas of interest, e.g. for
diff-in-diff analysis
- optionally, extract pixel time-series, e.g. for later usage with the
{mapme.classification}