Mode and Statistics

We provide three different general modes of operation:

  1. Download - Mode
  2. The most basic one would be the Download - Mode that simply downloads all raw satellite files. While it is a rather fast process the storage demands in this mode are massive!

  3. Composite - Mode
  4. The Composite - Mode merges the image stacks in the temporal space based on one or more statistics the user defines. To define your metrics, fill the statistics field accordingly, excluding any prefixes ( or parenthesis and separated with commas (e.g. median, std). In the graphical user interface, a multiple choice field makes the selection easier.

  5. Classification - Mode
  6. With the Classification - Mode the image composition is extended by a classification. Classifications will be carried out for each satellite independently but a merged combination result will be provided in case more than one satellite was selected.

Useful Tip:

Please refer to the Process Overview in the Introduction for detailed information on the general workflow!


Select the satellites you want to be considered for your run. Landsat is a shortcut for Landsat 5, 7 and 8, while All refers to all possible satellites. Double selections (e.g. selecting Landsat 7 and also All) will be reduced to exactly one.

Useful Tip:

Please refer to the Satellites in the Introduction for detailed information on the considered satellites!


Start and end dates can be selected manually by writing dates in format or by picking with the provided widget. The dates and their resulting interval must fulfill the following criteria:

  1. The end date must be after the start date.
  2. The end date must not be within the last three days or in the future.
  3. The end date must not be before 01/01/1984.
  4. The date interval must be less than a year.

Boundary File

The boundary file is a vector file that is already stored with your account, uploaded via the File Management Tool.

Important Note:

The boundary file must be a GeoJSON vector file! The unary union of all contained shapes is considered the boundary.

Sample Options

In case of a classification job, you have three choices for your input training dataset. The default is the MODIS Land Cover Classification (MCD12Q1). For downloading MCD12Q1 a NASA LAADS Key is required. If you want to bring your own dataset, select Samples and define the respective file stored with your account. There are two allowed kinds of input data:

  1. Point Data
  2. A CSV file containing integer class values per year and X/Y coordinates in the EPSG:4326 reference system (WGS:84). The columns are named per year ("yyyy"), "X" (for long) and "Y" (for lat). The year closest to the time interval will be considered. Only columns containing just digits are considered year-class columns.

    X Y 2018 2017 2016
    8.22 46.34 14 14 14
    7.54 46.95 15 15 14
    7.93 48.21 10 9 18


    Using the freely available QGIS Application, it is easy to convert other sample datasets to points.
    To convert points in different file types refer to link but select GEOMETRY: AS_XY.
    To convert polygons refer to link.

  3. Raster Data
  4. A TIF raster file containing three bands: (1) Class [int], (2) Probability [0-100] and (3) Cloud mask [0 = Cloud / 1 = No Cloud].


It is possible to change some of the default values. Define a different EPSG Code for the output coordinate reference system or the maximal cloud cover for a single satellite image to be considered in the process. Note that a lower threshold can improve speed in the merging process but does not prevent any previous processing steps for respective images. For the download mode these parameters have no effect, since there is no further processing step.

Useful Tip:

We strongly advise to attach a short but meaningful description string to every run for later reference!