1. Within Google Earth, draw a polygon around the urban area of interest. Once drawn, save the polygon as a KMZ file and import it into SAGA GIS.
2. Import the Petere urban areas dataset shape file. This can be downloaded by clicking here.
3. When both files have been imported into SAGA open the “Cut Shapes Layer” tool.

Geoprocessing -> Shapes -> Construction -> Cut Shapes Layer

Within this window set the options as:

Shapes >>                          ne_10m_urban_areas (Petere urban areas dataset)

Extent >>                          <create>

Method >>                        center

Extent >>                           shapes layer extent

In the following window set the “Shapes” option as the imported KMZ file.

4. Once the Petere dataset has been clipped to the area of intrest, open the “Shapes Buffer” tool.

Geoprocessing -> Shapes -> Tools -> Shapes Buffer

Within this window set the options as:

Shapes >>                          (The clipped Petere dataset)

Buffer Distance >>           <not set>

Default >>                         0.04504505

Buffer >>                           <create>

Leaving the remaining options as default.

5. Open the newly made buffered layer. Create a new rectangular polygon around the buffered zone:

Right click the shape file (Within the file manager) -> Edit -> Add shape

Draw the rectangular polygon by clicking the action button () and drawing around your buffered area of interest. Once drawn, right click the map and untick “Edit Selected Shape” to save the changes.

Right click the newly created polygon and select the “invert selection” option. Right click again and select the “Delete Selection” option, deleting the no longer needed buffered zone.

Save this new polygon as a KML file, named ROI.

Geoprocessing -> File-> GDAL/OGR -> OGR: Export Shapes to KML

This KML file can then be imported into google earth and used as the ROI to generate the training data