Training Data

Training data can be created in a variety of ways, but the algorithm expects a Feature Collection of points with unique land cover labels identified with a ‘label’ attribute. For example, forests can have a ‘label’ attribute of 1, agriculture as 2, and so on. One simple method to develop training data is to simply use the data on the Google Earth Engine and creating a Feature Collection for each land cover.

1. While in the Earth Engine, navigate to Geometry Imports -> + new layer.
2. In the layer configuration, given the layer a name, set the type to FeatureCollection, and give it a ‘label’ property with a unique integer identifier.

3. Using the data available on the Earth Engine, such as Landsat, Sentinel-2, and the high resolution background imagery, add training points that correspond to a certain time period. For example, all of the training points could correspond to the land cover from 2012-2014. See the official Earth Engine tutorials for information on finding and displaying Image Collections.

4. Create training data for all land covers in the study region. The land covers are necessary to differentiate between forest conversion and a disturbance that does not result in a change in land cover.
5. The training features can be combined with the ‘merge’ method. For example, assuming ‘forest’, ‘pasture’, ‘agriculture’, ‘urban’, and ‘water’ all represent Feature Collections:

var training = forest.merge(pasture)
                     .merge(agriculture)
                     .merge(urban)
                     .merge(water)

6. The outputs should then be saved as an Earth Engine asset.

Export.table.toAsset({
   collection: training,
   description: 'sample',
   assetId: 'sample'
})