Use ML to separate forested and non forested landbase with satellite covariates and bcmaps training points

[whmacken]

Separating the forested landbase from non-forested ecosystems is the first step in building models for identifying broad habitat groups. Most of the analytial tools and information sources required for doing this operation have been presented in the course and can be applied to separate out and map these two broad classes of habitats. Will feed into PEM methods research methods and wetland inventory methods.

[thengl]

Are you interested in land cover classification from raw Sentinel-2 imagery or you plan to use existing land cover for this? There are at least 3 land cover maps for BC and several international LC products (https://lcviewer.vito.be/). Using ICESat-2 data could also help probably.

[boshek]

Work happening here: https://github.com/bcgov-datasci-labs/BCForestNonForestPEM