i’d imagine it’s a mix of vetted scientific data (esp. for areas of specific interest like the Amazon), together with some automagical data-mining algorithms applied to satellite imagery. especially with longitudinal data, i imagine this could work reasonably well.
and yeah, that’s pretty much it. very interesting details here: http://www.globalforestwatch.org/sources
for example: “This base map depicts the highest points in the forest canopy. Its spatial resolution is 0.6 miles (1 km) and was validated against data from a network of nearly 70 ground sites around the world… The map was created using 2.5 million carefully screened, globally distributed laser pulse measurements from space…” cool!
