In this data set, “tree cover” is defined as woody vegetation with the height of 5 m and taller, and may take the form of natural woodlands, forests, or tree plantations across a range of canopy densities. Tree cover gain does not equate directly to restoration, afforestation or reforestation.

Due to variation in research methodology and date of content, tree cover, gain, and annual loss data sets cannot be compared accurately against each other. Accordingly, “net” cannot be calculated by subtracting figures for tree cover gain from the annual tree cover loss data set. Instead, the net tree cover change layer should be used, which was calculated exclusively from tree height data.

Integrated use of other products such as canopy cover density data also available on GFW should be performed with caution.

The authors evaluated the accuracy of the product, and the overall accuracy was found to be 99.3%, commission error (false positives) of 28.6%, and omission error (false negatives) of 42.2%. The accuracy does vary by biome and thus may be higher or lower in any particular location. As the omission error is higher than the commission error, this indicates that the product provides conservative estimates of forest dynamics.

There was confusion between forest enhancement (existing forest height increase) and forest gain (establishment of forests within the year 2000 non-forest land), and this was more prominent in boreal forest areas where forest height in the year 2000 was difficult to determine.

Tree Cover Gain

Tree Cover Loss - reference layer