Tree Cover Loss

This indicator uses satellite data from the Hansen/UMD/Google/USGS/NASA Global Forest Change dataset to track the loss of vegetation five metres or taller in areas with at least 30% tree cover density. As such, it captures the loss of forests, open woodlands, trees in agricultural settings and urban areas, and small patches of trees. Tree cover loss is defined as the disturbance of a forest patch or the complete removal of tree cover canopy at the pixel scale. A decision tree model is used to identify the main drivers of tree cover loss, including permanent agriculture, logging, shifting cultivation, wildfire, hard commodities, settlements and infrastructure, and other natural disturbances. The dataset does not indicate the stability or condition of land cover after tree cover loss, nor does it distinguish between natural and anthropogenic wildfires.

1. Exclusion of Certain Disturbances: The model does not include disturbances such as insect outbreaks, wind and ice storms, flooding, or rivers changing course. These disturbances were found to be highly localized and temporally restricted, affecting only 1% of all model validation sample cells​​.

2. Misclassification Issues: There was low model accuracy for the commodity-driven deforestation class in Africa, with much of this deforestation misclassified as shifting agriculture. In northern forests, especially in Russia, distinguishing between drivers was challenging in areas where wildfires spread through previously logged areas or where logging occurred after a fire event​​.

3. Lack of Detailed Differentiation: The study did not map changes in forest conditions over time in landscapes dominated by shifting agriculture, nor did it differentiate primary from secondary forest clearing within this land-use class. Differentiating key drivers like row crops from pasturelands in South America or tree plantations from disturbed natural forests in Southeast Asia could enhance the analysis​​.