Deforestation in Rwanda’s Protected Areas | Elise’s Remote Sensing and Land Use Portfolio

^{A visual analysis of tree cover (green represents gain and red represents loss) in Rwanda’s protected areas.}

My analysis of deforestation found that less protected areas in Rwanda experienced more tree cover loss than strictly protected areas. Strictly protected parks lost .0030393% of tree cover in 2020 while less protected areas lost .0031496% of tree cover in the same year. While this is only a fraction of a percent difference, inspecting the border of one of the strictly protected areas in Figure 1 shows that the border creates a distinct divide between areas that experience tree cover loss (outside the park) and no tree cover loss (inside the park). Although the degree of protection was determined by the IUCN category and status of the protected areas, there was still overlap between parks in the strictly and less protected area categories, which is a problem with the protected areas database and a source of error in my results. Parks with IUCN categories “not reported” or “not applicable” were considered less protected, but these designations are simply a lack of data and do not indicate how protected an area is.

^{Figure 1. A strictly protected area in Southeastern Rwanda. Tree cover loss is concentrated just outside the border of the protected area.}

I use the term “tree cover” loss as opposed to “forest cover” loss because Hansen’s designation of forest as simply as tree cover ignores different land uses that could be mistaken for forest. For example, a plantation of trees that is rotationally harvested will have drastic fluctuations in forest cover over the short time period (relative to a forest) of twenty years in this analysis. This is not the case within Rwanda’s protected areas, but Figure 2 is an example of this discrepancy in another area that experienced tree cover loss in 2020 but did not have any tree cover in 2000. Furthermore, the percent tree cover for a pixel to be classified as forest and the percent tree cover loss in a pixel to be considered as forest loss is important to define in a deforestation analysis using the Hansen’s data set. In my analysis, a pixel needed at least 30% tree cover to be forested and at least 30% tree cover had to be lost for the pixel to experience forest loss. These thresholds should be made specific to landscape type, location, and scale in order to provide the best analysis of tree cover loss.

^{Figure 2. These maps show an unprotected area in Southern Rwanda in 2000, 2013, and 2020. Tree cover increases and decreases in these three frames, likely due to tree harvesting.}

https://code.earthengine.google.com/52ffc1525f5c83ab8a1a59fa4a0c4b39