In Rwanda, most forests are located far away from urban settlements.
By far the largest shares of forest cover were located farther than 2km from urban points, with the most forests found in the 5km to 10km range. This is partially because these zones were the largest in the analysis. However, even when zooming in on the rings within 2km of urban settlement, there is a clear positive relationship between distance from urban points and the amount of forest in the region.
This makes since given the existing literature on land use in Rwanda. As Rukundo et al. note, multiple forces in Rwanda, including population pressure, a lack of employment opportunities, and farm fragmentation have led forests in the region to be converted not into to settlement, but agricultural land (2018). This expanding agricultural land creates a buffer between settlements and forests as demonstrated in Figure 3. In addition, the reliance on wood and charcoal as fuel in the region, coupled with the aforementioned population pressures, has led an increase in demand for wood and a strain on resources (Akinyemi 2017). This factor has caused deforestation, pushing forests away from settlements as people travel from settlements to the edges of forests to collect wood.
One way this analysis could be improved would by systematizing the way the urban areas are chosen, rather than selecting them by hand. In order to do this, the following steps could be implemented:
- Export a Sentinel-2 built-up/urban layer from a previous analysis and import this into the new script, making sure the new layer could adequately distinguish urban and barren land.
- Convert this image into a vector layer using image.reduceToVectors(), being sure to set the geometryType to ‘polygon.’
- Proceed as before, creating and examining buffers around the urban polygon layer rather than urban points.
This adjustment would be beneficial for a few reasons. To start, it creates clear criteria for which settlements are included in the analysis. Due to my lack of knowledge of urban patterns in East Africa, there may be important or unusual settlements in the study area that I missed when collecting urban points. In addition, it makes more sense to treat urban areas as polygons rather than points in this analysis. A point near the center of a larger city like Goma or Kigali could easily be more than a kilometer from the edge of the settlement, which skews the data away from showing forest cover near settlements. Thus, it makes more sense to assess forest proximity to the edge of a settlement rather than a point or points within the settlement. This would also allow analysis of the effects of settlement size on proximity to forests by separating settlement polygons into two classes based on area and creating different charts for the two classes.
Akinyemi, F. O. (2017). Land change in the central Albertine rift: Insights from analysis and mapping of land use-land cover change in north-western Rwanda. Applied Geography, 87, 127–138. https://doi-org.ezproxy.middlebury.edu/10.1016/j.apgeog.2017.07.016
Rukundo, E., Liu, S., Dong, Y., Rutebuka, E., Asamoah, E. F., Xu, J., & Wu, X. (2018). Spatio-temporal dynamics of critical ecosystem services in response to agricultural expansion in Rwanda, East Africa. Ecological Indicators, 89, 696–705. https://doi-org.ezproxy.middlebury.edu/10.1016/j.ecolind.2018.02.032