Portfolio Problem IV

December 08, 2020

Is there a relationship between tree cover area and distance from human settlement?

In this course, we have discussed two possible relationships between settlements and tree cover. One states that there is an inverse relationship between proximity to human settlement due to a need to clear agricultural land and gather timber and firewood. The other finds a positive relationship between tree cover and proximity to settlements as citizens sometimes purposefully enhance tree cover (Fairhead and Leach, 1996), often in rural agricultural communities (Brandt et al., 2018). In this problem, we attempt to analyze the relationship between total area of tree cover and distance from developed areas in Malawi.

Compare tree cover to developed areas

We compared a tree cover layer derived from Sentinel-2 data from the year 2019 (± 1 year) to a series of points marking developed or built-up areas. To determine the relationship between distance from development and area of tree cover, we created a series of buffers around each of the developed points, with each buffer representing a distinct distance interval from the point of origin (Fig. 1). The buffer intervals were 0-300m, 300-500m, 500-1,000m, 1,500-2,000m, 2,000-5,000m, and 5,000m-10,000m.

Fig. 1. This image illustrates the different buffer classes. The furthest left image shows all buffers, which cover a distance from 0 – 10,000m away from the point classified as “built.” The center image shows the 0-300m buffer, the 500 – 1,000m buffer, and the 1,500 – 2,000m buffer. The rightmost image shows the 5,000 – 10,000m buffer. Link to Code

We then summarized the total pixel area within each buffer class. Our results indicate that tree cover substantially increases with distance from developed areas (Fig. 2).

Fig. 2. Results for total tree pixel area in each of the buffer increments. X-axis values correspond to the buffer increments, with 0 representing the smallest (0-300m) and 6 the largest (5,000 – 10,000m). Clearly, the largest buffer increments contains the most tree cover pixels. Link to Code

Interpreting our results using the literature

In their analysis of deforestation in the Mwazisi district of Malawi, Ngwira and Watanabe (2019) sought to determine the underlying drivers of deforestation associated with the primary causes of forest loss: agricultural expansion, tobacco growing and curing, and brick burning.  A number of demographic factors, including poverty, rapid population expansion, lack of awareness, and lack of resources, were found to be driving land use changes that lead to decreased tree cover area. Kamanga et al. (2009) found that in the Chiradzulu district of Malawi, access to forests led to higher income among rural residents, with fuelwood bringing in the most money to poor- and medium-income households. This indicates that substantial income is an essential livelihood in rural areas with access to forest resources.

Ngwira and Watanabe (2019) also note that land management strategies in Malawi also contribute to tree cover loss. Customary lands, which account for approximately 85 percent of Malawi’s total area, are managed by rural communities whose strategies may differ substantially from one another. A low level of awareness regarding the conservation and management of forest resources was found among rural residents utilizing forest resources on customary lands.

It is important to note that many of these buffers, especially those measuring 5,000- 10,000 m from the development point, overlap with tree cover pixels that are in strictly protected areas (Fig. 3).

Fig. 3. Built points (grey), buffers (red), and tree cover pixels (dark green) compared to strictly protected areas (black outline). Many buffers overlap with tree cover pixels in these protected areas where trees are, in principle, mostly inaccessible for citizen use. Link to Code
Fig. 4. Land inside a protected area boundary (left side of black line) and land outside the boundary (right side of black line).

In order to account for tree cover pixels in these areas being, in principle, unavailable (Fig. 4), we should eliminate them from our analysis. We could do this by mapping strictly protected areas (as per Portfolio I), creating a geometry feature of Malawi exclusive of these strictly protected areas, and then running our buffer analysis on only the pixels that fall within this new geometry. Comparison with the original analysis would tell us if our interpretation of the relationship between developed areas and tree cover in Malawi holds true. We should also convert our tree cover analysis to a percentage of cover within each buffer to normalize the data between the different buffer areas.

References:

Brandt, M., Rasmussen, K., Hiernaux, P., Herrmann, S., Tucker, C., Tong, X., Tian, F., Mertz, O., Kergoat, L., Mbow, C., David, J., Melocik, K.A., Dendoncker, M., Vincke, C., and Fensholt., R. (2018). Reduction of tree cover in West African woodlands and promotion in semi-arid farmlands. Nature Geoscience, 11:328-333. https://doi.org/10.1038/s41561-018-0092-x

Fairhead, James & Leach, Melissa. (2000). Misreading the African Landscape; Society and Ecology in a Forest-Savanna Mosaic. Science. 71. 10.2307/3034551

Kamanga, P., Vedeld, P., and Sjaastad, E. (2009). Forest incomes and rural livelihoods in Chiradzulu District, Malawi. Ecological Economies, 68(3): 613-624. https://doi.org/10.1016/j.ecolecon.2008.08.018

Ngwira, S., and Watanabe, T., (2019). An analysis of the causes of deforestation in Malawi: a case of Mwazisi. Land, 8(3):480. https://doi.org/10.3390/land8030048