Following up from Exercise 2, where I determined how NDVI loss relates to distance from village center, I now wanted to explore whether or not the NDVI loss I observed could be explained or attributed to another factor: the distance of a village from a primary road.

To do this, I acquired a shapefile of roads from OpenStreetMaps for Senegal, which I used alongside my locations of mining and non-mining villages, ArcGIS Pro for some simple spatial work, and Excel to calculate a simple correlation.

First, I imported the shapefile of roads in Senegal to ArcGIS Pro, and first determined what the nearest actual road to each village was (fig. 1). Based on this first-pass glance, it appears the more mining sites in my sample than non-mining sites are nearest to “primary” paved roads, and that the majority of non-mining villages are near “tertiary” roads. This simply means that, out of my sample, more mining happens closer to paved roads, which would make sense: gold needs to be transported from gold sites, and equipment/supplies need to be transported in.

Fig 1

Next, I manipulated the shapefile to only include features which were designated “primary” roads (fig. 2). I then used ArcGIS Pro’s “Near” tool to generate new attributes for each set of villages which listed the shortest distance to the nearest primary road in meters.

Fig 2: Green dots are non-mining villages; red dots are mining villages

With these distances in hand, I plotted them in excel against the NDVI % loss values for the first ring (0-250m) around each village, as a simple first pass of understanding how distance from a primary road may be related to the amount of NDVI loss being experienced at a particular road (fig. 3).

Fig 3

Then, within Excel, I calculated a simple correlation (a la fig. 4) to determine how distance from a nearest primary road relates to amount of NDVI loss experienced by a village.

Fig 4: x, in this case, is distance (m) from nearest primary road and y is NDVI loss at the first ring.

For mining villages, I calculated a correlation of -.804, and for non-mining villages, I calculated a correlation of .686. This means that, for this sample at least, the NDVI loss experienced at a mining village is negatively correlated with distance from a primary road; that is, as a mining village’s distance from a primary road increases, its NDVI loss should decrease. This makes perfect sense: mining villages closer to primary roads are easier to access both by miners and by other users, and to leave as well, encouraging larger groups of people to live there, which would decrease NDVI as a result of their activities.

Non-mining villages, on the other hand, are more difficult to explain. One explanation, which may be plausible, is that more distant villages have less opportunities for other livelihoods which are not environmentally dependent, and so are more obligated to work in and with the environment, perhaps encouraging vegetation loss as users rely on and drawdown more on their environment.

Overall, I think this was a useful approach for examining another possible explanation for why mining villages may have higher rates of NDVI loss than non-mining villages. Mining villages are closer to primary roads, which means they are more accessible to potential miners, and thus more likely to grow larger than non-mining villages, and drawdown more on their environment. That being said, as can be seen in Fig. 3, this is not a perfect relationship by any means, and a more robust analysis would be useful to say much more. But, as a first pass, it’s useful for illustrating some additional explanation.