When The Lights Go Out
Using Nighttime Satellite Imagery to Survey the Effects of Conflict with the Islamic State in Iraq
If you have kept up at all with news last year regarding the prolonged conflict with the Islamic State (IS) in Iraq and Syria, then you’ve likely seen the devastating images of once large, beautiful cities like Mosul and Aleppo bombed to unrecognizable expanses of rubble. In Iraq alone, more than 2.5 million people have been internally displaced as of 2017, according to the Internal Displacement Monitoring Centre. Millions more sought refuge in nearby Jordan, Europe, or elsewhere.
For a project last year, I was interested in trying to find some new perspectives on the conflict and its effects. This being for a GIS class, I looked at the possibility of risk mapping or outlining migration patterns of refugees, but eventually I settled on seeing if I could use satellite images of light pollution caused by human activity to “capture” the conflict in Iraq.
Admittedly there is a fair degree of obviousness inherent in a study like this — if bombs drop in a given city and knock out power out for a prolonged stretch of time, of course that’s going to be visible with satellite imagery. But, with some luck, perhaps during the project I could catch upon something interesting and worthwhile. Either way I found it a fun, albeit nerdy way to pass a weekend or two. And I felt a little bit like an intelligence officer to boot.
The primary data source was limited by what was available to me as a cheap college student, but I managed to find a cache of high-quality raster files from the Visible Infrared Imaging Radiometer Suite (VIIRS), a cluster of sensors on the Suomi-NPP satellite which are capable of detecting the light of a single streetlamp (I would link to the NOAA page I downloaded the files from, but it’s currently down due to the president’s temper tantrum erm government shutdown). I’m using four monthly composite files, starting from the height of the conflict with IS in 2014, to 2017, when the bulk of the conflict was over (this was also the most recent composite file that was available as of when the study was conducted). Areas of significant losses or gains were cross-checked against Google Maps or OpenStreetMap, in addition to other references.
Just looking at 2014 vs. 2015, you can see a marked difference in the north and west of Iraq, areas that were under IS control. Mosul and Ramadi, both areas of high combat activity, became significantly dim, and a few major transit routes, like the one leading from Mosul on south to Tikrit, have fallen dark. Another route leads faintly from Qa’im, a major border crossing between Iraq and Syria, to cities east. Qa’im was one of the last major towns controlled by IS to be liberated in 2017. In 2015 this route was almost totally dark as well.
If you look closely at the eastern outskirts of Mosul, you can see a line of activity, as if the light from inside the city spread outward and away. Very likely this corresponds to the locations of IDP camps at the time, as it correlates with an infographic I found from the REACH Initiative:
I did run a quick analysis depicting the amount of losses or gains between 2014–2015, but it didn’t appear as conclusive as I had hoped. That may have been due to the way I was rendering the raster files in ArcGIS. I got a better sense of change by graphing by city, e.g. Mosul had a decrease of almost 90 percent.
A significant limitation of the study is determining what changes in light intensity can actually be attributed to conflict as well as what amount of change is significant versus completely normal. For instance, the vertical string of bright lights in southern Iraq near Basra is one of the brightest objects on the map, but a quick look on Google Maps indicates they aren’t towns or cities, but rather industrial or oil production facilities:
Similarly, another huge concentration of light near the Iran/Iraq border immediately west of Baghdad (and also one of the few areas of net gain), shows a desolate area with little if any development on Google Maps, but on OpenStreetMap is listed as the location of a military outpost, the former Combat Outpost Shocker:
The full link to the project, complete with a full write up, can be found here. Also, a quick acknowledgement to Stratfor for this blog post that looks pretty similar to mine, but post-dates it by about a year. (Hey Stratfor, if you happen to need another GIS person on staff, feel free to give me a shout!)
