Remote Sensing and Mass Graves Detection 101
When you use the term “mass graves” a couple of things tend to spring to mind. First off, you think about Nazi atrocities during the Second World War, and then secondly, if you are someone who is interested in the world of remote sensing (RS) you tend to think about Srebrenica.
In 1995, the US State Department released a series of rather grainy RS images, which purported to identify suspected mass grave sites outside the beleaguered city in Bosnia. The resulting public outcry is often credited with having spurred the international community into action. The Srebrenica case proved the utility of RS imagery for such purposes, and as a result, the connection between RS technology and mass grave identification became a kind of “holy grail” ideal. Significantly, the concept of the mass grave as undeniable evidence of a war crime also gained headway, leading to a sometimes problematic push to identify sites without necessarily analyzing context. After all, you need to remember that a mass grave is only evidence of a mass burial — it does not immediately identify who is buried there, how they died, or what reasons there were for mass inhumation. Mass burials have been a standard practice for thousands of years, as a response to major death events and fears of disease, sanitation concerns, or even simply as a psychological need to remove the bodies of the fallen from the streets. In other words, caution is warranted when identifying such a site, if only to ensure that what you have found is not subsequently tampered with by well intentioned parties or other, more nefarious actors.
Having said all that, clearly RS tech does provide an exciting opportunity to try and identify such locations. Ever since the early 1970s, researchers have been using terrestrial sensors such as Ground Penetrating Radar (GPR) and thermal imagery to identify suspected crime scenes. While this is not the kind of RS that we often think of, the accuracy and utility of these ground based systems is self-evident. The problem is, you already need to have some idea of where to look, in order for the sensors to be useful.
The best way to identify new potential mass grave sites is to develop an MO for the perpetrators in question. Most war criminals follow specific formulas for choosing their grave sites. This allows researchers to develop crisis specific criteria that are then used for the purposes of GIS/potential site analysis. As one leading authority puts it:
[P]reliminary findings suggest that graves are typically located 1–10 km from the place where the victims were detained (the point of origin); that the graves are almost always within 100 m of a principal road; that these locations are usually not visible from the point of origin; and that pre-existing features such as wells, mines, and ravines are regularly utilized. Through such pattern analysis, a greater understanding of the factors that influence the location of these types of graves could result in more effective searches.
The International Commission on Missing Persons (ICMP) is one of the main authorities on the use of such techniques for the discovery of mass graves, given their substantial experience working in the field in the former Yugoslavia and elsewhere.
The next step up in RS detection consists of sub-orbital sensor systems, such as airborne LiDAR and HyperSpectral Imagery (HSI). Traditionally, this has been done with huge, expensive sensor arrays strapped to small aircraft, and as a result has had only limited impact in the field of mass grave identification. However, recent developments in regards to sensors and aerial platforms suggest that this is about to change considerably.
For a start, the exponential rise in the UAV/Drone arena means that an ever increasing number of groups are gaining access to high resolution imagery, without the need for major funding and/or having to fight for airtime with multiple other projects. Secondly, in 2017 it is possible to purchase thermal, LiDAR, and HSI sensors that can be attached to small-to-medium scale drone systems, meaning that researchers are able to obtain high resolution data sets quickly while on location, and at relatively low cost. This would allow for multiple sites to be processed during field deployment, and with the added bonus of not having to disturb the site physically, always supposing you have permission to fly over the site in the first place.
LiDAR systems are designed to give spectacularly accurate topographic models. These are useful if you are looking to identify ground subsidence due to the grave site “collapsing” in on itself, or otherwise anomalous identifiers that might indicate the disturbance of a particular area. Small scale LiDAR sensors (or “pucks”) for use with UAVs are relatively cheap (anywhere from US$7,000–10,000 and upwards). Thermal imagery has been used since the 70s in criminal investigations in order to try and locate the presence of bodies, although there tends to be a limited window of opportunity — ostensibly about 2 weeks — before thermal indicators begin to fade drastically, depending on the local geophysical conditions. This might appear to be a major obstacle for human rights based investigations, but on the other hand, both military and civilian investigators following the current push back against ISIL in Iraq have often been on-site within days of a massacre allegedly taking place. Theoretically, thermal scanning of recently liberated regions could help to identify potential grave sites. Drone mountable thermal cameras can be had for as little as US$1,000.
HS imagery stands out as a real potential game-changer. A number of recent academic papers have successfully explored the use of HSI/Spectral Resonance for grave detection using aircraft based data sets. HyperSpectral systems capture data into hundreds of narrow “bands” within the electromagnetic spectrum. As a result, they are extremely useful for identifying the precise spectral signatures of individual compounds and elements within the soil that might indicate the presence of bodies, the anomalous growth of vegetation, or even soil composition changes due to the disturbing of the earth. A paper published in 2014 was able to successfully identify the locations of two single “proxy” grave sites in this manner, suggesting that the scope for further research with this technology is potentially very significant. On a positive note, research groups at McGill University, the University of Tennessee, and the University of Arkansas are all actively engaged in this area, which bodes well for those of us interested in taking this technology into the field as soon as possible. Drone mountable HSI sensors are expensive (in the roughly US$50,000 to 70,000 range), but are clearly going to play an important role going forward.
The third tier for RS use is orbital platforms. Of course, Very High Resolution (VHR) images are something of a staple in the human rights world, and are often held up as the poster child for identifying (and publicizing) rights violations from afar. However, mass grave identification from RGB, “true color approximation” images or panchromatic shots (such as the ones of Srebrenica) requires the subjective analysis of a trained analyst in order to identify potential grave sites. This can be problematic for a variety of reasons, not the least of which is the need for ground verification to ensure that what you have marked out as a mass grave is not simply ground clearance or construction for some other purpose. This is often when understanding the MO for mass burials in the specific region in question is essential. Very few studies appear to have used orbital sensors, although a couple of them have used VHR imagery as secondary data sets for already identified locations. It should also be noted that there are a number of orbital thermal and HSI sensors, but these lack sufficient ground resolution to be practically useful for grave detection at present.
On the plus side, one area that might prove interesting going forward is the use of orbital Synthetic Aperture Radar (SAR) interferometry. Using radar systems has a couple of advantages — they use active sensors (meaning you both send and receive electromagnetic energy) which allows you to be very specific in what you are looking for, and they work at night-time and despite the presence of cloud cover. Interferometry allows you to compare before and after shots of a particular location, and can identify even relatively slight changes on the ground. A research team at NASA/JPL working on humanitarian applications for this technology has been able to chart landslides, building collapses, and even identify individual house extensions in urban environments, suggesting that there is potential here for use in identifying gravesites. As far as I am aware, no one has attempted to use SAR-I in this manner, perhaps due to the increased complexity and precision necessary for effective interferometry work, but it could be an interesting avenue to explore in the future.
Ultimately, RS in all its many guises has a distinct advantage in the search for missing bodies — namely, that it is able to look for evidence of mass inhumation without having to disturb any potential site in question. This is significant for both legal and ethical reasons, and should prompt greater interest in the development of practical techniques from a variety of stakeholders (academic, activist, legal, and governmental). Another notable advantage is the RS data access over inaccessible regions of the world. As the technology develops making access to sensors and other essential platforms cheaper, easier to acquire, and more ubiquitous, human rights advocates of all stripes will start to adopt these systems and monitor the nearly daily changes in the environment without the need for ground access. Unfortunately for us all, the world as it stands will continue to provide no lack of opportunities to refine our techniques. The fact that we potentially stand at the dawn of a golden age of mass grave detection is, in the last instance, rather bitter-sweet.