Symposium on Forensic and Counter-Forensic Approaches to International Law: Looking Down and Looking Away – The Aerial Perspective in Investigations of Armed Conflict and Genocide

[Christiane Wilke is a Professor in the Department of Law and Legal Studies at Carleton University, Ottawa and collaborates with Airwars on a project examining legal and factual claims in US military civilian casualty assessments]

How do we look at genocide, and how does the vantage point shape what we see? Armed conflicts and genocides are frequently represented using the aerial perspective: satellite images, drone video footage, and maps showing patterns of displacement and destruction. This essay probes the politics and limits of the aerial perspective for representing violence by juxtaposing the work of Forensic Architecture with the US military’s assessments of civilian casualty allegations.

The aerial perspective is powerful and seductive, but it also has important limitations. First, bird’s eye view is unfamiliar to human eyes: we are not used to seeing from above. As a consequence, when we look at images taken by satellites or drones, we often rely on explanations contained in captions, annotations, and arrows superimposed on images. Second, the view from above is associated with abstraction, power, and dehumanization. The drone does not see faces. Palestinian writer Mohammed Al Kurd cautioned: “When I speak of dehumanization, I am referring to the West’s refusal to look us in the eye.” This statement challenges us to think about whom and what we see when we look at violence. Is the aerial perspective that dominates the visual evidence of genocide and aerial warfare automatically dehumanizing? Or can the view from above be used to see otherwise, to be accountable to the victims on the ground? Forensic Architecture has consistently scrutinized and challenged the use of the aerial perspective by the Israeli military, as documented in the Cartography of Genocide report, their assessment of visual evidence presented by the Israeli legal team at the International Court of Justice, and previous work.

Seeing Like a Military: The Civilian Casualty Files

A juxtaposition of the use of the view from above by militaries with the work of civil society organizations allows us to understand under which circumstances the view from above has dehumanizing effects. The “civilian casualty files” obtained by New York Times journalist Azmat Khan that document US military assessments of civilian casualty allegations arising from Operation Inherent Resolve in Syria and Iraq in 2014-2018 allow us to understand how the US military interprets their own aerial surveillance footage. Of 1,311 allegations of civilian harm represented in the files, only 213 were found “credible,” often due to alleged “mistakes” in targeting. The vast majority of the 1,091 allegations deemed “non-credible” were based on ground-level testimony or visual evidence, often forwarded by Airwars. As explored in a recently published book chapter I co-authored, those accounts were often rejected because the US military was unable to match the allegation with a specific strike in the military’s own databases (which were known to be unreliable and incomplete). The civilian casualty files illustrate the US military’s reliance on the aerial perspective and its effects on the ability to see violence and civilians.

The US military recognizes civilian harm largely on the basis of aerial visual evidence (most of which had been produced by US and Coalition equipment). In a sample of 53 cases in which the US military had conceded civilian harm, 35 had only video evidence from the aerial perspective, 2 only from the ground, and 5 had visual evidence from ground as well as aerial perspectives.

In a number of attacks discussed in the files, a person walked into the target area in the time between the weapon release and impact. These persons – almost always thought to be men – were reluctantly but consistently identified as civilians. For example, one file assesses an incident in which “the weapon struck the moving truck just as the vehicle passed by a transient person” and determines: “… all personnel are assumed to be non-combatants unless they are positively identified as valid military targets. Based upon the available information, CJFLCC-OIRTAC Erbil cannot positively identify the uncharacterized, transient person engaged by this strike as a valid military target. Therefore, it is more likely than not that second engagement resulted in the death of a non-combatant.”

Here, the civilian remains an “uncharacterized” person, recognized as a civilian only because they had not been “positively identified” as a legitimate target. Law shapes who and what is seen: International Humanitarian Law establishes visual cues for spotting combatants (weapons, uniforms) and mandates the presumption of civilian status “in case of doubt” (AP I, Art. 50(1)). The US military’s internal Rules of Engagement specify the process for the “positive identification” tracking of persons who might be legitimate targets. Recognizing a person as a civilian or combatant requires the use of visual capabilities as well as legal rules and categories.

The files show that even when the assessors interpret the pixels on their screens as persons, they struggle to fully humanize them in their narration:

“FMV [full motion video] was recorded by [redacted], which observed the kinetic engagement and up to seventeen (17) minutes preceding munition impact. At approximately [redacted] one (1) ADM [adult male] is observed entering the DFP [defensive fighting position], accompanied by one (1) person of unknown age and gender who was substantially shorter than the ADM [adult male], just over 1/2 the height of the ADM [adult male], and was consistent with how a child would appear standing next to an adult. Munitions are detonated at approximately [redacted], and the DFP [defensive fighting position] was assessed as destroyed. FMV [full motion video] coverage continues for an additional one (1) minute and forty (40) seconds following the strike. No personnel are observed in vicinity of the DFP [defensive fighting position] following the strike. Through the available FMV [full motion video], the age and gender of the second person cannot be determined.” 

From the aerial perspective, humans appear as collections of pixels, their gender and age indicated by the shape of their clothing and the length of their shadows. Their deaths are obscured in the observation that no activity was observed after the strike. In most cases where the US military concedes civilian harm, the civilian victims remain anonymous. Out of a sample of 53 incidents, the US military knew the names of some or all victims in ten cases, but only one victim (on whose behalf New York Times journalists inquired and advocated) was offered any compensation. In short, while the recognition of unknown persons (usually adult men) as civilians had become routine in these assessments, it appeared schematic and limited. The aerial perspective allowed assessors to see civilians as abstract pixelated persons they would not feel accountable to. Used this way, the aerial perspective avoids facing the victims of the violence it documents.

Re-Humanizing from Above? Forensic Architecture’s Work

Two aspects of Forensic Architecture’s deep engagement with the aerial perspective stand out. First, the team re-views visual evidence submitted by states and militaries to challenge the findings, thereby demonstrating that visual evidence is often indeterminate. The meaning of photos and videos of atrocities is established through interpretation and narration . Second, Forensic Architecture often matches visual evidence from the aerial perspective with visual and testimonial evidence from the ground level. Viewed together, these two dimensions of Forensic Architecture’s work can help us understand which forms of dehumanization are inherent in the aerial perspective and which forms of refusing to “look [the victims] in the eye,” to slightly modify Mohammed El-Kurd’s phrase, have a different source.

In January 2024, the Israeli legal team relied on visual evidence in a hearing of the International Court of Justice, trying to rebut the accusation brought by South Africa that Israel was committing genocide against Palestinians in the Gaza Strip. All the photos they used (including video stills, satellite photos, and footage from drone cameras) feature the aerial perspective. The images look clinical, detached, objective. They are annotated with words and lines superimposed on the photos. The Israeli legal team used the visuals to argue that Palestinian fighters had embedded themselves in schools, hospitals, and residential buildings, providing legitimation for the relentless attacks on civilian infrastructure in Gaza. Only weeks later, Forensic Architecture published a report that re-views, re-contextualizes, and re-interprets visual evidence, illustrating the power and limits of visuals materials using the aerial perspective. In the hands of the Israeli military, the images were evidence of Palestinian violations of International Humanitarian Law. In the hands of Forensic Architecture, they became evidence of misrepresentation of visual evidence through misleading interpretation, contextualization, and annotation.

Two examples can illustrate how Forensic Architecture re-contextualized and re-interpreted the images. One photo showed an armed man (annotated as “terrorist with an RPG launcher”) walking towards a building. A white line superimposed on the image marks the alleged boundary of Al-Quds hospital. This visual argument was used to justify Israeli attacks on this and many other hospitals in Gaza. Using both the aerial perspective and triangulation with ground-level evidence, Forensic Architecture shows that the photo misrepresented the hospital boundaries. The aerial perspective showed that the roofs of hospital buildings were marked with Red Crescents, but this building had no such marker. Ground level footage from the area revealed the presence of a shop sign advertising desserts, not a hospital entrance.

In other cases, the Israeli legal team misattributed traces of destruction visible from above. In one example, Forensic Architecture found that the Israeli legal team had been “labelling a crater – a trace of Israeli-inflicted destruction – as a Palestinian rocket launch site.” The mislabelling of existing destruction was used as a pretext for further violence. It also illustrates that while traces of destruction might be clearly visible from above, the question of causation requires careful analysis. A picture might be worth a thousand words, but it might also need a thousand words to be adequately explained and contextualized.

The Forensic Architecture analysts complement the aerial view of violence with visual and testimonial evidence from people on the ground. This contrasts with the US military’s studied rejection of ground-level visual and testimonial evidence. The Cartography of Genocide report illustrates the power of combining both perspectives. The online platform contains a map of Gaza that the viewer can populate with different layers of visualization: destruction of civilian infrastructure, density of urban destruction, destruction of agricultural and water resources, and more. The incident database allows the user to locate an incident on a map that illustrates these larger patterns as well as to see photos of what the destruction looked like on the ground. The clinical list of incidents becomes devastating when we can see the photos of people looking in horror at a mosque in flames, university buildings demolished, and multi-story homes flattened to the ground. Forensic Architecture’s use of the aerial perspective allows for an overview, for the representation of patterns of destruction. Yet the ground level perspective immerses the viewer in the destruction, pain, and uncertainty at a human level. Grasping the enormity of the genocide in Gaza might be impossible, but Forensic Architecture is giving us important tools to try. Used by Forensic Architecture, the aerial perspective is animated by the desire to show the vast scale of destruction, not to render human pain into abstract collections of pixels.

Challenging the View from Above

The aerial perspective can be powerful and seductive, but it can also be disorienting and alienating. The US military’s narration of its own aerial surveillance footage shows that the aerial perspective is compatible with seeing but not fully recognizing civilian victims on the ground. Forensic Architecture’s engagement with Israeli visual evidence reminds us that far from being “objective,” aerial surveillance footage is easily put in the service of an ideology that degrades, discredits, and refuses to face its victims.

Is the aerial perspective inherently dehumanizing? The view from above makes it harder to see individual people on the ground, to recognize their humanity, and to imagine what the war looks like from their perspective. Yet, as the work of Forensic Architecture shows, it is possible to read aerial images otherwise and to stitch them together with ground-level perspectives. The aerial perspective doesn’t cause the dehumanization of the victims on the ground. Yet it can become a convenient vehicle for those who demonize or disregard the victims of violence, their testimony, and their perspectives.