The Use of Explosive Weapons in Videos and Images

Using open-source investigative methods, Airwars has reviewed hundreds of videos and images of explosive weapons use by the Israeli military to expose the impact of airstrikes on civilian populations in Gaza. This article explains the methodology in depth, which is intended to be fully replicable and scalable for other researchers also looking to investigate explosive weapons use in conflicts. This article also presents the findings of the Airwars investigation, where more than 400 civilians were likely killed in just 17 strikes.

Introduction

Since the war in Gaza erupted on 7 October 2023, the Israeli military has posted an unprecedented number of video clips of airstrikes on social media. These videos are typically blurry and greyscale, often with five or more strikes combined into a single video.¹ The clips typically do not provide details on targets and fatalities, and the impact on civilians is omitted.

Over the past year, Airwars – a watchdog for civilian harm from airstrikes – has been rigorously documenting allegations of civilian harm from airstrikes in Gaza. This has resulted in hundreds of incident-by-incident reports of civilians publicly reported killed or injured including, where possible, exact coordinates of the incidents.² 

The release by the Israeli military of hundreds of video clips of their own strikes prompted Airwars to match these two perspectives – of the official footage of the strike as released by Israeli forces from above and the documented human impact on the ground as reported by local citizens.

The investigation revealed the location of more than 70 strikes included in footage published by the Israeli military.³ In 17 of these incidents, Airwars was able to match the footage to the exact geolocation of a documented civilian harm incident. In these 17 strikes alone, Airwars identified more than 400 civilian fatalities.⁴

The interactive map from The Killings They Tweeted, displaying an Airwars incident connected to IDF strike footage. The highlighted strike clip shows the targeting of the Ramlawi family home, allegedly killing four civilians.⁵ Clicking on the ‘full incident report’ takes the user to the specific case in the Airwars civilian casualty archive. © Airwars 

Methodology

Locating IDF explosive weapons use

The process began by identifying and digitally preserving more than 600 clips of strikes released by the Israeli military in the first month of the war on their social media accounts. Each post varies in level of detail, with some posts including specific time periods relevant to the strike (such as ‘during the night’ or ‘in the past day’) or information as to the strike target (such as a tunnel or the individual targeted). Clips also included multiple locations spliced together to make one longer video.

Of the hundreds of videos released by the Israeli military on its own social media platforms, Airwars narrowed down a sample of less than a hundred clips that had sufficiently distinctive visual signals to support a process called ‘geolocation’, whereby user generated content (like an image or video posted on X) is reviewed and matched against satellite imagery to pinpoint the location of a captured event. For example, a visual signal might be a distinctive shape of a building (such as domes on a mosque), that make it more likely to find the matching distinctive shape when looking at satellite imagery (see Figure 1). 

Figure 1Each video clip was assessed and categorised in order of distinctiveness of the visual features through a scoring system. The team would assign a score of one to five to each clip, with five reflecting incidents with unique or noticeable features, and one being videos that contained more generic or blurry footage.

 Documenting civilian harm

As a separate process carried out by Airwars since October 7th, Airwars teams have been documenting hundreds of civilian harm events in Gaza based on open-source claims. This process has been led by Airwars’ specialist casualty recording teams, who have now amassed a public database of hundreds of incidents of harm - including from the first month of the war.⁶ Using a trusted methodology applied across multiple conflicts and honed over the last ten years, this approach also includes a geolocation element, enabling Airwars to pinpoint in time and place where and how civilians have been killed, airstrike by airstrike (see Figure 2).⁷

Figure 2

Matching datasets

With these two databases amassed, the hundreds of categorised clips of airstrikes released by the IDF, and the civilian harm incidents as reported by local voices, Airwars then began a process of reconciliation using the following methods: 

1. If the strike hit civilian infrastructure, the clip was cross checked with the Airwars archive. For example, if a video of a strike on a mosque was published on 17 October, all strikes on mosques that were documented by the casualty recording team on 17 October were reviewed. All mosques listed in internal Airwars research documents were geolocated. Satellite images of the mosques were then compared with IDF footage to see if there was a match. Early in the process, it became clear that the IDF also published footage of strikes that occurred several days prior. The date range was then expanded when cross checking civic buildings with the Airwars archive.

2. Every geolocated incident of civilian harm completed by the Airwars geolocation team was compared with ‘high scoring’ IDF strike footage. For example, if a strike was geolocated to a residential building by Airwars on 17 October, satellite imagery was compared against clips of strikes published by the IDF on, and around, 17 October. This was the most direct method of linking published IDF footage with cases of civilian harm.

3. In other instances, ‘high scoring’ IDF footage was geolocated by manually searching across satellite imagery of Gaza. There was either little, or no, geographical information provided alongside the IDF strike footage. So, by analysing the density of the area pictured in the clip, the investigations team assessed whether the strike was in a rural area or city environment, narrowing down on neighbourhoods of Gaza to parse through. By matching distinct features with satellite imagery, exact locations of IDF strikes were identified.

The investigation uncovered the location of 72 IDF strikes that were posted by official Israeli military accounts on social media. From these, 17 were matched to civilian harm incidents, revealing that at least 448 civilians were allegedly killed from Israeli strikes posted by the IDF online. The total number of reported civilian casualties includes 204 children and 123 women. This means that, at a minimum, 70 percent of civilians reportedly killed in the 17 strikes were women and children.⁸ The remaining 30 percent were not necessarily men, as a large number remain unidentified. 

Visualisation

Having uncovered these findings and established links between the two bodies of evidence, Airwars then produced a short 20-minute film that presents these findings alongside interviews with survivors and relatives of one of the strikes, as well as legal and academic experts to explain the implications of the findings.

An example of geolocation evidence. The left image is a screenshot from an IDF strike clip. The annotation corresponds to the buildings pictured in the image on the right, which is a screenshot from Google Earth. The white box is the al Taj tower, a residential building that is shown to be destroyed in the IDF footage. In this incident, Airwars counted tallied OR identified at least 101 civilians allegedly killed.⁹ © Airwars

 A key tenet of open-source investigations is also to be transparent in how they are undertaken and to encourage others to interrogate these findings and replicate the research. Airwars therefore built a fully interactive map alongside the video, displaying each matched incident (see Figure 3).

Figure 3

Conclusion

This investigation revealed not only the widespread impact on civilians of the use of explosive weapons in populated areas by Israeli forces, but also that it is possible to critically interrogate military narratives of precision when it comes to deployment of such munitions. 

Many of the videos were produced alongside, and to complement, the assertion by the Israeli military that they were taking unprecedented steps to protect civilians.¹⁰ Using the same footage that was intended to support this claim, this investigation reveals a very different reality for those civilians who the IDF were ostensibly trying to protect.