Forensic Architecture: how to investigate human rights violations using spatial analysis
Written by Sofia Craveiro, Gerador
Using 3D spatial modelling, satellite image analysis, the organisation of multimedia archives, and the collection of community testimonies, Forensic Architecture reconstructs events and reveals evidence that has been presented before international courts and used by communities seeking justice for human rights violations.
This interdisciplinary collective, based at Goldsmiths, University of London, brings together architects, programmers, filmmakers, lawyers, journalists, artists and specialists from multiple fields who join forces to analyse dramatic events in detail. Together, they produce complex investigations drawing on three‑dimensional technology, audiovisual materials and testimonies from people who survived violent attacks or high‑impact incidents.
The scope of their investigations is global. Many times, they work on situations challenging official findings or contradict established narratives. Among the cases that demonstrate the breadth of their work are the analysis of the development of the Grenfell Tower fire in London, carried out through 3D modelling and situated accounts, the reconstruction of the massacre of humanitarian workers in Tel al‑Sultan, Gaza, combining satellite imagery with survivor testimonies, and the project A Cartography of Genocide, which documents Israeli military actions in Gaza after 7 October 2023.
Natalia Sliwinska, filmmaker, multidisciplinary researcher and video editor of the collective, explained to GeoJo the challenges inherent to developing each investigation and how spatial analysis can be decisive in reaching robust conclusions.
At FA, you basically recreate environments, in order to get closer to the truth. So you use architecture as a basis, but you actually recreate many things.
Yes, we base ourselves on that. So, if it concerns a building or an environment, we use different forms of media to then build it in a 3D space. We work with satellite images, we work a lot with archives, we work a lot with open‑source information. Obviously, this needs to be verified in several ways. We analyse metadata, if we can get direct access to the original material. If it is not the original material, then we cross‑reference it with other things.
We also work with any type of open‑source documents. In the case of larger investigations, like the Grenfell Tower fire, we worked with survivor testimonies that were available through the public inquiry. There was an entire website with the individual testimonies of each person. So we consolidate all our sources and then establish connections between them to see what really makes sense.
How do you choose the situations you want to analyse? I know that people can contact you and report that something happened and is not being properly investigated, but do you have other criteria? I imagine many people ask for your help…
Yes, that’s true. Normally, we work at the request of communities facing direct human rights violations. And another criterion is understanding whether our techniques are useful in any way for that specific investigation, and whether there is an incentive to further develop our techniques. Several criteria need to be met, so that we can also advance as an organisation and support those communities in their pursuit of justice.
What are those criteria, the ones that make you realise your methods are useful for that community?
It mainly has to do with the fact that we use architectural tools. These are cases involving reconstructions of buildings or environments that we believe are possible to reconstruct with our tools, or when there is media analysis within a certain environment.
A few years ago, I worked on a police‑violence case in Amsterdam and all we had were police testimonies. It involved the death of a young man at the hands of the police, and we also had local residents who filmed from multiple angles. In that case, we knew we could investigate because we didn’t only have the police account. We obviously wouldn’t work solely with that. We use it to develop certain aspects or counter their testimony because we had visual elements. We had available videos from which we could analyse certain movements from different angles. That’s what we do: we collect visual material, synchronise it based on metadata. If it is a photo, if it is a video, we can place it on a timeline and cross‑reference those moments. So that would be the criterion. Also the amount of sources available to carry out an investigation.
How can the spatial analysis and visual reconstruction methods used by Forensic Architecture be useful for journalists who work in this area?
By using satellite images or extracting certain geospatial data, we are able to track different changes in the environment. In our work, for example in projects related to Gaza, we have observed the destruction of the environment and, in particular, the destruction of agriculture in that territory. We were also able to directly monitor conflicts and identify changes in the militarization of certain regions by analyzing those images, as well as site changes, such as the construction of illegal settlements, etc.
By extracting certain geospatial points, we were also able to verify specific events. If an image is found on X or any social media platform, sometimes, if we get to the right source, we can extract geospatial information to verify where and when the picture was taken, helping ground the evidence in a specific space.
These methods are also useful for building a narrative, right? It can also become a kind of storytelling tool?
Yes, exactly. In some of the work we’ve done in the past, we’ve done that, to get to the bottom of an issue or certain environmental changes. By analyzing contemporary satellite data alongside historical maps — such as in our projects in Louisiana — we were able to identify mass graves or cemeteries that were later transformed into something else, in a contemporary setting. So, layering out different types of data points allows us to track these changes over years or even centuries, helping us understand the terrain and its history.
What principles do you consider essential to ensure that spatial visualizations — such as maps, reconstructions, diagrams — do not reinforce biased narratives?
For projects like the one we did for Gaza, it’s all set in QGIS [geographic information system]. We’re using tools that are publicly used for research. They’re used in schools, universities, et cetera. We rely on tools and datasets that are publicly accessible and widely used in research environments. With that, I think we are very transparent and we show that we are not taking out resources out of nowhere.
When working with archival maps, we always attach the source and clearly indicate the date, so viewers understand exactly where each element comes from. I think when it comes to working with multiple sources, you always need to have a clear annotation where the source comes from, and [show] that it’s not just a randomly found image. Maintaining clear documentation of where every source comes from is one of the most important things.
In the case of Gaza, everything we have been tracking is based on information that emerges during the ongoing war campaign. For example, the way we trace evacuations of people from specific neighbourhoods is based on actual evacuation orders issued by the Israeli army. These documents are publicly available and can be searched for. Our analysis is built directly on those sources, ensuring that the research accurately traces and references its origins.
In many cases — maybe all — you go further than the investigation conducted by the competent authorities.
Yes.
But you also face difficulties in having your evidence accepted by courts in some contexts, right?
In some cases, yes. I think it is becoming easier nowadays, especially because our founder, Eyal Weizman, is on the technological board of the ICC [International Criminal Court]. So now the evidence we produce, or that citizens produce, is increasingly recognised in certain cases. And, as I said in my presentation, we are supporting the South African legal team in the case against Israel. This evidence is being recognised, but in that case, we work in collaboration with the lawyers. We don’t just go to court on our own. That is also how our organisation is built: we are a multidisciplinary team of architects, filmmakers, programmers, lawyers, journalists. We work through these collaborations, just as we handle the sources available. We also approach work methods in dialogue with as many specialists as possible. If it is an investigation into a bombing or an explosion, we consult ballistic experts or other types of specialists, for example.
But the main focus is always spatial information, spatial reconstruction?
Yes, and more recently, we’ve been working a lot with sound. Our partner organisation, Earshot, also works forensically but mostly with sound. We’ve done many sound investigations together. Things that combine space and sound, that allow us to understand how sound travels and what can be detected through sound. But yes, our investigations mostly concern specific environments.
As you said, many of your investigations rely on open‑source data. What best practices would you recommend for journalists who are verifying and using this type of data?
First of all, it is essential to keep a clear record of all sources you’re using. Maintaining a spreadsheet listing the information each data point contains — for instance, everything embedded in an image — is very important. I think for us it is also [important] trying to verify it with as many other sources [as possible].
If a source was reposted by an established news organization, for example, we take that into account. Individual posts by citizens are obviously really important but, in order to ground it better, we also look at what other people, other media outlets or additional reposts have been made of this particular moment in time, and then we verify it.
The Cartography of the Genocide platform is the best example of this [approach]. Whenever we track incidents in Gaza, we include a visible verification level so that it is clear whether an incident has been confirmed or is still pending verification. So you can see if an incident is verified or if it’s still to be verified, clearly. We always make sure we are transparent with our research. We always make clear to what extent we are certain with a specific event taking place.
Another important practice is keeping it extremely organized. While monitoring conflicts when incidents, you get more and more incidents throughout the time. So keeping it clearly organized, annotated and linked to its sources [is crucial].
It is also helpful to download images and maintain a personal database, because posts on social media are often deleted or accounts become unavailable. So, as a journalist, you can also download these sources and keep them organized and annotated in your own archive. That helps moving forward because I don’t think you can always rely on the internet or social media to keep a clear record of evidence. You never know [what’s going to happen]. Accounts get banned, accounts gets deleted, evidence gets taken down. So I think the best way is to try and gather as much of that for your research.
Thanks for reading this new GEOJO issue!
If you’re new here, subscribe and check out our upcoming course.