In February 2021, three months into the war in Tigray in northern Ethiopia, researchers noticed something that would confirm their worst fears about the nature of the conflict.

With a communications blackout and access to the region largely closed off, the team from research organisation Vigil Monitor turned to satellite data to monitor the conflict. On the 10 February, after noticing a NASA fire management readout of a single fire in the middle of a road in a rural village called Adi Chilo, Vigil’s alarm bells went off and a rapid response satellite image was requested. The devastation in the area was stark and upon first inspection, the images of burnt-out vehicles, houses, and food stocks pointed towards damage caused by battle, says Alexander Lee, Vigil’s director and lead investigator.

But something didn’t look right. The structures were burned from the inside out without connecting burn scars, and the damage was far too extensive. From above the scars resembled a shotgun spread-like pattern that indicated a possible systematic door-to-door burning of homes and livelihoods.

The team analysed high- and low-resolution satellite imagery from different dates and triangulated this with other data – witness testimonies, social media, local and expert insights, and media reports of mass killings on 10 February. Vigil deduced there had been two incidents, not one: damage to vehicles inflicted on 9 February during a military ambush; and fire damage to houses and grain stores on 10 February. The second incident appeared to be a deliberate retaliation on the civilians in the immediate area by the security forces who were ambushed. This atrocity was by no means unique. The Vigil team applied their mixed methods research in multiple locations to demonstrate that the tactic of following up military ambushes with civilian attacks was, in fact, a trend across Tigray. Incident by incident, they developed an overall picture of widespread and systematic human rights abuse of thousands of civilians across the Tigray region.

A New Direction for Conflict Analysis

Satellite and open-source data are now reshaping how we understand the drivers, dynamics and direction of conflicts – particularly in regions off-limits to international actors and subject to information blackouts.

This wasn’t always the case. For decades, satellite imagery was the exclusive domain of governments, originating in the military and intelligence operations of the Cold War and the Space Race. Then, in 2000, regulations were relaxed and satellite data (also known as remote sensing) opened up to multiple sectors. For those researching and investigating conflicts, it transformed the monitoring of atrocities – providing visual evidence of violence, demolitions, mass graves, secret prisons, and other abuses.

Two decades on, the use of remote sensing research in conflict analysis is much broader. A growing number of companies provide ever-cheaper data to diverse organisations – research institutes, universities, media, governments, civil society, commercial risk analysis, and others. Researchers, including many working on the XCEPT programme, increasingly combine satellite and open-source data analysis with traditional research methods to dig deeper into why violence occurs, how conflicts connect across borders, the fallout for communities, and ways to build peace.

And, crucially, this research feeds into conflict and humanitarian response policies and interventions.

“We’re looking to use evidence and information so that we can track evolving situations, in order to prevent violence and conflict,” says Moazzam Malik, former Director General, Africa, at the UK Foreign, Commonwealth and Development Office (FCDO). As well as documenting atrocities and laying the groundwork for accountability, policymakers use this mixed methods research to “understand the motivations and entry points for a whole range of conflict actors, so that one can begin to think about resolution, mediation, and… address underlying grievances.”

Amid information wars, ‘fake news’, and authoritarian media crackdowns, sifting truth from lies is difficult. Research that draws on multiple datasets to triangulate evidence, cut through the misinformation, and arrive on policymakers’ desks in a timely, concise and reliable fashion, is vital, helping shape policy responses more finetuned to meet conflict challenges.

Melding Old and New

With the use of remote sensing technologies in the commercial and civil society space booming, it is crucial that researchers engage with them. Finding the optimal mix of new and established research methods to identify conflict trends takes time, tenacity, expertise, and sound contextual knowledge. It typically involves honing in on an incident, understanding the specifics of it as thoroughly as possible – including by triangulating research with other data sources – and then by expanding to finding similar events.

– What is remote sensing? How is it used in conflict research?

Remote sensing research involves accessing data on Earth and its atmosphere using sensors located away from the object of study. There is a vast range of government-owned and commercial satellites and data types:

Passive remote sensing satellites produce data by detecting energy emitted by or reflected back from Earth. Those monitoring solar energy need sunlit and cloudless conditions to operate, while Night Light imagery captures lights from human activity at night.

Active remote sensing satellites send signals to Earth and measure those received back. They include Synthetic Aperture Radar (SAR) satellites, which use radar to produce images. SAR works at night and in cloudy conditions.

Infrared satellites detect heat energy and display objects based on their temperature level.

LIDAR technologies on satellites use laser pulses to measure ranges to Earth and generate precise 3-D imagery of Earth’s surface. LIDAR also detects toxic gases and biological warfare agents in the air.

Radio frequency (RF) data is captured by satellite systems such as HawkEye 360. These monitor radio frequencies emitted by multiple objects – ships, for example – to identify and geolocate objects.

Polar orbiting satellites orbit the Earth and provide data on different locations.

Geostationary satellites have orbits that do not move relative to the ground, remaining above the same place on the Earth’s surface and providing data on the same location over time.

The Global Positioning System (GPS) is a global navigation satellite system (GNSS) owned and managed by the US government. It provides geolocation and time information free-of-charge to anyone with a GPS receiver.

Satellite images are produced in low, medium, high, and very high resolution (VHR).

Satellite imagery resolution can be spatial, spectral, temporal, or radiometric.

Multispectral imagery visualises electromagnetic wavelengths and frequencies invisible to the human eye, such as infrared and ultraviolet.

Geographic information system (GIS) tools are computer systems that create, manage, analyse, and visualise information with a geographic location. They enable the integration and analysis of the satellite imagery with other geospatial data such as digital terrain models, road networks, land cover maps etc.

– What is open-source data in conflict research?

Open access and open-source data in conflict research includes:

Cartographic data provided by organisations such as NASA Earth Observatory, Humanitarian OpenStreetMap (HOT), and the UN OCHA Humanitarian Data Exchange.

Transponder data (AIS – Automatic Identification system) from shipping and aircraft.

Trade data from trade databases such as UN Comtrade.

Open access research made available free-of-charge by universities, research institutes, civil society organisations, governments etc.

Open-source software is any programme that is released alongside its source code, which users can use and modify for free. Examples in conflict research include the mapping programme Ushahidi and GIS tool QGIS.

News media in multiple formats – text, audio, video, web etc.

Social media including text, images, graphics, audio and video sent via Twitter, Facebook, Instagram etc.

Sometimes researchers work with data that the human eye cannot perceive. Infrared and false colouration, for example, enable the analysis of incidents and trends that would otherwise go unnoticed, while Synthetic Aperture Radar (SAR) satellites use radar to produce images, working at night and in cloudy conditions. In February, academics working with SAR and road traffic data from GPS were the first to spot that Russian troops amassed on the Ukrainian border were preparing to invade.

Pushing the boundaries

Satellite and tech-informed research methods enable a more holistic view of how multiple phenomena intersect to drive and resolve conflict. For example, they illuminate how climate change, water scarcity, food insecurity and conflict intersect and exacerbate each other – a critical area of research given that 14 of the 25 countries most affected by climate change are beset by conflict.

Research by XCEPT and RHIPTO on the tri-border area of Chad, Libya, and Sudan uses satellite and other data to show how falling oil prices have catalysed alternative livelihoods, including artisanal gold mining and people smuggling. It reveals how control over people, gold, and other commodities ebbs and flows between different powerbrokers and armed groups across borders, redrawing battle lines and affecting wider stability – the Libyan ceasefire, for example. And it shows how broader national, regional, and global factors affect these transnational dynamics. For example, COVID-19’s curbs on air travel led to a surge in illegal overland migration.

Mixed-methods research incorporating satellites has also revolutionised our ability to understand change over time. Researchers can access a “multi-year time machine that allows you to detect change”, in spectrums beyond the human eye, says Nathaniel Raymond, lecturer at Yale University’s Jackson Institute for Global Affairs. This is a game-changer for conflict analysis, expanding our understanding of conflict’s short-term shocks and long-term effects. By combining aggregated satellite imagery with other data, researchers can unlock correlations that previously would have been impossible.

This analysis helps shape policies more attuned to the granular detail, triggers and trends that make up the big picture over time. It sharpens policies and interventions on conflict resolution, humanitarian aid, peacebuilding, and long-term stability, shedding light on how to balance political, development and security priorities and better integrate local, national and regional responses. Alongside predictive peacekeeping tools, it can help forecast when conflicts might occur and how and when to intervene, enabling pre-emptive interventions rather than reactive responses, which typify peacebuilding today.

With much research being open access, it has rapidly become a public good benefiting multiple communities and governments.

Remote methods are new tools in researchers’ existing methods toolkit. They are most valuable when blended with traditional methods. The optimal mix leverages these new approaches alongside traditional qualitative and quantitative field methods. Ground truth data collection, such as interviews and surveys, enables calibration, framing and sense checking of remote methods.

A Revolution Requiring Careful Handling

Of course, remote sensing research is not without its challenges. As a new and growing sector, it faces a raft of issues – including a lack of expertise, outdated ethical frameworks, misuse and misinterpretation, and the constant danger of actionable, real-time information being exploited by bad actors to cause harm. The challenge now is to address these problems, be laser-focused on the problems at hand, and ensure robust ethical frameworks for research.

Misinterpretation is a major challenge. Often, expertise within academia has not kept pace with the sophisticated tools now available, while misinterpretation by proliferating users outside of expert networks is also widespread. Incorrect assumptions and false correlations are common pitfalls where satellite imagery and open-source data aren’t properly contextualised and analysed.

In the wrong hands, open-access data and research can cause severe harm to vulnerable populations and reshape conflict dynamics to devastating effect. For example, Raymond says the publication of a 2012 report by Harvard’s Satellite Sentinel Project (where he then worked) on road construction and military deployments in southern Sudan contributed to Sudanese rebels kidnapping 29 Chinese road builders.

Researchers are not Nostradamus, they cannot predict the future, he says. They need to recognise that information could be used in unpredictable ways and take the potential second-order effects of using and releasing tech-gleaned data much more seriously.

“We have to continually reckon with the balance between the need to disclose and the need to minimise harm”, says Lee. “Civilian communities overwhelmingly pay the price in war –  their protection… should extend to what information is shared about them and … [where they are]”.

Start With Problems, Not Solutions

The hype surrounding new tech is another issue. Researchers can be tempted to start with a new tech-based solution and find problems to apply it to, rather than the other way around. But technology rarely offers ‘magic keys’ and should be treated as an asset, not a panacea – and one requiring expert handling.

Moving away from ‘solutions-led’ to ‘problem-driven’ approaches is far more likely to lead to robust interventions, says Heather Marquette, professor of development politics at the University of Birmingham and a specialist in corruption and anti-corruption evidence and interventions. For researchers, this means “being driven by the research question and not the tool or the method – technology-based or otherwise”.

“If we keep starting with the research questions, built around solving real-world problems, we can help avoid investing in tech for tech’s sake, and all the potential consequences that could go with that”, argues Professor Marquette.

There is also a danger that remote sensing becomes a comparatively cheap and easy proxy for informed engagement with vulnerable communities and raises issues of consent and anonymity that require far more attention. If researchers make decisions based on remote data rather than community consultations, their research risks further disenfranchising – and potentially endangering – the very people it strives to support. As with all research, but particularly in sensitive conflict settings, “community engagement … can’t just be replaced by an appropriate or a convenient tech solution,” says Charlotte Watts, the Chief Scientific Adviser at the FCDO.

Spotlight on Ethics

All research requires the constant evaluation of risk and ethics, but the stakes are particularly high for conflict-related research: if undertaken or communicated poorly, it has the potential to cause harm and suffering on an enormous scale.

As remote sensing research gets more ubiquitous, researchers must use it responsibly from start to finish – understanding and mitigating its dangers, following ‘Do No Harm’ principles, and ensuring robust ethics frameworks are in place. Researchers need to justify using satellite imagery and adhere to research ethics principles such as anonymity, openness of purpose, and ensuring risks to researchers and informants are mitigated, says Marquette.

Research institutes and governance bodies also need to step up and overhaul ethics frameworks. One major problem, says Raymond, is the focus on personally identifiable information (PII), which has seen institutional review boards (IRBs) give the greenlight to research because it does not include PII. But this research might include what Raymond terms “demographically identifiable information (DII)” – group characteristics like ethnicity or gender – which he calls the “engine … of remote sensing”. Because IRBs don’t think about populations in this way, they overlook the harms that research involving DII could cause – revealing information that could lead to ethnic cleansing, for example. And this oversight puts the onus on individual researchers to self-police.

Research bodies could take the lead in reassessing the ethics of remote sensing research and providing guidance. Organisations like the Economic and Social Research Council could help work through dilemmas and provide frameworks for practitioners and researchers – including by developing best practices, publicising examples of harm, and ensuring systems adapt over time to reflect new realities.

None of this will remove research risks entirely. But being clear about research questions, adhering to high ethical standards, and striving to anticipate potential use and misuse could mitigate them substantially. This culture shift needs to happen fast. For, as conflicts grow increasingly complex, strategies based on ethically robust research that mobilises myriad data types to deepen our understanding of wars and ways to resolve them, are becoming ever more vital.