Attachments
Francisco Rowe, Carmen Cabrera, Elisabetta Pietrostefani, Matt Mason, Rodgers Iradukunda, Andrea Nasuto and Emiliano Beltran | Geographic Data Science Lab, University of Liverpool
Executive summary
Conventional data sources to monitor population displacement in Iran following the outbreak of war on 28 February 2026 are limited due to existing sanctions, active hostilities, information controls and restricted humanitarian access. This situation report uses digital trace data as an alternative way to measure likely population dynamics in near real time during the reporting period [1].
We leverage Cloudflare HTTPS request data as the main proxy for relative changes in population presence at the provincial level. Despite Internet blackouts, Cloudflare HTTPS request data can still provide informative signals of population activity because these blackouts rarely affect the entire network all at once. Small levels of external connectivity often persist through selected networks, partial routing or brief restoration periods, and fundamental services such as banking and medical services remain in operation.
Our Cloudflare-based estimates capture relative shifts in population presence across provinces - not flows or absolute counts. Readers are strongly encouraged to consult the caveats before interpreting findings. We assessed and validated the patterns from our Cloudflare-based estimates against Farsi Wikipedia pageviews and Iran Strike Map event data.
Our Cloudflare-based estimates display a clear geographic pattern and evolution of displacement as the conflict evolves. They indicate early displacement towards border provinces near Turkey, Armenia and Azerbaijan, with later concentration in North-central and Eastern provinces — complementing IOM DTM cross-border data recording approximately 40,000 departures between 3 and 10 March, primarily to Afghanistan, Pakistan, Turkey and Azerbaijan [2]. A more complete analysis will follow in a subsequent report.
Insights
• Apparent early population displacement towards border provinces near Turkey, Armenia and Azerbaijan in the Northeast
• More prominent population hotspots in provinces bordering Afghanistan and Tehran as the conflict evolves.
• Tehran shows a modest positive population change against its pre-war baseline, consistent with prior research on urban concentration during conflict [3].
• Central and to south western provinces emerge as the most prominent origins of displacement.
• Cloudflare data is calibrated against a pre-war baseline, adjusted for network disruptions, and validated against strike exposure and Wikipedia data.
• Findings are proxy evidence - confidence is stronger where changes align with the geographic patterns of strike exposure and Wikipedia data.