Integrating satellite imagery with text data reshapes conflict research and humanitarian response

Armed conflicts are at unprecedented levels, yet most quantitative research still depends on text-based datasets that record fatalities and suffer from uneven media coverage. Satellite-derived observations, while still developing, offer an alternative view of war-related damage to infrastructure, agriculture and the natural environment. The authors describe three ways to merge these sources: improvement (adding missing features such as spatial precision), enrichment (linking complementary variables like killings and building loss) and fusion (jointly modeling both modalities).

Case studies from Ukraine’s frontline war and Myanmar’s 2017 Rohingya campaign demonstrate how integration reveals patterns—such as differing targeting by Russian and Ukrainian forces or prolonged displacement after initial massacres—that single-source analyses miss. The authors call for broader adoption of these methods, expanded training data for automated detection, and interdisciplinary safeguards to ensure ethical use.