As a more than twenty-year-old citizen science project, OpenStreetMap (OSM) has become a widely used geospatial platform for applications such as urban planning [3] and humanitarian response [4]. Its data can reach the same quality level as authoritative data in western countries [2]. In many developing regions, it represents the main source of geographic information regarding roads or buildings, where no national mapping agency exists.
OSM has proven particularly valuable in the context of short-term natural disasters [1] such as in Haiti or Japan. There, the roads can be updated rapidly to support rescue missions. But the past decade has brought the return of a large-scale, long-duration human catastrophy named "war". Unlike natural disasters, these events unfold over years and across vast territories, raising new questions about how mapping communities respond over time: how does editing activity evolve in relation to conflict dynamics, and how close to the active frontline are OSM data updated?
This study analyzes OSM contributions in Ukraine during the ongoing war, with the objective of understanding how mapping activity responds to long-duration armed conflict. Two main axes are explored: (i) the spatio-temporal relationship between OSM building edits including deletions, ruin-related tags, aerial bombardment events, and (ii) the evolution of OSM contribution levels in the vicinity of the active frontline over time. The analysis focuses primarily on buildings, which represent the most informative object type for this context with nearly 7.5 million objects in Ukraine according to OSM Taginfo.
Area of study. Our area of study is the whole country of Ukraine within its internationally recognized pre-war borders, including Crimea, as all our data providers treat it as part of the country. Ukraine was chosen over other active conflicts (Gaza, Iran) because its mapping culture aligns closely with western OSM practices, resulting in high baseline contribution levels and a richer edit history to analyze. Russia and Belarus are excluded, as we hypothesized that OSM contributions in both countries may have declined during the war.
Data and methods. The data and methods are summarized in Figure 1. The first column features the data, the second the methods and the third the theme of the research.
OSM history data were retrieved using the ohsome API, which exposes the full edit history of OSM objects. We also used Geofabrik to extract Ukraine's administrative boundaries. Conflict data were sourced from the Armed Conflict Location & Event Data (ACLED) global monitor, providing georeferenced records of air strikes and shelling events. Frontline geometries were assembled from three complementary sources to cover the full conflict period: Institute for the Study of War (ISW) shapefiles for February 2022, ACLED battle events as a spatial proxy from March 2022 to June 2024, and DeepState daily files from July 2024 onwards. For each time step, the distance from OSM-modified building centroids to the contemporary frontline was computed. This allowed the spatial relationship between editing activity and conflict intensity to be tracked over time.
Results. Preliminary results reveal a strong temporal signal following the February 2022 invasion. As shown in Figure 2, a sharp spike in building-related edits occurred in the weeks immediately after the invasion, concentrated primarily around Kyiv, before activity progressively redistributed across the country. Year-on-year comparisons between February and March 2022 and the same period in 2023 confirm that this initial surge was not sustained. This suggests that the humanitarian mapping response was most intense in the earliest phase of the conflict.
Spatial analysis shows that OSM contributions do not concentrate exclusively in safer areas: significant editing activity is observed near the frontline, even as it shifted over time. Areas experiencing intense bombardment display heterogeneous behaviors, with both increases — consistent with damage mapping by remote volunteers — and local decreases, likely reflecting contributor displacement or infrastructure disruption. Figure 3 shows the perfect example of an area inside the frontline (in green) with multiple clusters of contributions. A moderate positive Pearson correlation between monthly conflict event counts and OSM modification counts per spatial unit in the first year of the conflict suggests that bombardment events act as a partial trigger for mapping activity. This also highlights the significant role of remote contributors whose activity is decoupled from local ground conditions.
These findings provide new insights into the behavior of volunteered geographic information (VGI) communities in long-duration conflict settings and open avenues for further research on the role of VGI in crisis-affected territories.
Roads, with nearly 3 million objects according to OSM Taginfo, are considered as a perspective for future work. Indeed, roads also have strategic importance in war, so updating them may give an advantage to the enemy. Another perspective of this project is the analysis of contributors to see if these edits are done on the field by locals or by image analysis. All scripts and processed datasets used in this study are openly available at a repository to ensure full reproducibility of our results: https://anonymous.4open.science/r/osm-war-ukraine-2737. Raw conflict data from ACLED are not included in the repository as they require a dedicated access request through the ACLED researcher tier, available on their website. Although, it is simple to obtain access to that data thanks to the ACLED team.