Operationalising EMS-98 Damage Classification: A UAV-to-GIS Pipeline for Macroseismic Survey Support

Giovanni Galli , Marco Dubbini, FIlippo Bernardini, Luca Arcoraci

Post-earthquake macroseismic surveying often relies on ground-based visual inspections that are slow, costly, and difficult to scale in the immediate aftermath of a seismic event. Deep-learning damage detectors have advanced substantially in recent years, yet their outputs are rarely translated into operational deployment tools that yield a georeferenced dataframe of buildings aligned with the European Macroseismic Scale 1998 (EMS-98). This study presents the Macroseismic Survey Mapper (MSM), a prototype end-to-end pipeline that links unmanned aerial vehicle imagery acquisition, instance-level damage segmentation across the five EMS-98 severity grades, per-building aggregation under a worst-observed rule, and export to an OGC GeoPackage layer ready for direct ingestion into geographic information systems. On the held-out test set the model reaches 64.6% exact-match accuracy; treated as a binary triage classifier, it recalls 90.3% (65/72) of buildings carrying actionable damage (EMS-98 grade S3 or higher) while over-grading only 13.5% (5/37) of lower-grade buildings, with all but two mis-classifications falling within one grade of the ground truth. The pipeline is exercised on a real UAV deployment over Piedilama (Arquata del Tronto, Central Italy), mapping the surveyed block at 3.78Å}0.32 s per image. The MSM establishes a deployable end-to-end workflow; future work targets UAV-native training data to quantify operational accuracy at scale.