Observability Core Vector Trajectories for Structural Health Monitoring: From Stable States to Structural Change

Abstract What does infrastructure look like to radar as it changes, is repaired, reconstructed, or fails? We extend our infrastructure-observability framework into the temporal domain using the Observability Core Vector (OCV), which combines six Sentinel-1 descriptors of cross-acquisition echo consistency and spatial echo-mask geometry. We examine whether OCV trajectories provide complementary information for structural health monitoring across stable operation, controlled state transitions, reconstruction, repair, major configuration change, and documented failures. Physical processes produce distinct, weak, or absent OCV changes depending on the asset, acquisition geometry, measurement support, and temporal reference. Observable area and fragmentation vary across different physical processes but do not share a common health direction, and apparent long-term evolution can depend on the measurement layer. The cases do not support transferable life-cycle trajectories or a universal failure signature. We use the OCV to characterize temporal change in an asset's radar representation and provide complementary information for infrastructure monitoring. Structural interpretation requires independent measurements or validated physical models.