Event-Catalogue-Guided Multi-Event Sentinel-1 SAR Analysis of Flood Recurrence and Urban Vulnerability in Kisangani, Democratic Republic of the Congo

Introduction: Sentinel-1 synthetic aperture radar (SAR) is widely used for flood mapping, but its application remains limited in data-scarce tropical African cities. In Kisangani, Democratic Republic of the Congo, structured georeferenced flood inventories are lacking and, to the best of our knowledge, no previous study has conducted a multi-event Sentinel-1 assessment of urban flood recurrence and spatial vulnerability. This study links an independently reconstructed catalogue of documented floods with event-specific SAR mapping.

Methods: Seventeen documented flood events between 2017 and 2025 were reconstructed from Sentinel-1 imagery using pre- and post-event VH backscatter change detection, refined Lee speckle filtering, and masks for permanent water and steep terrain. Each event was retained individually to derive recurrence, relative frequency, spatial persistence, and inter-event similarity. Flood recurrence was integrated at 100 m resolution with geospatial inventories of exposed urban elements, including schools, health facilities, transport infrastructure, settlements, services, and roads. Documentary sources were used for qualitative corroboration of mapped results.

Results: Across the 2,194.162 km² study area, 50.65% (1,111.387 km²) was detected as flooded at least once, 19.15% (420.255 km²) at least twice, and 6.10% (133.910 km²) at least three times. Local recurrence reached 13 of 17 events, while 7.97% (174.945 km²) was affected during at least two successive events. Among 221 named geographic features, 147 were directly exposed, and 75 named roads contained affected sections. The balanced vulnerability model identified 1,707 evaluable 100 m cells, including 135 high- and 29 very-high-vulnerability cells, together covering 0.075% (1.640 km²) of the study area.

Conclusions: The main contribution is an event-catalogue-guided, multi-event Earth observation framework that reconstructs undocumented flood dynamics and transforms individual Sentinel-1 observations into recurrence, persistence, and location-specific urban vulnerability information for a poorly documented tropical African city.