Integrating morphometric susceptibility and socio-behavioral analysis to understand flood risk perception in the rift flank west of Lake Kivu

Résumé fourni par la source

The rift flank west of Lake Kivu in the eastern Democratic Republic of Congo experiences recurrent flash floods exacerbated by protracted conflict. To inform sustainable disaster risk reduction (DRR) in this fragile context, we investigate the determinants of Flood Risk Perception (FRP) across three flood-prone sub-basins. We employ mixed-methods approach integrating: (1) a morphometric flash flood susceptibility index (MFFSI) derived from Digital Elevation Models, and (2) a hierarchical regression analysis of structured household surveys data ( n = 363) across Lukungula, Renga and Mubimbi sub-basins. Morphometric results reveal that although the sub-basins exhibit elongated geometries (Form factor = 0.326–0.333), theoretically associated with attenuated peak flows, extreme relief ratios (up to 100.8 m/km), orographic rainfall and deforestation-driven sediment dynamics override these geometric protections transforming these headwater systems into efficient flash flood generators. The MFFSI ranks physical susceptibility as Renga (0.552) ≈ Mubimbi (0.550) > Lukungula (0.466), an order paradoxically inverses to the observed FRP gradient. MFFSI yielded a significant negative main effect on FRP (β = −0.112, p = 0.018), identifying a morphometric blind spot: where physical susceptibility is high but direct flood experience is limited, the terrain's hazard potential remains cognitively invisible. Interaction effects between MFFSI and experiential factors were non-significant ( p > 0.05), confirming that the psychological imprint of direct flood experience acts as a universal driver of FRP regardless of basin configuration. Experiential factors, spatial proximity, sociodemographic characteristics, and adaptive capacity collectively explained 63.2% of FRP variance. Flood damage severity was the strongest predictor (β = 0.290, p < 0.001), followed by structural mitigation (β = 0.214) and flood frequency (β = 0.208). Our findings demonstrated that despite near-universal risk awareness (M = 4.61/5), preparedness remains critically low (M = 2.47/5). A dedicated preparedness regression formally decomposed this paradox: psychological threat appraisal explained only 9.7% of preparedness variance, while socio-structural factors dominated by income (β = 0.389) and IDP status (β = −0.298) added a further 30.1%. This confirms that preparedness is constrained not by ignorance but by a capacity ceiling dictated by structural barriers. We identify community-based early warning and MFFSI-informed anticipatory risk communication as the most immediate interventions for resilient DRR in fragile rift settings.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.