EUVIMEDEUVIMED entdecken
Crossref

Determination and analyses of rainfall thresholds for landslide initiation in the Greater Durban Metropolitan Region, Eastern South Africa

Sibonakaliso G. Chiliza, Molla B. Demlie, Egerton D. C. Hingston

Natural Hazards · 2026-11 · 122 (22)

DOI: 10.1007/s11069-026-08435-9 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Natural Hazards, online erschienen am 23.10.2026. Verfasst von Sibonakaliso G. Chiliza und 2 weiteren Autor:innen. Die Arbeit nennt 73 Literaturangaben. Gefördert von: Council for Geoscience.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This study investigates rainfall thresholds for landslide initiation in the Greater Durban Metropolitan Region, eastern South Africa. The rainfall threshold has been established using quantile regression analysis applied to 63 event-station pairs and five-decade landslide and rainfall datasets spanning 1970 to 2022. A 70th percentile cumulative event rainfall (E)-duration (D) threshold of E = 108D0.67 was established. Leave-one-out cross-validation yielded a 31.7% detection rate for this threshold, positioning it as a confirmatory, high-confidence tier within a multi-level early warning system. By contrast, the 50th percentile, with a 50.8% detection rate, provides a more sensitive first-tier advisory threshold. Correlation analysis of antecedent rainfall across 5–40-day windows demonstrated the important role of pre-event wetness, particularly over the preceding 10–30 days, in preconditioning slopes for failure, with event rainfall correlating most strongly with 10-day antecedent rainfall (Spearman ρ = 0.756). Combined duration–antecedent quantile regression using 20-day antecedent rainfall was a statistically significant predictor of event-rainfall magnitude at the 60th percentile and above. Area-normalised spatial analysis indicated strong lithological control on slope instability, with the Pietermaritzburg Formation shales exhibiting a greater intrinsic susceptibility than the dominant Natal Group sandstones. Temporal probability analysis indicates a three-year recurrence interval for significant landslide-triggering events; however, this estimate assumes rainfall stationarity despite a statistically significant decrease in annual rainfall over the study period. Integrating tiered rainfall thresholds and antecedent conditions into early warning systems can improve prediction and support the mitigation of landslide-related socioeconomic losses. Evolving precipitation patterns and rapid urbanisation nevertheless necessitate periodic reassessment of these thresholds, ideally complemented by real-time soil-moisture monitoring.

Förderung

  • Council for Geoscience

Zitieren

APA

Chiliza, S. G., Demlie, M. B., & Hingston, E. D. C. (2026). Determination and analyses of rainfall thresholds for landslide initiation in the Greater Durban Metropolitan Region, Eastern South Africa. Natural Hazards, 122(22). https://doi.org/10.1007/s11069-026-08435-9

Vancouver

Chiliza SG, Demlie MB, Hingston EDC. Determination and analyses of rainfall thresholds for landslide initiation in the Greater Durban Metropolitan Region, Eastern South Africa. Natural Hazards. 2026;122(22). doi:10.1007/s11069-026-08435-9

BibTeX
@article{Chiliza2026,
  author = {Chiliza, Sibonakaliso G. and Demlie, Molla B. and Hingston, Egerton D. C.},
  title = {Determination and analyses of rainfall thresholds for landslide initiation in the Greater Durban Metropolitan Region, Eastern South Africa},
  journal = {Natural Hazards},
  year = {2026},
  volume = {122},
  number = {22},
  doi = {10.1007/s11069-026-08435-9}
}
Kostenlos starten →