An assessment of the spatial distribution of shallow landslides induced by Hurricane María in Puerto Rico

ABSTRACT

Rainfall associated with Hurricane María triggered more than 70,000 shallow landslides over a 6,400 km² area of Puerto Rico for a mean density of 11.2 slope failures per km². When compared to other single storm landslide inventories with similarly sized search areas in tropical and sub-tropical regions, HMA’s mean landslide density is eleven times greater than the average. This study relied on frequency and zero-inflated negative binomial regression analyses to explain the spatial variability of landslide densities induced by HMA in relation to soil, geologic terrane, land cover, total storm precipitation, slope, and elevation. Results show that landslide densities tended to be exponentially higher for those areas with the highest rainfall totals and greater on igneous intrusive and other non-limestone sedimentary geologic terranes. Additionally, maximum densities occurred on relatively moderate slopes of 35–50% and not on the steepest hillslopes. Although a resulting regression model over-predicts landslides at lower elevations and underestimates high landslide density areas, it provides an acceptable fit for the landslide densities at a 0.25 km² scale. The results presented here can guide future landslide susceptibility analyses to adequately contextualize the heterogeneity of landslide activity caused by extreme tropical cyclones in Puerto Rico and elsewhere.

KEYWORDS:

• Tropical cyclone
• landslide
• mass wasting
• debris flow
• Caribbean

Acknowledgments

No funding was solicited nor required for the completion of this research. Our sincere gratitude to Dr. Odin Marc (University of Strasbourg) for providing us with electronic version of the worldwide landslide inventory, Alejandro Soto for providing guidance in the preparation of the geologic terranes database, and to Caroline Garnett (UT-Austin) for assisting in the preparation of databases and maps. Our sincere gratitude to all anonymous reviewers for helping improve earlier versions of this manuscript.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplemental data

Supplemental data for this article can be accessed here.