Estimation of empirical rainfall thresholds for landslide triggering using partial duration series and their relation with climatic cycles. An application in southern Ecuador

Abstract

Rainfall-induced landslides constitute a major cause of damage and fatalities throughout the intramontane basins of the Andes. The geological and climatic setting plays a key role in the generation of a high number of landslides in this area. For this reason, a greater understanding of the relationship between landslide frequency and climate conditions is necessary to mitigate human and economic losses.Accordingly, this paper presents an analysis of rainfall variables associatedwith a series of dated landslides (153 in total) in the southern Ecuador basin of Loja. This analysis was performed by applying an affordable empirical method that enables the calculation of critical rainfall threshold (CRT) curves. This calculation is based on an in-depth examination of rainfall parameters, such as cumulative precipitation and mean intensity, linked to a wide range of rainfall duration (from 1 to 90 days). The inspection of these parameters was addressed considering their frequency, which was calculated by using partial duration series (PDS), taking into account the entire rainfall record. This work has revealed that only 24% of landslides were triggered by rainfall conditions with maximum return periods greater than 1 year, whereas the rest did not exceed that return period. After finding the best correlation between the maximum return periods and the maximum mean intensity, a minimum power law function was adjusted to the CRTcurve that correlates duration and cumulative rainfall. The values for this CRT function resulted in 5.14 and 0.83 for its scaling constant (α) and shape parameter (β), respectively. In addition, a spectral analysis was conducted to detect climatic cycles on the entire rainfall record. In general, a clear correlation could not be established between climatic frequencies and significant rainfall events inducing landslides, although similarly return periods were found for a critical rainfall event of March 2015 (10.4 years) and the SUNSPOT cycles (10.5-12 years). The results derived from this research are significantly valuable for the prevention of future mass-movements, although additional data will be crucial to update and calibrate CRT curves to study the influence of climate on landslide event frequency and magnitude in Loja.