Rain and small earthquakes maintain a slow-moving landslide in a persistent critical state

Large, shallow earthquakes (Mw > 5.5) trigger widespread landsliding in mountainous regions, significantly impacting the mass balance of earthquakes. Most earthquake-triggered landslides are activated co-seismically by moderate to large earthquakes (magnitude > 4). Several mechanisms explain co-seismic triggering: increased shear stress from dynamic loading, progressive soil weakening from ground motion cycles, and a rapid drop in shear resistance due to grain crushing (particularly when combined with high precipitation). This last mechanism, known as undrained loading, can lead to co-seismic reactivation or rapid triggering. Observations indicate varying time delays between seismic shaking and landslide activation, often influenced by precipitation. Changes in groundwater conditions due to microfractures can take days to reach the sliding surface, leading to delayed landslides. Increased rates of rapidly triggered landslides have also been observed months to years after major earthquakes. Slow-moving landslides are similarly affected, with increased velocities observed years after earthquakes. This is linked to decreased rock strength due to earthquake-generated micro/macrofractures, which create preferential paths for precipitation infiltration. The interplay between shaking, precipitation, material damage, and permeability remains poorly understood, with existing hypotheses based on qualitative observations. This study aims to quantitatively document these mechanisms using in-situ measurements of a slow-moving landslide in Peru, where earthquakes and seasonal rainfall combine.

Existing research on earthquake-triggered landslides primarily relies on regional inventories and post-event assessments, offering limited insight into the complex interactions between seismic activity, rainfall, and landslide mechanics. While studies have qualitatively observed the influence of rainfall on the timing and magnitude of landslide events, a quantitative understanding of how these factors interact to influence landslide behavior has remained elusive. Studies have highlighted the role of undrained loading, seismic weakening, and changes in groundwater conditions, but lack comprehensive, in-situ measurements to verify proposed mechanisms at the landslide scale. The temporal scales involved in the healing process following seismic events and the impact of smaller magnitude earthquakes are also not well understood. This study addresses these gaps through in-situ monitoring.

The study focuses on the Maca landslide in Peru, a large, slow-moving landslide situated in a seismically active area with seasonal rainfall. The landslide's velocity is primarily driven by precipitation and river erosion, with a rainfall threshold above which deformation is triggered. Regional earthquakes can also accelerate motion. A hut housing a GPS and a broadband seismometer was installed on the fastest part of the landslide in December 2015. These instruments continuously monitor surface displacement and relative seismic velocity changes (dv/v) using ambient seismic noise. GPS campaigns, conducted every three months since 2013, provide additional displacement data. The study analyzed surface GPS displacement and seismic activity during rainy and dry seasons. The influence of undrained loading, water table variations, and soil damage due to earthquake shaking was investigated using poroelastic models. Relative seismic velocity changes were analyzed to assess soil rigidity and its recovery after seismic events. The impact of smaller earthquakes (PGV < 1 cm/s) on landslide dynamics was also examined. Peak Ground Velocity (PGV) was used as an indicator of seismic shaking intensity, obtained either from direct seismometer measurements or approximated using Ground Motion Prediction Equations (GMPEs). Meteorological data (precipitation, temperature) were used to calculate effective precipitation.

The study found that the combined effect of earthquakes and rainfall produces greater landslide motion than either factor alone. A larger earthquake (MI 5.5) caused a co-seismic slip of 1 cm, followed by 11 cm of post-seismic displacement over 30-40 days. A smaller earthquake (MI 5.0) during the rainy season resulted in much greater displacement (80 cm over 5 months). Comparisons with other rainy seasons (2014 and 2017) with less seismic activity demonstrated that the increased displacement in 2016 was attributable to the combination of earthquakes and rainfall. Analysis of relative seismic velocity changes (dv/v) showed a co-seismic drop in velocity exceeding 2% after the MI 5.5 earthquake, attributable to soil damage due to shaking. This drop was less pronounced for the MI 5.0 earthquake. Modeling excluded undrained loading and water table variations as primary causes for this velocity drop. The post-seismic recovery of dv/v was logarithmic, similar to previous findings. However, this recovery was noticeably delayed during the rainy season and in years with high small-earthquake activity, even when the cumulative precipitation was not significantly higher. This suggests that frequent smaller earthquakes prevent soil healing, maintaining the landslide in a 'critical regime' where even minor forcings can trigger motion.

The findings demonstrate that the interaction between rainfall and earthquakes drives landslide dynamics, confirming previous qualitative observations and providing quantitative support for the hypothesis that soil damage from shaking facilitates water infiltration, thus increasing landslide motion. The study highlights the importance of considering the timing and magnitude of earthquakes and rainfall, and the cumulative effect of smaller seismic events. The persistent critical state of the landslide is linked to the balance between seismic damage and the limited healing allowed by persistent water saturation and frequent smaller earthquakes. The observed mechanisms are likely applicable to a wider range of slow-moving landslides, despite the specific geological context of the study site. Differences in geology, thickness, and pore saturation will likely influence the quantitative aspects, however, the fundamental mechanisms are transferable.

This study provides quantitative evidence that the combination of rainfall and earthquakes, including smaller-magnitude events, significantly impacts the dynamics of slow-moving landslides. The findings underscore the importance of soil damage, increased water infiltration, and the temporal interaction of various seismic events and rainfall in sustaining a persistent critical state for landslides. Future research should focus on extending these findings to other landslide types and geological settings, investigating the specific thresholds for seismic triggering under varying rainfall conditions, and developing predictive models incorporating these complex interactions. These insights have implications for landslide hazard assessment and the understanding of landscape evolution in seismically active regions.

The study focuses on a single landslide; while this landslide exhibits representative features, the generalizability of the findings needs further investigation through studies on diverse landslide types and geological settings. The approximation of PGV using GMPEs introduces uncertainty, especially for larger events that saturated the seismometer. More refined methods to estimate the PGV for large earthquakes would improve the accuracy of the analysis. The study's temporal scope of three years may not fully capture long-term trends in landslide behavior; longer-term monitoring would enhance the robustness of the conclusions.