Landslide hazard cascades can trigger earthquakes

Mass movements like landslides are significant hazards in mountainous regions, often triggered by factors such as rainfall, snowmelt, and human activities. In seismically active areas, large earthquakes can induce widespread mass wasting, further exacerbating the risk. When these movements dam rivers, creating LDLs, subsequent dam breaches lead to catastrophic outburst floods. The impact of earthquakes on surface processes has been extensively studied, focusing primarily on how earthquakes trigger landslides and initiate hazard cascades. However, the potential for surface hazards to trigger earthquakes remains under-researched. Studies have shown that processes altering the Earth's crustal stress state, such as water storage behind dams, hydraulic fracturing, and lake filling, can induce earthquakes if the stress increase exceeds a critical threshold. Given that mass movements redistribute sediments and alter water storage, leading to stress changes, it's plausible that these could trigger earthquakes. This study investigates this bidirectional interaction using the 2018 Baige landslide hazard cascades as a case study.

Existing literature extensively documents the impact of earthquakes on triggering landslides and subsequent hazard cascades, including the formation of LDLs and outburst floods. Studies have quantified and predicted the spatial distribution of mass movements and analyzed the geomorphic evolution after earthquake events to improve hazard mitigation. However, research on the reciprocal effect—landslides triggering earthquakes—is limited despite evidence suggesting that surface loading and fluid diffusion from various sources can influence the stress state of the Earth's crust and trigger seismic activity. Previous studies have investigated the impact of processes such as earth tides, water storage in reservoirs, hydraulic fracturing, waste fluid disposal, and lake filling on earthquake triggering. This study builds upon this existing research by focusing specifically on the potential for LDLs, formed by landslides, to induce earthquakes.

The study uses the 2018 Baige landslide hazard cascades on the Tibetan Plateau as a case study. The researchers built an earthquake catalog using continuous seismic data from 13 nearby stations (2014-2023), employing machine-learning (PhaseNet) and cross-correlation-based (PyOcto) methods. This resulted in approximately 3970 earthquakes within a 1° × 1° region. The spatiotemporal evolution of seismicity and stress changes due to the hazard cascade were analyzed. Coulomb failure stress changes (ΔCFS) were modeled, considering the combined effects of direct gravitational loading and pore pressure diffusion from the LDLs using GeoTaos software. The model uses a three-fault system (Boluo-Tongmai, Jinshajiang, and Gangtuo-Yidun faults) with assumed geometries informed by regional earthquake focal mechanisms. A global LDL database was used to explore the potential for LDLs to trigger earthquakes. The statistical significance of observed seismicity rate changes was assessed using various statistical tests, including Z-statistic and P-statistic tests, considering declustering to remove aftershock sequences. Uncertainty analyses were conducted considering aspects such as earthquake location uncertainties, fault geometry, friction coefficient, and hydraulic diffusivity.

The study found a statistically significant increase ( >99% confidence) in earthquake activity (local magnitude ≤ 2.6) within 10 km of the Baige LDLs as the second, larger LDL approached peak water level. This increase coincided with calculated increases in ΔCFS due to the combined effects of direct loading and pore pressure diffusion. The earthquake activity decreased to background levels after the dam breach. About 90% of the earthquakes were located in areas with positive ΔCFS. The first, smaller LDL did not trigger a significant increase in seismicity, possibly due to a lower peak ΔCFS (0.007 MPa) compared to the second (0.024 MPa). Comparing the Baige LDLs to a global database of reservoirs that induced earthquakes, the study found that many LDLs globally have similar heights and durations to those that have triggered earthquakes. Modeling of a typical 30-m deep LDL showed that ΔCFS could exceed 0.01 MPa (a threshold associated with earthquake triggering) within 11 days, further supporting the potential for widespread earthquake triggering by LDLs. The pore pressure diffusion was found to be the primary triggering mechanism for many events, especially after the dam breach.

The findings demonstrate that LDLs can trigger earthquakes, primarily through pore pressure diffusion. While direct loading can cause stress decreases in some areas, pore pressure consistently increases ΔCFS. The study's focus on a relatively seismically inactive region helped to isolate the impact of the LDLs on seismicity. The relatively low magnitude and number of triggered earthquakes are likely related to the region's low background seismicity, suggesting larger or more frequent earthquakes might occur in seismically active regions with similar LDLs. The study highlights the bidirectional nature of earthquake-surface hazard interactions. Earthquakes can trigger landslides, which then form LDLs; the LDLs, through pore pressure diffusion and direct loading can, in turn, induce earthquakes. This bidirectional interaction suggests a potential feedback loop, particularly in regions with high seismic activity and frequent landslides, like the Wenchuan earthquake area which saw many large LDLs forming after the 2008 event.

This study provides compelling evidence that landslide-dammed lakes can trigger earthquakes, adding a crucial dimension to our understanding of earthquake-surface hazard interactions. The bidirectional nature of this interaction should be incorporated into geological hazard assessments and management strategies, particularly in mountainous regions. Future research could focus on investigating this phenomenon in more seismically active regions and exploring the role of other surface hazards, such as glacial lakes, in earthquake triggering.

The study focuses on a single case study, limiting the generalizability of the findings. The exact fault geometries in the study area were not fully constrained, requiring assumptions about fault dip and slip directions. The chosen hydraulic diffusivity value could influence the modeled ΔCFS, although a range of values was explored and confirmed consistent results. While various statistical tests support the significance of seismicity rate increases, the relatively low background seismicity rate could amplify the apparent significance.