A knowledge-aware deep learning model for landslide susceptibility assessment in Hong Kong.

Despite the success of the growing data-driven landslide susceptibility prediction, the model training heavily relies on the quality of the data (involving topography, geology, hydrology, land cover, climate, and human activity), the structure of the model, and the fine-tuning of the model parameters. Few data-driven methods have considered incorporating 'landslide priors', as in this article the prior knowledge or statistics related to landslide occurrence, to enhance the model's perception in landslide mechanism. The main objective and contribution of this study is the coupling of landslide priors and a deep learning model to improve the model's transferability and stability. This is accomplished by selecting non-landslide sample grounded on landslide statistics, disentangling input landslide features using a variational autoencoder, and crafting a loss function with physical constraints. This study utilizes the SHAP method to interpret the deep learning model, aiding in the acquisition of feature permutation results to identify underlying landslide causes. The interpretation result indicates that 'slope' is the most influential factor. Considering the extreme rainfall impact on landslide occurrences in Hong Kong, we combine this prior into the deep learning model and find feature ranking for 'rainfall' improved, in comparison to the ranking result interpreted from a pure MLP. Further, the potency of MT-InSAR is utilized to augment the landslide susceptibility map and promote efficient cross-validation. A comparison of InSAR results with historical images reveals that detectable movement before their occurrence is evident in only a minority of landslides. Most landslides occur spontaneously, exhibiting no precursor motion. Comparing with other data-driven methods, the proposed methods outperform in accuracy (by 2 %-5 %), precision (by 2 %-7 %), recall (by 1 %-3 %), F1-score (by 8 %-10 %), and AuROC (by 2 %-4 %). Especially, the Cohen Kappa performance surpasses nearly 20 %, indicating that the knowledge-aware methodology enhances model generalization and mitigates training bias induced by unbalanced positive and negative samples.

Earthquake-induced soil landslides: volume estimates and uncertainties with the existing scaling exponents.

Quantifying landslide volumes in earthquake affected areas is critical to understand the orogenic processes and their surface effects at different spatio-temporal scales. Here, we build an accurate scaling relationship to estimate the volume of shallow soil landslides based on 1 m pre- and post-event LiDAR elevation models. On compiling an inventory of 1719 landslides for 2018 Mw 6.6 Hokkaido-Iburi earthquake epicentral region, we find that the volume of soil landslides can be estimated by γ = 1.15. The total volume of eroded debris from Hokkaido-Iburi catchments based on this new scaling relationship is estimated as 64-72 million m3. Based on the GNSS data approximation, we noticed that the co-seismic uplift volume is smaller than the eroded volume, suggesting that frequent large earthquakes (and rainfall extremes) may be counterbalancing the topographic uplift through erosion by landslides, especially in humid landscapes such as Japan, where soil properties are rather weak.

Combining geophysics, remote sensing and numerical simulation to assess GLOFs: Case study of the Namulacuo Lake in the Southeastern Tibetan Plateau.

The current highest glacial lake outburst floods (GLOFs) risk level is centered in the eastern Himalaya. GLOFs represent a serious threat to downstream inhabitants and ecological environment. In the context of climate warming on the Tibetan Plateau, such GLOFs will continue or even intensify in the future. Remote sensing and statistical methods are often used to diagnose glacial lakes with the highest outburst probability. These methods are efficient in large-scale glacial lake risk assessment but do not take into consideration the complexity of specific glacial lake dynamics and triggering factor uncertainty. Therefore, we explored a novel approach to integrate geophysics, remote sensing, and numerical simulation in glacial lake and GLOF disaster chain assessments. In particular, geophysical techniques are rarely applied to the exploration of glacial lakes. The Namulacuo Lake located in the southeastern Tibetan Plateau is considered as the experimental site. The current status of the lake, including landform construction and identifying potential triggering factors, was first investigated. Secondly, the outburst process and disaster chain effect were evaluated by numerical simulation based on the multi-phase modeling frame proposed by Pudasaini and Mergili (2019) implemented in the open source computational tool r.avaflow. The results allowed verifying that the Namulacuo Lake dam was a landslide dam with an obvious layered structure. Also, the piping-induced flood might have more severe consequences than the short-term ultra-high discharge flood caused by surge. The blocking event caused by a surge disappeared faster than that caused by piping. Therefore, this comprehensive diagnostic approach can assist GLOF researchers to increase their understanding of key challenges they are facing regarding GLOF mechanisms.

Imminent threat of rock-ice avalanches in High Mountain Asia.

Upsurge of glacier-related hazards in High Mountain Asia (HMA) has been evident in recent years due to global warming. While many glacial-related hazards are instantaneous, some large landslides were preceded by slow gravitational deformation, which can be predicted to evade catastrophes. Here, we present robust evidence of historical deformation in 2021 Chamoli rock-ice avalanche of Himalaya using space imaging techniques. Multi-temporal satellite data provide evidence of a precursor event in 2016 and expansion of a linear fracture along joint planes, indicating 2021 rock-ice avalanche is a retrogressive wedge failure. The deformation history shows that the fracture propagated at a velocity of ~0.07 m day-1 until September 2020, and with an accelerated velocity (~0.14 m day-1 on average) lately. Analysis of recent similar cases in HMA supported our inference on global warming-induced glacier retreat and thermomechanical effects in enhancing the weakening of fractured rock masses in tectonically active mountain belts. Recent advances in Earth observation and seismic monitoring system can offer clues to the location and timing of impending catastrophic failures in high mountain regions.

Surface temperature controls the pattern of post-earthquake landslide activity.

The patterns and controls of the transient enhanced landsliding that follows strong earthquakes remain elusive. Geostatistical models can provide clues on the underlying processes by identifying relationships with a number of physical variables. These models do not typically consider thermal information, even though temperature is known to affect the hydro-mechanical behavior of geomaterials, which, in turn, controls slope stability. Here, we develop a slope unit-based multitemporal susceptibility model for the epicentral region of the 2008 Wenchuan earthquake to explore how land surface temperature (LST) relates to landslide patterns over time. We find that LST can explain post-earthquake landsliding while it has no visible effect on the coseismic scene, which is dominated by the strong shaking. Specifically, as the landscape progressively recovers and landslide rates decay to pre-earthquake levels, a positive relationship between LST and landslide persistence emerges. This seems consistent with the action of healing processes, capable of restoring the thermal sensitivity of the slope material after the seismic disturbance. Although analyses in other contexts (not necessarily seismic) are warranted, we advocate for the inclusion of thermal information in geostatistical modeling as it can help form a more physically consistent picture of slope stability controls.

Unraveling the drivers of intensified landslide regimes in Western Ghats, India.

The Western Ghats (WG) mountain range in the Indian sub-continent is a biodiversity hotspot, now faces a severe threat to the valley population and ecosystem because of changing rainfall patterns and land-use changes. Here, we use the 2018-2019 landslide inventory data together with various geo-environmental factors and show that the landslide activity in the WG region is amplified by anthropogenic disturbances. We applied a generalized feature selection algorithm and a random forest susceptibility model to demonstrate the major topographic controls of landslides and the risk associated with them in the WG region. Our results show that road cutting and slopes modified to plantations are the strongest environmental variable (50% - 73% within 300 m buffer distance) related to the landslide patterns, whereas short-duration intense precipitation in the high elevated terrain, profile concavity, and stream power contributed to the initiation of landslides. The susceptibility models made for the present, and Global Climate Models (GCM) under the representative concentration pathway (RCP) 8.5 scenario predicts the vulnerable nature of WG for future climate extremes. Our results highlight the impacts of Anthropocene hazards and sensitivity of the WG ecosystem, and a greater focus therefore should be placed to reduce the vulnerability and increase preparedness for future events.