Takeaways from the International School on Landslide Risks Assessment and Mitigation (LARAM)

I had the opportunity to attend the International School on Landslide Risk Assessment and Mitigation (LARAM), organized by the Geotechnical Engineering Group (GEG) at the University of Salerno, Italy, from September 9th to 20th, 2024. 39 PhD students and three early-career researchers from 30 universities across Europe and beyond, representing 20 different countries, participated in the event to share their research on landslide disasters. Renowned professors and stakeholders working in landslide disaster risk mitigation from various countries, including Italy, Austria, Norway, China, Switzerland, Spain, the Netherlands, Germany, Serbia, Croatia, and Greece, shared their insights. They discussed and showed tutorials on key topics such as an introduction to landslides, landslide risk theory, landslide modelling, risk analysis and zoning, monitoring and mitigation, risk management, and governance.

Here, I share some key takeaways from LARAM School, with a focus on conceptual insights rather than the detailed physical and mathematical aspects discussed during the sessions.

Human Influence and Landslide Dynamics

Nowadays, landslide disasters are no longer considered purely natural; human activities significantly influence them. Actions such as altering landscapes, unplanned urbanization, deforestation, and climate change (leading to extreme rainfall and temperature shifts) can trigger landslides in various ways. Among climatic factors, unevenly distributed heavy rainfall is a major contributor, as it causes slopes to become oversaturated and increase pore water pressure leading to slope instability. Landslides can be likened to human cancer: just as cancer is triggered by uncontrolled cell growth, landslides occur when soil becomes unstable due to various triggers. Of the many types of landslides, flow-like landslides are the most dangerous, as they can propagate over large areas. Therefore, proper land management is essential to mitigate landslide risks, as these events cause significant disruptions, affect ecosystems, and highlight the importance of early detection and intervention.

Landslides do not occur randomly on hills or mountains but tend to happen near weak points such as cracks, fault lines, joints, and drainage channels. A first failure (landslide) usually occurs at these weak points, triggered by external factors. This initial failure weakens the slope, altering its structure, and creating conditions for secondary failures in the surrounding areas. These secondary landslides increase the susceptibility of the region, leading to a feedback loop that can trigger additional, larger landslides. Hence, geological and geomorphological knowledge is crucial for landslide research.

Monitoring and Case Study

After a landslide, the stratigraphy or the layered structure of the soil and rock in the affected area undergoes significant changes. These alterations can disrupt the stability of the landscape, making previous risk assessments unreliable. This is why it becomes essential to update risk maps based on new geological and geotechnical conditions. For instance, landslides can expose deeper, less stable layers or alter the composition of surface materials, which may now be more prone to further movement.

From a soil science and physical science perspective, it is crucial to assess the presence and behaviour of water tables in the affected zone. Water tables influence slope stability, and multiple water tables at different depths can increase the complexity of the risk. The type of water table (confined or unconfined) determines how water moves through the soil, affecting its weight and cohesion.

Monitoring pore water pressure over time is a key factor in evaluating the soil or rock’s shear strength. Pore water pressure refers to the pressure exerted by water within the pores of the soil. When this pressure increases due to rainfall, groundwater infiltration, or changes in drainage it reduces the friction between soil particles, weakening the overall structure. This reduction in shear strength, combined with increased shear stress from external forces like gravity or added water weight, can trigger landslides or make the area more susceptible to future failures. Understanding these dynamics is critical for accurate hazard assessments and mitigation strategies.

Modelling and Mapping Landslide Risks

Recent scholars have modelled landslides to mitigate risks using three widely employed methods: mathematical, constitutive, and numerical. However, an important step that is often overlooked is the observation of phenomena through community surveys, stakeholder interviews, field visits, laboratory tests, and reviewing existing research. A purely physical science-based model may produce statistically excellent results but may significantly deviate from reality. Therefore, incorporating field observations is essential to make the results more practical and reliable, a principle applicable to other disasters as well.

Landslide inventories are the foundation for modelling landslide disasters. Today, satellite imagery is extensively used to create these inventories, which helps geologists and geomorphologists. However, it remains crucial to cross-check satellite-generated inventories with field visits, as artificial intelligence (AI) used to detect landslides in images is not 100% accurate. Additionally, the nature of landslides varies by location, catchment area, and even slope, making geological knowledge necessary for inventory classification.

There are two main approaches to mapping landslide susceptibility: bottom-up and top-down. It is always recommended to perform a back-analysis of past landslides to calibrate models. The bottom-up approach, although thorough, is time-consuming and impractical for large areas within 3-4 years. AI typically uses this approach, but it would be beneficial for researchers to guide AI in employing top-down methods as well. The top-down approach is efficient for large-scale landslide risk mitigation, allowing researchers to first map a large area and then categorise it into different zones based on homogenous patterns of landslides, geology, hydrology, and other factors. This makes it easier for risk managers to implement similar mitigation strategies across similar zones, offering a more practical and effective solution than the bottom-up approach.

Early Warning Systems and Mitigation

Monitoring landslides and providing early warnings (EW) are vital to reducing landslide risks. Although satellite imagery is widely used for landslide monitoring, it is not ideal for fast-moving landslides (e.g., rock falls), which can occur in minutes, making it impossible to analyse their velocity due to the fixed revisit times of satellites. Multi-hazard monitoring is also essential, as landslides can be triggered by other disasters such as earthquakes, floods, and storms, which compound the overall damage. Developing an effective EW system requires collaboration between academics, hydrologists, geologists, geomorphologists, climatologists, programmers, and stakeholders. Although effective EW systems can reduce landslide risks, they depend largely on the willingness of the affected communities to act on the warnings. Often, people are unwilling to relocate even after receiving an EW, putting their lives in danger. Moreover, early warnings are not always disseminated widely; in Switzerland, for example, only around 30% of affected people receive warnings, with this figure dropping to as low as 2% in some cases.

It is crucial to ensure that the recipient of the EW fully understands the different alert levels and their implications. Clear communication at each stage helps the receiver take appropriate actions based on the level of threat, reducing confusion and improving response effectiveness.

Beyond early warnings, mitigation strategies are crucial. Before implementing any structural mitigation measures, it is wise to consider the long-term consequences of such structures at least 15 years down the line. While a structure may provide an immediate solution to current problems, it could cause more harm than good in the future (long-term). Measures that work perfectly in one area may not be effective in another, even within the same region. It is essential to design mitigation strategies for the unique conditions of each location, using place-specific risk analysis to ensure the most effective approach for local hazards and vulnerabilities. Nature-based solutions, such as planting trees, are effective for all types of landslides, but they may take a long time to yield results.

---

Roquia Salam is a first-year PhD student at UCL RDR, supervised by Dr Bayes Ahmed (primary) and Professor Peter Sammonds (secondary). Her PhD research topic is “Explainable AI-Driven Digital Twin Technologies for Developing Rainfall-Induced Landslide Forecasting Systems in Bangladesh.” Her work explores hazards, capacity, vulnerability, and disasters using both qualitative and quantitative approaches, with a focus on promoting disaster resilience.

---

• Read more RDR Student Blogs
• Follow RDR on Twitter/X and LinkedIn
• Read other RDR Blogs

The views expressed in this blog are those of the author(s).