PhD Defence Hakan Tanyas

rapid assessment of earthquake-induced landslides

Hakan Tanyas is a PhD student in the Department of Earth Systems Analysis. His supervisors are dr. C.J. van Westen and prof.dr. V.G. Jetten from the Faculty of Geo-Information Science and Earth Observation.

Earthquakes may cause severe impacts in both urban and rural areas, especially in less-developed countries, due to inadequate spatial planning and building control. In seismically active mountainous regions, the impact of seismic shaking is aggravated by secondary hazards, of which earthquake-triggered landslides are often the most damaging phenomena. Many studies confirm that earthquake losses due to landslides and related ground failures can be very high.

For reducing earthquake disaster losses in mountainous regions, it is important to predict the areas that might be affected by earthquake-induced landslides (EQIL), in order to use this in risk management. It is also very important to predict if landslides may be triggered immediately after an earthquake, in order to facilitate the rescue operations, before the landslides can be actually mapped using satellite imagery. This research focuses on the development of methods for rapid assessment of earthquake-induced landslides, based on knowledge obtained from a database of historical events. 

Frequency-area distribution (FAD) of landslides can be used to derive summary statistics regarding an EQIL-event, which could help us to provide valuable information in the emergence response phase. The power-law relation for the FAD of medium and large landslides (e.g., tens to millions of square meters), which has been observed by numerous authors, provides the basis to model the size distribution of landslides and to estimate landslide-event magnitude (). Using a rapid prediction of landslide-event magnitude immediately following an earthquake, we can evaluate the severity of a landslide-event in near real-time. We can estimate the total landslide area and volume based on empirical relations proposed by previous studies.

For many EQIL inventories the FAD of small landslides diverges from the power-law distribution, with a rollover point below which frequencies decrease for smaller landslides. Some studies conclude that this divergence is an artifact of unmapped small landslides due to lack of spatial or temporal resolution; others state that it is caused by the change in the underlying failure process. An explanation for this dilemma is essential both to evaluate the factors controlling FADs of landslides and also power-law scaling, which is a crucial factor regarding both landscape evolution and landslide hazard assessment.

Rapid assessment of the spatial distribution of EQIL could provide valuable information in the emergency response phase. Previous studies proposed global analyses with the aim of predicting EQIL distributions in near real-time. However, all previous studies are based on grid cells as basic mapping units, which do not reflect the physical properties of terrain units and whose size do not match the resolution of existing thematic data at a global scale. Moreover, none of the existing analyses considers sampling balance between different inventories or categorizing the inventories to construct a training set with higher statistical representativeness. Also, most of the previously proposed models are based on a limited number of historical EQIL inventories.

EQIL inventories are essential tools to extend our knowledge of the relationship between earthquakes and the landslides they can trigger. Unfortunately, such inventories are difficult to generate and therefore scarce, and the available ones differ regarding their quality and level of completeness. Moreover, access to existing EQIL inventories is currently difficult because there was no centralized database.

To address these issues, we compiled EQIL inventories from around the globe based on an extensive literature study. The database contains information on 363 landslide-triggering earthquakes and includes 66 digital landslide inventories. To make these data openly available, we created a repository to host the digital inventories that we have permission to redistribute through the U.S. Geological Survey ScienceBase platform. The hope is that it will grow over time as more authors contribute their inventories. We analyzed the distribution of EQIL events by time period and location, more specifically breaking down the distribution by continent, country, and mountain region.

Additionally, we analyzed frequency distributions of EQIL characteristics, such as the approximate area affected by landslides, the total number of landslides, the maximum distance from fault rupture zone, and distance from epicenter when the fault plane location is unknown. For the available digital EQIL inventories, we examined the underlying characteristics of landslide size, topographic slope, roughness, local relief, distance to streams, peak ground acceleration, peak ground velocity, and Modified Mercalli Intensity. We developed an evaluation system to help users assess the suitability of the available inventories for different types of EQIL studies.

Using the compiled inventories, we analyzed the frequency-area distribution (FAD) of EQIL inventories. We developed an updated method for estimating  and its uncertainty that better fits the observations and is more reproducible, robust, and consistent than existing methods. We validated our model by computing  for all of the inventories in our dataset and compared that to the total landslide areas of the inventories. We demonstrated that our method is able to estimate the total landslide area of the events in this larger inventory dataset more successfully than the existing methods.

We proposed a method to predict landslide-event magnitude, using five predictors, both morphometric and seismogenic, which are globally and readily available. These predictors were used within a stepwise linear regression and validated using the leave-one-out technique. We demonstrated that our approach successfully predicts landslide-event magnitude values globally and provided results along with their statistical significance and confidence levels. The proposed approach can provide information globally and in near real-time, by retrieving data from the USGS ShakeMaps, along with topographic and thematic information. The results may provide valuable information regarding landscape evolution processes, landslide hazard assessments and contribute to the rapid emergency response after earthquakes in mountainous terrain.

We also examined the factors controlling the FADs of landslides and propose that the successive slope-failure process is the main reason for the underestimation of small landslides and thus the divergence from a power-law. This reveals that the divergence from the power law is not necessarily attributed to the incompleteness of an inventory. Because of the subjectivity of mapping procedures, the total number of landslides and total landslide areas in inventories differ significantly, and so do the shapes of FADs.

Finally, we developed an improved global statistical model that overcomes the drawbacks of previously developed methods to estimate the probability of the occurrence of EQIL. We used slope units, which are terrain partitions attributed to similar hydrological and geomorphological conditions and to processes that shape natural landscapes. A set of 25 EQIL-events were selected and categorized based on the similarity between causal factors to determine the most relevant training set to predict a given landslide-event. As a result, we developed a specific model for each category. We sampled an equal number of landslide points from each inventory to overcome the dominance of some inventories with large landslide population. We used seven independent thematic variables for both categorizing the inventories and modeling, based on logistic regression. The results show that categorizing landslide-events introduces a remarkable improvement in the modeling performance of many events. The categorization of existing inventories can be applied within any statistical, global approach to earthquake-induced landslide events. The proposed categorization approach and the classification performance can be further improved with the acquisition of new inventory maps.