Once again the field course for master students in risk took place in the beautiful valley of Ubaye this year. During 11 days (end of May – beginning of June 2019), 12 students from environmental risk or geological risk had the opportunity to study the landslide of Lavalette, work on rockwall stability and discover the processes leading to debris flows. We had the pleasure to be hosted at the scientific center of Seolane in Barcelonnette.
Debris flow channel of the Riou Bourdoux torrent
The usual fieldtrip for bachelor students in risk took place in Les Diablerets . About twenty students took part to this course on 1-3 May 2019. The Pont Bourquin landslide was used for a mapping exercise and as a starting point for a risk analysis.
Example of a landslide map produced by students during 1 day mapping
Ryan Kromer is PhD graduate of Queen’s University and a post doctoral researcher at the Colorado School of Mines. He was a visiting PhD student at the University of Lausanne during 2015 and 2016 and is now visiting the Risk group from April to June 2019. During his visit, he will be conducting research on automated monitoring of landslides using terrestrial LiDAR and photogrammetry. The research visit is supported by the Herbette Foundation. Ryan is looking forward to another fruitfull visit with the group.
Active mountain fronts are subject to large scale slope collapses which have the capacity to run long distances on piedmont areas. Over time, fluvial activity and other gravitatory processes can intensively erode and mask primary features related to the collapses. Therefore, to reconstruct the history of their occurrence, further analyses are needed, like sedimentologic analyses. This work focuses on the occurrence of large rock avalanches in the Vinchina region, La Rioja (28°43’27.81” S / 68°00’25.42” W) on the western side of the Famatina range (Argentina). Here, photointerpretation of high-resolution satellite images (Google Earth) allowed us to identify two rock avalanches, with their main scarps at 2575 and 2750 masl. There are no determined absolute ages for these deposits, however by comparing their preservation degree with those dated further north (in similar climatic and landscape dynamics contexts), we can suggest these rock avalanches took place during the Pleistocene.We carried out a fieldwork survey in this remote area, including classical landslide mapping, structural analysis, deposits characterization and sampling. The deposits reach the valley bottom at around 1700 masl. with runouts about 5 km and 5.3 km long. In one of the cases, the morphology of the deposit is well preserved, allowing to accurately reconstruct its extension. However, in the second case, the deposits are strongly eroded by sources draining from the mountain front, therefore further analyses should be done to reconstruct its extension. In addition to morphologic interpretations, a multiscale grainsize analysis was done to di↵erentiate rock avalanches from other hillslope deposits: first 3D surface models of surface plots (5x5m) were built with SfM (structure from motion) photogrammetry; then classical sieving and finally laser grain-size analysis of deposits were performed. Samples were collected on different parts of the slope, but also along cross sections throughout the avalanche deposit. This deposits characterization was combined with results from mapping and image analysis to provide a first description of the sequence and the extension of events related to the evolution of this mountain front. The collected data helped to create a series of propagation models with the software DAN3D, developed by Hungr (2009). We chose a Voellmy rheology for the model with f = 0.10 ± 0.05 and ξ = 100 (m/s2) ± 50 (m/s2) for the rock avalanches and f = 0.10 ± 0.05 and ξ = 500 (m/s2) ± 200 (m/s2) for the debris flow. The results show a good propagation with more dispersion that we can see on the field. Part of the cover on the numerical model is not visible on the field, probably due to erosion and transport having moved the material, resulting in the current landscape.
Author: José Pullarello
Director: Marc-Henri Derron, Michel Jaboyedoff.
External Supervisor: Ivanna Penna.
Director: Prof. Dr. Michel Jaboyedoff
Jury: Dr. Marc-Henri Derron, Dr. Andrea Pedrazzini, Dr. Nicolas Pollet, Prof. Dr. Suren Erkmann
Geological and hydro-geological natural hazards, as landslides and floods, are a threat to many transport networks built in mountainous areas. These risks, that are often small in intensity, are poorly evaluated or unknown.
This doctoral thesis focuses on the characterization and quantification of natural hazards that impact Swiss communication tracks. It also explores various remote sensing techniques dedicated to the survey of areas around traffic lanes.
A database has been created to characterize the many small events that regularly affect roads and railway lines in order to compensate the lack of knowledge. The number of events and their trends – such as their spatio-temporal distribution, the weather, the geology, the direct damages or the types of affected tracks- are thus known at the national level over a period of five years (2012-2016). It shows that a natural event, on average, disrupts the traffic every 2.1 days and that the events occur mostly during the months of June and July, late afternoon. Direct costs were estimated at CHF 4 million per year, with an average cost per event estimated at 23 400 francs.
In order to characterize the approaches to roads and railways, we have developed and tested the photogrammetric technique “on-motion Structure from Motion”, whose cost is reasonable. This remote sensing technic makes possible to obtain colorized and georeferenced 3D point clouds from images taken by four action cameras placed on a moving vehicle. Its accuracy has been evaluated in laboratory conditions and on many sites. It was also compared with seven other traditional surveillance techniques to identify advantages and disadvantages.
This work highlights the impact and consequences of small-scale natural hazards that, taken as a whole, are not negligible for society. In addition, this study demonstrates that low-cost survey technic can compete with more expensive traditional survey techniques.
Keywords: Natural hazards, transportation networks, on-motion Structure for Motion, topographical survey.
François Noël, Teresa Gracchi and Emmanuel Wyser of the Risk Analysis Group (UNIL) went to Barcelonnette (France) from the 23rd to the 27th of September for an amazing experimental work in the field dealing with rock fall trajectories and associated topics (e.g. impact response, energy transfer, etc.), amongst other French researchers from Active Deformation Group of the University of Strasbourg (EOST, leaded by Jean-Philippe Malet) and the IRSTEA Institute of Grenoble (leaded by Franck Bourrier).
Several (30) rocks were thrown down into a short but steep gully while seismic signals (EOST) and high-speed imaging (IRSTEA & UNIL) were acquired. High resolution 3D imagings were acquired thanks to TLS and SfM.
Accelerometers (ISTE & EOST) were also included in few rocky blocks to monitor and acquire data to better understand impact response, angular velocity changes and other exciting data.
In collaboration with the Fondazione Montagna Sicura, Michel Jaboyedoff, Antoine Guerin and Li Fei went to Entrèves (Aosta Valley, Italy) on 23 October 2018 to investigate the 1997 Brenva rockslide scar (3’870 m, Mont-Blanc massif), which reactivated in September 2016. A helicopter flight of about 25 minutes allowed acquiring hundreds of pictures (digital and thermal) of the rock mass in exceptional conditions, as the high mountain was dry in late autumn 2018. A high-resolution Structure-from-Motion model was then generated using these pictures, allowing us to analyze in detail the structural features and rockfall activity of the Sperone della Brenva.
View of the Mt BLanc and Brenva spur
3D point cloud model obtained by photogrammetry
Part of the Risk Analysis Group participated to the International Symposium on Rock Slope Stability 2018 in Chambéry, France, on 13 to 15 November 2018. It was a great opportunity to exchange about many topics related to natural hazards and rockfalls. The group presented two contributions as first authors whose titles are below:
- About the “rockfall” failure hazard assessment. Michel Jaboyedoff et al.
- Coupling 3D rockfall propagation to the spatio-temporal frequency for a realistic rockfall hazard mapping. Cécile D’Almeida et al.
The symposium concluded with very interesting field visits where we did learn about the history of the Mont Granier and its previous rock collapses and about the different instabilities that affects the gorges of Arly and how the local institutions manage them.
The 3rd Virtual Geosciences Conference took place in Kingston (Ontario) on 22-24 August 2018. This conference is at the intersection of geomatics, visualization, computer vision, graphics and gaming, as well as virtual and augmented reality with applications to a range of geoscience subfields, such as geological mapping, geomorphology, geohazards, glaciology, volcanology, tunnelling, and mining. It was organized at Queen’s University, by Ryan Kromer, a former PhD student of Lausanne and Queen’s universities.
MH Derron, S. Buckley, J. Chandler, M. Jaboyedoff and R. Kromer (Chairman of the 3rd VGC)
The first VGC conference was in Lausanne in 2014 and then in Bergen in 2016. These events are fantastic opportunities to learn how new technologies can be used in geosciences, gathering together people from different horizons.
Optimizing the use of 3D point clouds data for a better analysis and communication of 3D results. François Noël, Marc-Henri Derron, Michel Jaboyedoff, Catherine Cloutier Jacques Locat
Infrared Thermal Imaging for Rock Slope Investigation – Potential and Issues. Marc-Henri Derron, Antoine Guérin, Michel Jaboyedoff
On contrary to hazards which have defined return time for establishing natural hazard maps (for example rock falls or floods), there is no similar methodology for shallow and spontaneous landslides. One way to improve the current methodology is proposed by Cedric Meier, Marc-Henri Derron, Michel Jaboyedoff from RISK-UNIL and Christian Gerber, Veronica Artigue and Melanie Pigeon from the Vaud county administration. It includes the definition of 7 pilots zones based in Jura, Plateau and Alps, on riverbanks or mountain slopes. Based on the new airborne LiDAR acquisition, a former inventory from 1889 to 2013 and basics documents such as geological and topographical maps, air photos, about 110 landslides were registered.
Shallow landslide in the area of Ollon in 2018
The parameters of the source zone of the landslide, like length, width, estimated depth, area, slope angle and propagation angle (Farböschung) were recorded. For each landslide, 3 different volumes (with half-ellipsoid method, elliptical paraboloid method and Sloping Local Base Level or SLBL method, method developed and applied currently at the Institute of Earth Sciences, ISTE – UNIL) were calculated. A volume-frequency distribution, approximated by the Power Law site specific, but also depending on the slope type was developed. Figures showing the probability of the estimated depth or the volume depending on the area of the source zone were also prepared. For the propagation, only 4 % of the landslide have a propagation angle greater than 13°.
Probability of max depth in function of the surface area of the landslide
Jaboyedoff M., & Derron M.-H. 2005: A new method to estimate the infilling of alluvial sediment of glacial valleys using a Sloping Local Base Level, Geogr.Fis.Dinam. Quat., 28, 37-46.
VD (2017) : http://www.geo.vd.ch/theme/dangers_nat_thm