5 December 2011

AGU 1: Landslide characterisation and forecasting

Posted by Dave Petley

My intention in terms of blogging at AGU is to post on those talks and sessions that I find interesting.  This is not going to be comprehensive, and if I miss your talk it does not mean that I didn’t enjoy it or value it.  It is just that there is so much…

So first up this morning was an oral session in landslide characterisation and forecasting, which was wide-ranging and very engaging.  My thoughts went to Susan Cannon of the USGS, who had the misfortune to talk first (8 am on Monday morning), covering debris flow initiation in the aftermath of wildfires in Southern California.  They have classified the landslides according to the intensity of the events (volume and number of debris flows), and then compared this with the intensity / duration relationship for the triggering rainfall event.  They achieved impressive relationship between the two, providing an opportunity to develop a warning system that is potentially useful for decision-makers.  In many ways the most interesting aspect of this is the ways in which it can be actually used by the emergency services, but unfortunately there was less detail than would be ideal.  Sometimes the very short time-slots at AGU (15 mins) are a real constraint.

Next up was a talk by Silvia Peruccacci and her colleagues from CNR-IRPI in Italy, who talked about the development of rainfall intensity – duration thresholds for landslide initiation.  This is a well-trodden path, but pleasingly there was a really new element to this work.  This was the application of a statistical approach to determine the threshold, a significant step forward.  This then allowed objective comparison of thresholds between different areas, between different lithologies in the same area, and between different seasons.  I think there are two immediate upshots from this.  First, it was interesting to note that in their field area in Italy the winter and the summer thresholds were different – quite a surprising result, and one that I am not sure that I understand.  Is this the effect of vegetation I wonder?  Second, the overall thresholds proved to be somewhat lower than other studies in the same area had shown.  This is great work.

Third into bat was Doug Stead and his colleagues from Canada, who talked about mechanisms of rock slope failure as shown by numerical modelling.  Most pleasing about this talk was an emphasis on the importance of the (under-estimated) process of progressive failure, and a recognition of the role of time dependent processes.  They are working to analyse that amazing Cornwall cliff collapse video (which gives me an opportunity to show it again):

Doug reported that they have slowed the video down to generate a 15 minute movie that documents the failure in detail.   That is something I’d like to see.  It is really good to see that a YouTube video is now helping scientific understand of the failure process.

And that was a nice prequel to a talk by Greg Stock and colleagues from the National Park Service of Yosemite and the USGS, which focused on hazard assessment for rockfalls from the cliffs in the canyon.  The National Park is quite small but at times houses as many as 30,000 visitors on a single day – a single ill-timed failure could have serious consequences.  The biggest concern comes from large boulders that fall from high cliffs and then bounce down the talus slopes, which allows them to run out some quite large distances.  Greg outlined the hazard assessment process they have adopted, the results of which are being used to identify buildings at high levels of risk.  Those buildings are being closed.

And speaking of acute rockfall hazard, Mauri McSaveney came into bat on behalf of himself and Chris Massey from GNS Science in New Zealand.  They reported on their hazard assessment work for rockfall events in the Port Hills area of Christchurch in New Zealand.  This is work right at the frontline as the decisions are determining whether individual houses are safe to inhabit. The interesting conclusion from this fascinating work was that many of the houses were at an unacceptable level of risk even before the earthquake.  The potential for further seismicity (in the form of aftershocks) has raised this hazard level.  This is work of the very highest quality.

A really eye-opening talk was then presented by Valentin Gischig and his colleagues from ETH Zurich, who talked about the role of temperature in the initiation of rock slope failures.   Remarkably, they were able to provide compelling evidence that there was a temperature driven effect, resulting in landslide displacements at as deep as 68 m below the surface at the Randa landslide site in Switzerland.  They suggest that this temperature-driven effect may be important in controlling the way that the slope is accumulating strain, and thus progressively weakening.  I think we had always assumed that temperature effects occur at shallow depths; it is surprising and intriguing to discover that this is not the case.

The penultimate speaker was Ping Lu and colleagues from Tongji University in China and the University of Florence.  Over the years I have seen many (maybe even too many) presentations on the use of InSAR for landslide monitoring, but this one had quite a twist.  In this case the study used InSAR data to drive a hazard model by combining displacement data with a susceptibility analysis.  I suspect that this points to what will become a very important research field in the future as InSAR data becomes better and easier to process.  For now it allowed a very interesting estimate of the magnitude of losses from landslides in the Arno River basin in Italy, which came out at 3.22 billion Euros over a 30 year timescale.  Ouch!

Finally, Andreas Gunther from BGR Germany, and various colleagues from across Europe, described the process that they are using to develop a pan-Euriopean first order landslide susceptibility map.  This is really interesting work that feeds straight into the move towards a European Directive on soils.  Perhaps the most impressive element is that they are compiling the European catalogues on landslides – they have over 100,000 landslides in the combined inventory to date – and comparing this with their analysis.  The results are quite encouraging, even though the quality of the data is so patchy.  Their dataset for landslides in Switzerland for example numbers just 280 – an astonishing under-estimate.

So, all in all this was a great session with eight terrific talks.  The poster session walk through is in 20 minutes, so that is where I am heading next.