ResearchPod

Advances in catastrophic slope failure forecasts

ResearchPod

On most days, a weather forecast is a convenient answer to simple questions, like : do I need an umbrella? How windy will it be?  

For high stakes questions around landslide hazards, how do we deal with slope failure forecast? And are there any new tools improving our capabilities without waiting to learn from another disaster?

Andrea Manconi, research scientist at the WSL Institute for Snow and Avalanche Research SLF joins us to discuss new satellite dataand methods to get ahead of the next slope collapse.

Read the original paper: https://doi.org/10.5194/nhess-24-3833-2024

00:00:05 Will Mountford 

Hello. I'm Will. Welcome to researchpod. 

00:00:08 Will Mountford 

For most people on most days, a weather forecast is a convenient answer to: do I need an umbrella? How windy will it be? 

00:00:16 Will Mountford 

Simple questions with simple answers that you can usually trust. 

00:00:21 Will Mountford 

So for high stakes questions around avalanche risk, how much trust would you put in a prediction? What confidence value would convince you that it was safe or not? And is there any way to improve our predictive tools without waiting to learn from another disaster? 

00:00:36 Will Mountford 

Today I'm speaking with Andrea Manconi, research scientist at the Swiss Federal Research Institute WSL, about their techniques and their teams looking to get ahead of the next slope collapse. 

00:00:50 Will Mountford 

Andrea manconi. Hello there. For myself and for everybody listening at home. Could you tell me a bit about you, your work, where it is you're located and what kind of areas your research covers? 

00:00:52 Andrea Manconi 

Hello. 

00:01:01 Andrea Manconi 

Yeah, I'm originally from Italy and specifically from Sardinia. So I studied my master, did my master in engineering there and after that I've been working as a PhD student in Germany, specifically in Potsdam, where there is a big Research Institute about geoscience. 

00:01:18 Andrea Manconi 

And after that I've been working back in Italy in a Research Center is called the National Research Council for about 7 years, and since eight years. Now I'm in Switzerland, working in the framework of the ETH domain. So six years first at the ETH cities in the university, basically working as a lecturer. 

00:01:39 Andrea Manconi  

And researcher. And since three years now I'm working in the WSL, so this is the Federal Institute for Forestry and Snow and Avalanche Research. And I'm based at the moment in Davos. 

00:01:51 Andrea Manconi 

Hike up in the mountains, 1500 meters, where we work mostly on Alpine research in different aspects. 

00:02:00 Will Mountford 

And to think about, I mean just the geography of your research has reflected a very Alpine region. Would you say that the geography that you've grown up in, the influence of nature has led to your research development or have you gone to those places seeking out mountains, slopes and that kind of environment to work in it? 

00:02:18 Andrea Manconi 

Yeah, it's. It's a kind of interesting path. So I've been always attracted by natural environments and also to understanding a little bit how the processes are dominating a little bit, what what we experience daily on the climate and also they say the natural phenomenal natural hazards like earthquakes or volcanic deformation and volcanic. 

00:02:39 Andrea Manconi 

Eruptions. 

00:02:40 Andrea Manconi 

And actually my PhD was more in that direction, so geophysical outserts by earthquakes and volcanic activity. And then I just specialized in natural hazards that are related to mountains, but they was not seeking for that. So there are many colleagues that they are passionated about mountains that they are always trying to work in these kind of environments. 

00:03:00 Andrea Manconi 

For me it was, I wouldn't say by chance. It's still a phenomenon that is really interesting to me, but I was not searching for it. So now I'm focusing on this kind of processes, but from time to time I still work also with former colleagues working on earthquakes and volcanic activity. So my interest is why both in this. 

00:03:20 Will Mountford 

And what is it that makes your research timely? What is it that makes the paper that we're discussing something that is worth bringing attention to now? Kind of what is the the running clock on? 

00:03:31 Andrea Manconi 

It. Yeah, I think it's quite clear that climate change is now attracting quite a lot of research because we are experiencing in the last years. 

00:03:40 Andrea Manconi 

Kind of dramatic change that I think in the everyday life, everybody is somehow experiencing with this extreme events that are usually affecting very delicate areas of of our planet and non. 

00:03:52 Andrea Manconi 

Lengths are actually one of the delicate areas and they are called the Sentinels of the major changes that we can experience in the future years. So in the high elevations we have quite critical conditions that changing, for example, the temperature or having extreme events in terms of precipitation. 

00:04:13 Andrea Manconi 

Or in terms of less precipitation of snow, for example, would modify completely the environment and also the perception that we have in this. 

00:04:24 Andrea Manconi 

Environment. So I think the research we are doing in the Alpine areas. So generally speaking, we talk about Alps usually referring to the European Alps, but this term can be also used to define. So the high mountain areas. So this is usually in research also used just to describe the environment that is above the tree. 

00:04:44 Andrea Manconi 

Line let's say. So this is where glaciers or rock glaciers or. 

00:04:48 Andrea Manconi 

This kind of phenomenon. 

00:04:49 Andrea Manconi 

Occur. 

00:04:55 Will Mountford 

The inciting incident behind the paper that brought your attention to us and our attention to you was centered out of the marmalade as a marmolada malada. 

00:05:04 Andrea Manconi 

Yeah, it's no, it's marmolada. You're you're pronounced correctly. So it's it's, it's the glacier. There's quite a famous glacier in north in Italy. 

00:05:11 Andrea Manconi 

In the Alps and in 2022, there was a big event that also attracted a lot of media. And maybe because there were some victims, usually this is also what happened. We have major events, let's say major failures that are occurring as natural events in many mountain belts. 

00:05:28 Andrea Manconi 

Across the world. 

00:05:29 Andrea Manconi 

But of course this comes to the news. 

00:05:31 Andrea Manconi 

Only when there are some victims or some damages that are affecting infrastructures, for example. And then this topic is secured to the media. 

00:05:40 Andrea Manconi 

And in 2022, there was this event, there was a failure of the glacier in summer period where this area is basically visited by a lot of tourists, hikers that are climbing up towards the top of this mountain and a portion of the glacier collapsed and basically was killing. 

00:06:00 Andrea Manconi 

Some people so during that that event of course I've been, let's say working in different scenarios where this happen. 

00:06:08 Andrea Manconi 

And at that time it was kind of mixed between the emotion. So because it's it was an an event occurring in Italy and of course this touched me, but also I was trying to think what are scientists in this domain and including myself are bringing to trying to avoid these kind of scenarios so. 

00:06:28 Andrea Manconi 

Many times in scientific conferences, when we present our results to our peers, we try really to deliver. So our advances in a form that could be helpful for the people. 

00:06:40 Andrea Manconi 

But I really felt that we had to provide something soon now and not just think at the longer perspective because these kind of events according to what we expect from the climatic modeling might increase over time. So we need to have some proper strategies to deliver solutions. 

00:07:01 Andrea Manconi 

That might be really applied for this kind of scene. 

00:07:04 Will Mountford 

I mentioned about doing something soon and I'd like to follow the idea of time because you know, mountains feel like they are eternal or forever to the bystander, like snow is seasonal every winter, but if we can think about how fast the circumstances on a mountain on a slope can change and how quick predictions or reactions. 

00:07:24 Will Mountford 

And be to see that you know their instant shift of, you know, snow builds up for months and months and months and then in a second an avalanche happens. What's kind of the clock? 

00:07:34 Will Mountford 

Again there. 

00:07:35 Andrea Manconi 

Yeah, this is true. So it's really one of the interesting points that we are trying to understand. So we know that there are some processes that are developing over hundreds or even thousands of years and suddenly or changes that might be really little. Then we experience a failure or a dramatic evolution or a very rapid evolution. 

00:07:56 Andrea Manconi 

Maybe it's not coming to a failure. So and this is actually the part of the research we are trying to develop, understanding the physical processes that are dominating this slow evolution towards a potential catastrophic evolution. 

00:08:10 Andrea Manconi 

It is true that glaciers being always moving, so we know that they are basically displacing all the time in somewhere because of the temperature, we know that some crevices they fade from time to time. According to the dermatologist and on the expectations of the future years. This increase in temperature. 

00:08:31 Andrea Manconi 

That also increase the instability in the high mountains. 

00:08:35 Andrea Manconi 

And might also increase the probability of occurrence of phenomena of this kind, like the marmolada glacier or all the events that have been also experienced in the Himalayas and in other mountain chains. 

00:08:47 Will Mountford 

And a lot of your work and this paper is centered around prediction rather than reaction, because, I mean, it makes sense that if you can prevent catastrophic loss of life or, you know, be there soon enough to rescue people, then that's better than too late. 

00:09:01 Will Mountford 

So to think about running that clock backwards into predictive territory, how do you how does the field at the moment predict slope failure? If we can break things down into a simple language, simple chunks of information and then build up to what your research is putting forward as the next necessary steps. 

00:09:20 Andrea Manconi 

Yeah. We basically look at the mountains like a doctor would look at the human body, let's say. So we try to monitor some parameters. And so for example, the monitor perimeters of the human body would be. 

00:09:32 Andrea Manconi 

Temperature or would be the blood pressure or all the parameters that you could basically analyze to have some quantitative information. The numbers that would be helpful and also in mountains and the slopes. What we can do is to measure these kind of parameters over time and understand whether there might be some changes. 

00:09:52 Andrea Manconi 

One of the parameters that we are used to to control with monitoring systems is the displacement or the deformation that is occurring at. 

00:10:03 Andrea Manconi 

So. 

00:10:04 Andrea Manconi 

This is something that is happening at the surface and it's relatively easy to measure with different kind of instruments. So and the idea is really to monitor the evolution of the displacements overtime to identify if there is an acceleration or if there is a change in the pattern that we would expect for specific seasons. 

00:10:26 Andrea Manconi 

And in case there is an acceleration then react and try to model what is the evolution in the. 

00:10:35 Andrea Manconi 

So I don't know if I can bring some relatively easy mathematical concept. So what we have been working on since the 80s is a very simple model that is called the inverse velocity model. So you basically calculate the inverse of the velocity of measurements that you have of the surface deformation. 

00:10:55 Andrea Manconi 

And probably everybody reminds so not lectures that if you calculate the inverse of number that is a really high really large number. You basically go towards. 

00:11:08 Andrea Manconi 

Zero. It would be smaller and smaller, so the inverse velocity modeling exploits this concept to calculate when we would reach the 0. 

00:11:19 Andrea Manconi 

So meaning that if we have an increase of the velocities overtime, we would approach the zero in a certain time window and we can try to predict when we would reach this. 

00:11:31 Andrea Manconi 

You know. 

00:11:32 Will Mountford 

It's interesting to think of, you know, being able to pin a countdown clock to something. Is it as easy as you know, putting a little stick in the snow and waiting until the timer hits zero? What's the necessary technology here? 

00:11:44 Andrea Manconi 

Yeah. So there are several. 

00:11:46 Andrea Manconi 

So the displacement can be measured with different techniques, so there are some techniques that are putting just points that can be measured over time through, for example cameras. So standard photo cameras. This would be one possibility, and then measure the change of position over. 

00:12:04 Andrea Manconi 

Time can be with the GPS technology also having points that are also changing their position over time. Another rather common instrument is what is called the tail light or a robotic total station. You probably have seen several times along the roads, some technicians that are making measurements. 

00:12:24 Andrea Manconi 

With some strange priests targets and they are basically measuring the distance between a sensor and the target, and basically having the location of this target, we can use the same principle also to track the evolution of points over time. 

00:12:41 Andrea Manconi 

They are very precise. You can go up to millimeters of deformation, and this is something that is really attractive. But the problem is that happen in the mountains sometimes it's not so logistically feasible to install instruments. So that's the reason why in the last year's remote sensing technology picked up. So I was making the example of cameras. 

00:13:04 Andrea Manconi 

With cameras, they have the disadvantage that they are in the optical domain, so when it's dark during night, you cannot use them when there is a bad atmospheric condition, you cannot. 

00:13:14 Andrea Manconi 

Use them and this would be the same if you have a camera installed on the ground at a certain distance from your monitor target, or if you have a camera or a sensor of the same kind in a satellite. So we would have the same problem. We can just monitor over the daylight and also when we have good atmospheric conditions. 

00:13:35 Andrea Manconi 

So that's the reason why, since almost 30 years, new technology has been dead. 

00:13:40 Andrea Manconi 

Out in the domain of radar. So this basically microwaves that can go through the atmosphere and reach the targets. And this is very convenient because you don't need the source of light to record the information. So it's an. 

00:13:57 Andrea Manconi 

Active. 

00:13:58 Andrea Manconi 

Microwave sensor that is. 

00:14:00 Andrea Manconi 

Sending waves to the ground and then recording the delay that these waves have while traveling across the. Also the atmosphere touching the target and getting back to the. 

00:14:13 Andrea Manconi 

Sensor. 

00:14:14 Andrea Manconi 

And with this principle, with radars we can meet. 

00:14:17 Andrea Manconi 

The displacements of several points. In this case we are talking about the mountains or a slope that is moving, or a glacier that is moving. This is basically what we are using more and more in these years. 

00:14:32 Will Mountford 

I'm going to keep coming back to the theme of time here because I know that if I look at the road that I live on on Google Maps, it's not fair that map hasn't updated yet and it's, you know, buildings have been here for several years. 

00:14:43 Will Mountford 

How quickly are the satellite radar scans that you're working with? Updating? How contemporary is your information? 

00:14:50 Andrea Manconi 

So this is exactly the point. So we we might have a very good spatial resolution. So the possibility really to map an image, some features with a very high curiosity in space. But what is really critical, especially for the topic we were discussing before about prediction is. 

00:15:08 Andrea Manconi 

Timing so meaning that we need some sensors that are passing as often as possible to catch changes and this is exactly what I developed after the event of the marmolada. So I wanted to use satellite data specifically in the reader domain that could bring us. 

00:15:28 Andrea Manconi 

Information. 

00:15:30 Andrea Manconi 

Possibly on a daily basis. So most of the satellites of this kind with radar, they provide you information every several days. So they are ranging between 40 days. So like in the range of months or up to six or ten days, one week, four days, but 4 rapid evolutions. 

00:15:50 Andrea Manconi 

Like what we have seen in events like the MARMOLADA, we need really to have updates at least. 

00:15:57 Andrea Manconi 

Really. 

00:15:58 Andrea Manconi 

This would be really an important achievement to try to predict what is happening over the next days. 

00:16:05 Will Mountford 

And this is somewhere that something like even civilian domestic drone technology is stepping up to fill that gap. 

00:16:13 Andrea Manconi 

Yeah. So there are some experiments that are also based on drones and the civilian techniques, of course, in the military domain, this technology is being used and is quite reliable already for this. 

00:16:24 Andrea Manconi 

Which is what is this video domain we have since few years. Also commercial satellites that are providing these data sets timely. So this is because the technology advanced a lot in the last years. The cost also for shipping supplies is much lower compared to the past. So meaning that we are now. 

00:16:44 Andrea Manconi 

Experiencing what is called the Golden Age of remote sensing and specifically of radar satellites, because we are starting having not just one satellite but the constellation of satellites. 

00:16:57 Andrea Manconi 

And this is allowing so the constellation is increasing the probability that one of these satellites is acquiring information over our target globally. So that's the main difference compared. 

00:17:09 Andrea Manconi 

To the past. 

00:17:11 Will Mountford 

Now there is some part of me that wants to weigh those advantages against there's got to be some disadvantage, like if the the amount of emissions for any rocket launch then contribute more to global warming, or if you're doing daily flowers with a helicopter. And that comes with emissions as well. Is there any trade off in terms of downsides or is it all just? 

00:17:31 Will Mountford 

Better and better as a golden age. 

00:17:33 Andrea Manconi 

Yeah, it's always one of these issues you have to balance a little bit for this, the advantage that you are bringing with this new technology and what would be the disadvantage, I'd never thought about doing a balance. So like a CO2 balance of what would be the benefit for for my research activity or for the acquisition of launching multiple. 

00:17:53 Andrea Manconi 

Utilized but for sure we are also attentive to this kind of problem. 

00:17:57 Andrea Manconi 

Now and remote sensing based on satellites is probably from this point of view less costly compared to helicopter flies or just fly with drones as well. Just because you can cover much bigger areas with a single acquisition. So you can cover several kilometers where you can cover. 

00:18:18 Andrea Manconi 

Globally, while if you would have to do the same with helicopter flights would be just a completely different amount of emissions that you would have to count. 

00:18:28 Will Mountford 

Well, there was a case study in your paper of putting some of these advances. Some of these practicalities in the field greens brinjals am I getting the pronunciation? They're correct. 

00:18:37 Andrea Manconi 

Yeah, prints is basically is a slope that is located close by to the holes where we are based and we are working since several years in this area is also an interesting case scenario because there is a village on top of the slope, so this is quite well known that there is a landslide going on there since many many years. 

00:18:57 Andrea Manconi 

And also in I think 8064 there was already some an evacuation of the village. 

00:19:04 Andrea Manconi 

Because of an acceleration of parts of this slope and recently we experienced also this kind of acceleration again of portion of this slope and that was basically the connection between my thoughts after the Marmolada glacier collapse where we basically arrived after collapse. 

00:19:24 Andrea Manconi 

While here in Prince we we experienced this acceleration and we could test some of these techniques that I mentioned with the radar and with the possibly a daily coverage with this technology. 

00:19:36 Andrea Manconi 

In an area that had also a similar health potential, so we acquired a lot of this imagery with radar overprints and in June last year. So in 23, there was finally a collapse of a portion of this slope. 

00:19:54 Andrea Manconi 

So what we basically have done is to process the data in a timely fashion. So that's showing that it's really possible to process the data almost in near real time. 

00:20:05 Andrea Manconi 

Time and provide information that would compare well with ground based sensors. So one of the issues with this new technologies is always that you have to validate their performance to be comparable with all the technologies that are there since many many years and they are reliable. We did this experiment and it was successful. 

00:20:26 Andrea Manconi 

Or the prediction capability that we had was similar to what you could achieve with ground based monitors. 

00:20:34 Andrea Manconi 

This means that in cases like brilliance, where we have full control of what is happening, we can validate this technique, and now we can try to approach with the same methodology. Also, all the slopes where we don't have full control, like here in brilliance or all the slopes in Switzerland or in Europe in general. 

00:20:54 Andrea Manconi 

Where we can reach relatively easily the slopes and we can also install additional. 

00:21:00 Andrea Manconi 

Answers, but you can imagine in places like in the Himalayas, where the size of these phenomena can be much bigger and also the vulnerability of the populations and of the infrastructure is a completely different level. If we can afford doing predictions by using remote sensing data. 

00:21:20 Andrea Manconi 

Instead of going there, she's logistically quite, quite complicated and also very expensive. 

00:21:26 Andrea Manconi 

If then we have the chance also to use this for predicting the evolution and maybe just trying to avoid these kind of disasters occur, or at least mitigate their impact on the population because in some cases there's nothing you can do. You can just warn people. 

00:21:46 Andrea Manconi 

Warned the authorities. But I mean, in the end what we are really keen on is is saving life if if possible, then of course infrastructures, you cannot move them if they are in a location that might be affected by it. 

00:22:01 Will Mountford 

Well, is there anything that you see in maybe future developments or I mean the big trend at the moment is to say what can AI do for this in terms of data analysis, monitoring and kind of the processing of the huge amounts of information going in and ratifying stratifying information going out. How do you think this could? 

00:22:20 Will Mountford 

Improve either response times or predictive power in those ideal situations in Europe, in the UPS like you say, compared to places like the Himalayas. 

00:22:29 Andrea Manconi 

I mean, one of the issues is of course that by having an increasing amount of data, we need to have some model source strategies to to process data and to really interpret this data in a timely fashion. And this is something that we are working on. So we have several projects that are also trying to develop methods in the AI domain to help us. 

00:22:49 Andrea Manconi 

In the mapping and interpretation of this kind of phenomena. 

00:22:54 Andrea Manconi 

On the other hand, I think there's usually also what I say to my students when we have lectures in this domain. One of the problems we have in remote sensing and in natural hazards in general is that we don't have sometimes enough data to train models and to train the machines to provide us the right answer. 

00:23:13 Andrea Manconi 

Meaning that we still have to learn a lot before teaching the machine. How to help us. 

00:23:20 Andrea Manconi 

Future. So at the moment we are still acquiring a lot of data and this would be for sure in the next year beneficial for the AI development in our domain science. But we are not yet at the point that we can really develop in AI model that could be really helpful for the interpretation and maybe the prediction. 

00:23:41 Andrea Manconi 

On where how big and when a new failure will occur? 

00:23:51 Will Mountford 

To keep coming back to the theme of time, we've talked about the past of developments and the present of your research and the future of where this might go, and we are now coming to the end of our time with you here today. So if anyone listening to this who is maybe interested in. 

00:24:07 Will Mountford 

Or research or in climate change, and in the glacial shift, what would you say would be the key take away for a member of the public? Let's say someone who is living on the slopes in Switzerland or in Italy and is worried about their future, or a different take away, perhaps for a politician, someone who would be in place to. 

00:24:24 Will Mountford 

To direct people in that disaster response. 

00:24:28 Andrea Manconi 

Yeah, this is a very important point because in many cases the products that we have are scientific papers that are delivered in the scientific community. And so there is a little distance between what we deliver in the scientific community and what actually the people are really taking as a message. 

00:24:45 Andrea Manconi 

So we are now also trying and I think this podcast today is also part of this process we're trying to deliver this information also in a way that the people that are not involved in the scientific process, they could also pick up these concepts and we have some papers or some products that might be. 

00:25:06 Andrea Manconi 

More accessible also to people without having the scientific background. 

00:25:09 Andrea Manconi 

And usually I'm talking now about the Swiss level. So there are the contents and the federal government that they are producing also some information that is really useful in terms of guidelines. So I think this is really now also part of the job that scientists have to do in cooperation with politicians and also other. 

00:25:29 Andrea Manconi 

Factors. 

00:25:30 Andrea Manconi 

Try to deliver this information in an easier way and also to avoid an excess of panic. This is also something that is really important, so we are working on trying to better understanding what is happening in the high months. We are not yet at the point that we can provide an interpretation to this final. 

00:25:50 Andrea Manconi 

We are not capable of predicting what, where and how big. As I said before and when something would occur. 

00:25:58 Andrea Manconi 

But we are trying to go in the direction of, like the meteorological predictions. So I mean everyone accepts that sometimes these predictions might be not accurate enough. So having more data, having more information will help us to go towards this potential prediction of large failures also in the mountains. 

00:26:19 Will Mountford 

And on a personal level for you, where can people find more from you, either on social media or on a university page to follow your work? 

00:26:27 Andrea Manconi 

So there is a website of the WSL which is www.wsl.ch door in this web page you can find a lot of information in different languages. 

00:26:40 Andrea Manconi 

About several projects we have in the research domain and most applied projects. 

00:26:45 Andrea Manconi 

So there is a specific web page also for all the research groups so that people can find out also specific information about the projects I'm working. 

00:26:54 Andrea Manconi 

And so I'm personally delivering most of my social media in LinkedIn, not in any other platform. So that's basically where people can find some more information, more updated information on my research. 

00:27:08 Will Mountford 

Thank you so much for your time, Doctor Manconi and all the best with hopefully getting ahead of those predictors and forecasts soon. 

00:27:14 Andrea Manconi 

Thank you very much.