GeoLog

mapping

GeoTalk: A smart way to map earthquake impact

GeoTalk: A smart way to map earthquake impact

Last week at the 2016 General Assembly Sara, one of the EGU’s press assistants, had the opportunity to speak to Koen Van Noten about his research into how crowdsourcing can be used to find out more about where earthquakes have the biggest impact at the surface.

Firstly, can you tell me a little about yourself?

I did a PhD in structural geology at KULeuven and, after I finished, I started to work at the Royal Observatory of Belgium. What I do now is try to understand when people feel an earthquake, why they can feel it, how far away from the source they can feel it, if local geology affects the way people feel it and what the dynamics behind it all are.

How do you gather this information?

People can go online and fill in a ‘Did You Feel It?’ questionnaire about their experience. In the US it’s well organised because the USGS manages this system in whole of the US. In Europe we have so many institutions, so many countries, so many languages that almost every nation has its own questionnaire and sometimes there are many inquiries in only one country. This is good locally because information about a local earthquake is provided in the language of that country, but if you have a larger one that crosses all the borders of different countries then you have a problem. Earthquakes don’t stop at political borders; you have to somehow merge all the enquiries. That’s what I’m trying to do now.

European institutes that provide an online "Did You Feel the Earthquake?" inquiry. (Credit: Koen Van Noten)

European institutes that provide an online “Did You Feel the Earthquake?” inquiry. (Credit: Koen Van Noten)

There are lots of these databases around the world, how do you combine them to create something meaningful?

You first have to ask the different institutions if you can use their datasets, that’s crucial – am I allowed to work on it? And then find a method to merge all this information so that you can do science with it.

You have institutions that capture global data and also local networks. They have slightly different questions but the science behind them is very similar. The questions are quite specific, for instance “were you in a moving vehicle?” If you answer yes then of course the intensity of the earthquake has to be larger than one felt by somebody who was just standing outside doing nothing and barely felt the earthquake. You can work out that the first guy was really close to the epicentre and the other guy was probably very far, or that the earthquake wasn’t very big.

Usually intensities are shown in community maps. To merge all answers of all institutes, I avoid the inhomogeneous community maps. Instead I use 100 km2 grid cell maps and assign an intensity to every grid cell.. This makes the felt effect easy to read and allows you to plot data without giving away personal details of any people that responded. Institutes do not always provide a detailed location, but in a grid cell the precise location doesn’t matter. It’s a solution to the problem of merging databases within Europe and also globally.

Underlying geology can have a huge impact on how an earthquake is felt.  Credit: Koen Van Noten.

Underlying geology can have a huge impact on how an earthquake is felt. 2011 Goch ML 4.3 earthquake.  Credit: Koen Van Noten.

What information can you gain from using these devices?

If you make this graph for a few earthquakes, you can map the decay in shaking intensity in a certain region. I’m trying to understand how the local geology affects these kinds of maps. Somebody living on thick pile of sands, several kilometres above the hypocentre, won’t feel it because the sands will attenuate the earthquake. They are safe from it. However, if they’re directly on the bedrock, but further from the epicentre, they may still feel it because the seismic waves propagate fast through bedrock and aren’t attenuated.

What’s more, you can compare recent earthquakes with ones that happened 200 years ago at the same place. Historical seismologists map earthquake effects that happened years ago from a time when no instrumentation existed, purely based on old personal reports and journal papers. Of course the amount of data points isn’t as dense as now, but even that works.

Can questionnaires be used as a substitute for more advanced methods in areas that are poorly monitored?

Every person is a seismometer. In poorly instrumented regions it’s the perfect way to map an earthquake. The only thing it depends on is population density. For Europe it’s fine, you have a lot of cities, but you can have problems in places that aren’t so densely populated.

Can you use your method to disseminate information as well as gather it, say for education?

The more answers you get, the better the map will be. Intensity maps are easier to understand by communities and the media because they show the distribution of how people felt it, rather than a seismogram, which can be difficult to interpret.

What advice would you give to another researcher wanting to use crowd-sourced information in their research?

First get the word out. Because it’s crowd-sourced, they need to be warned that it does exist. Test your system before you go online, make sure you know what’s out there first and collaborate. Collaborating across borders is the most important thing to do.

Interview by Sara Mynott, EGU Press Assistant and PhD student at Plymouth University.

Koen presented his work at the EGU General Assembly in Vienna. Find out more about it here.

GeoSciences Column: Mapping floods with social media

GeoSciences Column: Mapping floods with social media

Picture this: you are on your commute home, smartphone or tablet in hand, surfing the internet. You might quickly catch up on the latest news, check in with your friend’s on Facebook, or take to Twitter to share a morsel of information with your followers.

This scenario is common in the modern era of technology. No doubt we are all guilty of indulging in a serious session of internet navigation every now and then (and nothing wrong with that!). But what if your online persona could also make a contribution to better natural disaster management?

One of the many challenges during, and in the immediate aftermath of, natural disasters is being able to provide local populations with timely and reliable information about the extent of damage and/or disruption expected. Flooding events are a prime example: minimising and managing the financial, human and emotional cost of floods is key for researchers, local communities, policy makers and authorities alike.

Contributing to this effort, a team of German researchers have designed a tool which harnesses our desire to share snippets of our lives via social media to support the creation of rapid inundation maps during flooding events. The research was recently published in the EGU open access journal, Natural Hazards and Earth System Sciences.

Currently, measurements of flood water heights made by river gauges, hydrodynamic-numerical models and remote sensing data – such as before and after images acquired by satellites – are used to create rapid response flood maps. Despite their successful and wide-spread use, they are not without limitations.  River gauges only allow for narrow point information on water heights during a flood and require detailed topographical data to be validated. Hydronamic-numerical models aren’t very flexible: it is difficult to build unforeseen incidents into them (e.g. a dike breach). Remote sensing techniques have limitations when it comes to providing real time information; it can take up to 48 hours for the images to be delivered and processed before they can be used.

The study authors argue that eyewitness information about flooding events shared via social media can fill in some of the gaps. Using quantitative data, such as geographical location and flood water height, held in images shared via Twitter and Flickr, can provide information to make more detailed and accurate flood maps in almost real-time. The researchers put the theory to the test for the June 2013 Dresden floods.

The city of Dresden, with its 800,000 inhabitants, sits on the banks of the River Elbe, known for its long history of flooding. This means the city’s population is more aware of the hazard and, being an urban area, likely has a large number of social media users, making it a good case study candidate.

Location of useful photos retrieved with PostDistiller and inundation depths estimates. (Photos by Denny Tumlirsch (@Flitz- patrick), @ubahnverleih, Sven Wernicke (@SvenWernicke) and Leo Käßner (@leokaesner). Taken from J. Fohringer et al. (2016))

Location of useful photos retrieved with PostDistiller and inundation depths estimates. (Photos by Denny Tumlirsch (@Flitz-patrick), @ubahnverleih, Sven Wernicke (@SvenWernicke) and Leo Käßner (@leokaesner). For instance, photos 1 and 2 show inundated roads but a dry sidewalk. This context en- ables the analyst to estimate inundation depth in the order of approximately 5 cm Taken from J. Fohringer et al. (2016))

The research team created an inundation map using only information from photos filtered from Twitter and Flickr. To collate the flood data from social media, the team designed a computer programme. In the first instance a search for key words (in both English and German) related to floods was ran: “Hochwasser”, “Flut”, “Flood”, “inundation”, to name a few. The results were then filtered by the time frame of interest (from May 5th to 21st June 2013) as well as the geolocation of the posts. This yielded a total of 84 posts from which five inundation depths were derived (see the figure caption for details of how the team achieved this), in the space of no more than four hours. The depths calculated were then used to create the inundation map.

To test the robustness of the map, the team created a second map relying only on online data acquired from the Dresden river gauge. Comparing the two maps shows that the social media created map overestimates inundation height by decimetres as well as the geographical extent of the flooding. Despite that, the study authors argue that the errors are acceptable when it comes to providing rapid inundation maps, particularly in situations when no other information is available.

Inundation maps and inundation depths derived from online water level observations (a) and social media content (b) ; inundated area derived from the reference remote sensing flood footprint (c) ; and differences between inundation depths for overlapping areas in scenarios (a) and (b) (panel d ). J. Fohringer et al. (2016))

Inundation maps and inundation depths derived from online water level observations (a) and social media content (b); inundated area derived from the reference remote sensing flood footprint (c); and differences between inundation depths for overlapping areas in scenarios (a) and (b) (panel d). J. Fohringer et al. (2016))

The case study also highlighted some of the method’s shortcomings. It will be important to improve the vertical and horizontal accuracy of the social media created maps by supplementing them with more detailed topographical terrain data. The current method of acquiring data via social media is relatively passive and relies on users sharing images from a flooding event. Crowdsourcing data, where citizens are actively encouraged to share images, would improve the reliability of the data as well as the spatial coverage.

So when you next take a selfie or capture a stunning landscape to share on social media, who knows, the data held in your images and geolocation could have even more value than you might have originally thought!

By Laura Roberts Artal, EGU Communications Officer

 

 

 References

Fohringer, J., Dransch, D., Kreibich, H., and Schröter, K.: Social media as an information source for rapid flood inundation mapping, Nat. Hazards Earth Syst. Sci., 15, 2725-2738, doi:10.5194/nhess-15-2725-2015, 2015.

Rimkus, S. et al.  A Century of UK Flood Losses (conference abstract) Geophysical Research Abstracts Vol. 18, EGU2016-11905, 2016, EGU General Assembly 2016

Trejo Rangel, M.A., et al. How Can Flood Affect the Real Estate Market? (conference abstract) Geophysical Research Abstracts, Vol. 18, EGU2016-8977, 2016, EGU General Assembly 2016

Geosciences Column: Do roads mean landslides are more likely?

Geosciences Column: Do roads mean landslides are more likely?

Landslides have been in the news frequently over the past 12 months or so. It’s not surprising considering their devastating consequences and potential impact on nearby communities. Data collected by Dave Petley in his Landslide Blog shows that from January to July 2014 alone, there were 222 landslides that caused loss of life, resulting in 1466 deaths.

A recent paper, in the journal Natural Hazards and Earth System Sciences investigates, what the potential effects of human denudation can have on the occurrence of landslide events. There is no denying that landslide susceptibility has been increased by human activity. Global warming and greater precipitation are key contributing factors to the rise in the number of landslides which occur globally. On a local scale, the building of infrastructure, particularly roads and felling of trees to make way for agriculture are largely to blame for increased numbers of slides and slumps.

Overview of the study area with mean annual precipitation patterns (top panel), and its location in southern Ecuador (lower left panel). Highways Troncal de la Sierra E35 and Transversal Sur E50 extend in the north–south and east–west direction, respectively. The numbers along the street refer to the corresponding geological unit (1: unconsolidated rocks; 2: sedimentary rocks; 3: volcanic rocks; 4: metamorphic rocks; 5: plutonic rocks). The area of the detailed map (lower right panel) will be used as a sample area for the visualization of a predictive map in Fig. 5. Precipitation data are taken from the study of Rollenbeck and Bendix (2001). From Brenning et al., (2015)

Overview of the study area with mean annual precipitation patterns (top panel), and its location in southern Ecuador (lower left panel). Highways Troncal de la Sierra E35 and Transversal Sur E50 extend in the north–south and east–west direction, respectively. The numbers along the street refer to the corresponding geological unit (1: unconsolidated rocks; 2: sedimentary rocks; 3: volcanic rocks; 4: metamorphic rocks; 5: plutonic rocks). Precipitation data are taken from the study of Rollenbeck and Bendix (2001). From Brenning et al., (2015). Click on the image for a larger version.

The research presented in the paper focuses on landslides along mountain roads in Ecuador, where drainage systems and stabilisation of hillsides is often inadequate and is known to increase the likelihood of landslides. This problem is not exclusive to Ecuador and is often linked to poorer infrastructure and engineering in developing countries. In addition, the study area is a tropical mountain ecosystem, which is naturally more sensitive and prone to landslides. The key question here being: are more landslides likely to happen close to a road (in this particular case an interurban highways), or does greater distance from them offer some hazard relief?

The geology, and local climate and vegetation are important factors to also take into consideration when carrying out an assessment of this nature. Highways E35 and E50 run along Southern Ecuador and intersect the Cordillera Real, which creates a strong local climate divide and generates a precipitation gradient along the area studied. Páramo ecosystems are dominant towards the east, whilst tropical dry forests are common in the west. The geology is also variable across the area studied: dipping and jointed metamorphic rocks are dominant, but are in contact with horizontally layered sedimentary units of loose conglomerates and sandstones. Additionally, the hill sides running along the highways are often deforested to make way for coffee, sugar cane and banana crops. When they are not, they are commonly handed over to cattle for grazing.

By mapping, in great detail, all landslide occurrences within a 300m corridor along the highways, the researchers were able to digitise 2185 landslide initiation points! In total, 843 landslides were mapped and classified by recording the type of movement experienced, as well as the material type (soil, debris or rock) and whether the slide was still active, inactive or had been reactivated. The detailed data meant it was possible to statistically model the likelihood of landslides occurring in close proximity to the highway (25m) vs. some distance away (200m). The results showed that susceptibility to landslides increases by one order of magnitude closer to the highway when compared to areas between 150-300 m away from the mountain road. Furthermore, slides close to the highway were found to be more likely to be reactivated than those a greater distance away.

The study found that the local topography, geology and climate conditions had a lesser influence on the likelihood of landslides. However, the influence of stretches of mountain road constructed in the sedimentary units seems to enhance the hazard.

Landslides occurring along the investigated highways. (a) Typical landslides of the wet metamorphic part of the study area in the east. (b) Typical landslides of the semi-arid, conglomeratic part of the study area in the west. (c) Highway destroyed by landsliding. (d) A highway is cleared from a recent landslide occurrence. From Brenning et al., (2015).

Landslides occurring along the investigated highways. (a) Typical landslides of the wet metamorphic part of the study area in the
east. (b) Typical landslides of the semi-arid, conglomeratic part of the study area in the west. (c) Highway destroyed by landsliding. (d) A
highway is cleared from a recent landslide occurrence. From Brenning et al., (2015).

In future, the model can be used to predict locations where landslides are more likely to occur along the E35 and E50. Recently, engineering works have been carried out along the studied stretch of highways to stabilise the hillsides. The data collected as part of the research presented in the paper will be useful in the future to monitor the efficacy of the improvements. On a larger scale, further studies of this type could be used by local governments when planning new infrastructure and could lead to incorporation of cost-effective mitigation measures in new developments.

 

By Laura Roberts Artal, EGU Communications Officer

Reference:

Brenning, A., Schwinn, M., Ruiz-Páez, A. P., and Muenchow, J.: Landslide susceptibility near highways is increased by 1 order of magnitude in the Andes of southern Ecuador, Loja province, Nat. Hazards Earth Syst. Sci., 15, 45-57, doi:10.5194/nhess-15-45-2015, 2015.

GeoTalk: Steven Smith on fossil faults and fantastic faulting

This week in GeoTalk, we’re talking to Steven Smith, a Lecturer from the University of Otago. Steven takes us on an Earth-shaking journey, explaining how ancient faults reveal what’s happening under the Earth’s surface and delving into the future of fault zone research.

First, could you introduce yourself and tell us a little about what you are currently working on?

Last September I started as a Lecturer at the University of Otago in New Zealand. My work focuses on understanding the structure and evolution of tectonic fault zones in the continental crust. I graduated from Durham University (England) back in 2005 and remained there to complete my PhD, studying a large fault zone exposed on Italy’s Island of Elba. My Italian connection continued as I then moved to Rome for 4 years to undertake a post-doc with Giulio Di Toro and Stefan Nielsen at the INGV (National Institute of Geophysics and Volcanology). In Rome I was working on a large group project funded by the European Union. The project involved integrating field and experimental work to better understand some of the extreme deformation processes that occur in fault zones during earthquakes. When my post-doc finished, I relocated to New Zealand, attracted by the research opportunities available here and the wonderful places waiting to be explored!

Steve on the summit of Lobbia Alta (3,195 m), a peak in the Adamello area of the Italian Alps. This area is part of a large Tertiary magmatic batholith that is cut by several fault zones Steve studied as part of his post-doc work. (Credit: Steven Smith)

Steve on the summit of Lobbia Alta (3,195 m), a peak in the Adamello area of the Italian Alps. This area is part of a large Tertiary magmatic batholith that is cut by several fault zones Steve studied as part of his post-doc work. (Credit: Steven Smith)

Last year, you received a Division Outstanding Young Scientists Award for your work on the structure of shear zones. Could you tell us a bit more about your research in this area?

I work mainly on so-called “exhumed” fault zones – those that were once active and have been brought to the surface of the Earth by millions of years of uplift and erosion. By studying the structure of exhumed fault zones we can understand a lot about the physical and chemical processes that are active when faults slip. I have worked on fault zones in Europe that were exhumed from both the middle and upper crust. I’m particularly interested in the crushed-up rocks in the cores of fault zones (better known as fault rocks) and what these can tell us about fault zone rheology and deformation processes.

In the last few years, we have identified a number of previously-unrecognised textures in the cores of large normal and thrust faults in Italy. These include faults with highly reflective “mirror-like” surfaces and small rounded grains resembling pellets. Our work focused on trying to understand the significance of these textures, and in doing so we had to develop new experimental methods to deform fault zone materials under more realistic conditions. By comparing our field observations to the experimental data, we realised that the textures we identified probably represent ancient “fossil” earthquakes preserved in the rock record.

When thinking of a fault zone, you don’t often think small. How can you scale down what’s going on to create a laboratory-sized experiment?

That’s a very good question, and something that experimentalists have to be aware of all the time. It’s generally not possible to reproduce in the laboratory all of the conditions that a natural fault enjoys, which is why laboratory work ideally has to be integrated with other data, such as field observations of natural faults and theoretical modelling. But it is possible to perform experiments at quite realistic pressures and temperatures, and some deformation apparatus can deform fault zone samples over a very wide range of slip velocities – the sort you would expect during the seismic cycle along natural faults.

Laboratory experiments allow us to produce data on rock strength and fault zone behaviour that simply wouldn’t be possible by any other means. At the same time, it’s important to bear in mind that a small laboratory experiment might represent the behaviour of only a single point on a much larger fault surface – that’s where complementary field and geophysical observations come in. Looking at fault geometry and fault zone evolution over time helps put lab studies into context.

Admiring the inner workings of a fantastic fault on the Italian Island of Elba. (Credit: Bob Holdsworth, Durham University)

Admiring the inner workings of a fantastic fault on the Italian Island of Elba. (Credit: Bob Holdsworth, Durham University)

How does deformation differ across the world’s fault zones?

In the last few years, seismologists have detected different types of slip behaviour along active faults, from those that creep along steadily at rates of a few millimetres per year to those that fail catastrophically in earthquakes. In between there are other types of slip behaviour seen in seismological signals (such as seismic waves) from the deeper parts of fault zones like the San Andreas fault and the Alpine fault in New Zealand. Understanding the basic controls of this rich variety in fault slip behaviour is one of the key goals of modern earthquake science. Geologists studying exhumed fault zones have also long recognised that fault zone structure in the crust is very complex and highly dependent on factors like the type of host rock, level of exhumation, availability of fluids and so on. A future goal for geologists like me is trying to understand whether different types of fault slip behaviour recognised in modern-day seismic signals are preserved in the structure of fault zones in some way.

Is digital mapping an essential tool for the modern-day structural geologist? How does it help?

As in most branches of science, digital technologies are now increasingly used for geological mapping purposes, although a majority of universities still favour the traditional pen-and-paper approach to training students in geological mapping. In many mapping situations – ranging from reconnaissance-scale mapping to detailed surface topography measurements – there are a number of benefits to a digital approach, the main one being that field data are automatically located using GPS or laser-based technology. This opens up a range of possibilities to map and study 3D structures that would be difficult and time-consuming, if not impossible, using more traditional approaches.

I certainly think that digital technologies will play an increasingly important role in geological mapping and research in the future, and the technology will inevitably become more fit-for-purpose and affordable. I recently taught a short workshop on digital mapping to a group of 4th year students in Otago; after a few days they were all quite confident using handheld computers and GPS devices to map, and I think they were impressed by the ease with which the digital data could be analysed to better understand the inherent complexities of geological structure.

Finally, will you be working on faults in the future or moving on to pastures new?

I think there are a lot of important research problems to tackle in fault zones, so it’s likely I’ll continue working on faults in a broad sense in the future. The devastating Tohoku-oki earthquake and tsunami in Japan in 2011 highlighted that our understanding of fault zones is far from complete. My post-doc experience has convinced me that integrating field and experimental work is a promising way to understand fault zone processes. At the moment I’m particularly interested in the very-small scale frictional processes that occur on fault surfaces during earthquakes, and I’m currently using some of the analytical equipment here in Otago to study experimental samples that were deformed under earthquake-like conditions.

If you’d like to suggest a scientist for an interview, please contact Sara Mynott.