Volcano monitoring from a distance

In the past few weeks, there has been an eruption that keeps littering my inbox with emails: Bogoslof Volcano, on a tiny island of roughly 1 by 2 km out in the Bering Sea, west of the Alaska Peninsula.

View from a helicopter onto Bogoslof Island. Photo: Dan Leary, Maritime Helicopters

Despite the fact that it's effectively in the middle of nowhere (the nearest town is roughly 100 km away), Bogoslof is an interesting one. Being up in the Aleutian Chain, it sits along a very important corridor for international air traffic. If you remember the chaos all over Europe after the 2010 eruption of Eyjafjallajökull in Iceland, it's hardly surprising that monitoring volcanoes even in parts of the world as remote as Alaska is an important task. But how do you monitor a volcano that sits on an uninhabited, far away island?

An obvious answer would be to put a bunch of instruments onto the island. However, the island is so small, so far away from any population, in such a harsh environment, that the Alaska Volcano Observatory has to focus its limited resources elsewhere. In addition, the last eruption previous to this one had occurred in 1992, and it's been at least 40 years since the last eruption before that, so unsurprisingly the volcano was relatively low on the monitoring priority list.

This changed on 20th December 2016, when several pilots in the area reported an ash cloud that had risen up to over 10 km above sea level. Because there is so much air traffic going through the region, reports like that are an important part of monitoring volcanic activity in remote areas. Whereas the eruption had stopped within an hour or two, activity at the Alaska Volcano Observatory certainly wouldn't have.

Data had to be analysed, statements had to be published and scientists were looking for signs of any unrest that may have preceded the eruption. Indeed, looking back through the data, the volcanologists realised that Bogoslof had been showing signs of activity throughout the month of December, and the first explosion may have occurred as early as 16th December. So what kind of data can volcanologists use to monitor Bogoslof?

Even though there are no seismometers on the island itself, nearby Okmok and Makushin volcanoes have extensive monitoring networks. Because seismometers are very sensitive instruments, and volcanic eruptions make the ground shake with waves that can travel a long way, it is actually possible to look at seismic signals from Bogoslof on other islands.

Similarly, microphones recording "infrasound" (i.e. sound at frequencies much lower than the range we can detect with our ears) can detect pressure signals coming from far away, and volcanic eruptions often produce distinct infrasound.

Satellite image show the ash cloud at Bogoslof Volcano on 18th January 2017. Image: NASA Earth Observatory/Jeff Schmaltz

Satellites are also quite useful. A volcanic ash cloud can often be detected from space. Some satellites capture light of many different wavelengths, others can detect different types of gases in the atmosphere, some of which can be traced back to volcanoes. Visual observations by pilots, local residents or fishermen help to complement the picture we get from satellites.

Last but not least, volcanic lightning (i.e., lightning strikes in or around the ash cloud coming up in an eruption) has been an increasingly valuable tool to detect volcanic eruptions over the last few years. Volcanic lightning is still not fully understood and subject to active study by volcanologists around the world, but even without a complete understanding of the exact mechanism it is a spectacular sight and can be used for eruption detection. You can watch lightning happen all around the world through the World Wide Lightning Location Network if you're interested, almost in real time.

Spectacular eruption with volcanic lightning at Mt. Etna, Italy. Photo: Karl-Ludwig Poggemann

At the time of writing this post, Bogoslof continues to have explosions every few hours to days, and scientists are analysing these eruptions through all the different types of data mentioned above, even though there are no instrument directly on the volcano. Pretty amazing, isn't it?

Volcanic whistles and more

We've been talking about science and being a scientist and girl power for a while, so I reckon it's time to get back to volcanoes. In August 2012 I went to Hawai`i for the AGU Chapman conference on Hawaiian volcanism. The lovely people at the Hawaiian Volcano Observatory (HVO) were nice enough to let me stay for a bit to get familiar with the volcano and to get some of their data to play with. It was also a good opportunity to catch up and collaborate with my friend and colleague Jess Johnson again. My supervisor Mark and I had decided that Kilauea would be a good volcano to study earthquake recordings.

What were we trying to achieve by studying Kilauea Volcano? The main question we had was 

"Do the continuous earthquake recordings look different for different types of eruptions?"

Intuitively one might say, of course there are differences (or at least that's what I would have said). But we wanted some real evidence. Kilauea tends to erupt most of the time, and the style of activity varies, so this was an ideal place for us to go and test our hypothesis. 

I got data from the continuous recordings of earthquake activity around Kilauea volcano for an eruption in the eastern section of the volcano (called the East Rift Zone, close to a crater called Pu'u `O`o) in 2007, and a similar eruption in 2011. These eruptions were related to magma breaking and pushing open a big crack a few kilometres underground. This is called a dike intrusion.

I also got data from a series of more explosive eruptions in the western part of the volcano, close to a crater called Halema`uma`u, in 2008. During the year 2008 these explosive eruptions slowly formed a connection between the magma supply at Kilauea and the surface. Now there is a lava lake that's something like 200 m across, and visitors can no longer access the crater like they used to. The explosions probably only moved magma around that was a few 100 metres below the surface, not kilometres like in 2007 and 2011.

So we had data for two types of eruptions:

fissure eruptions in the east (i.e., lava fountains)

vs. explosive eruptions in the west

We used a technique called Fourier Transforms to find out what frequencies the earthquake waves were composed of at the different points in time (if you want an analogue explanation for what Fourier Transforms do check out this old post). That way we can make what's called a "spectrogram". You can learn a lot of things about the earthquake waves by doing that. For example, you can get an idea if the movement of the ground is from an earthquake that happened close by, or from an earthquake that was hundreds of kilometres away. The graphic below shows the seismic ground movement (the black wiggly thing) and the spectrogram (the rainbow coloured rectangle) for a few days in 2011. I've added some labels to explain in a bit more detail what we can see on it.

Seismic ground movement and "spectrogram" from Kilauea Volcano during an intrusion and fissure eruption. The coloured spectrogram shows how much of each frequency we have in the seismic wave at each point in time. Red means a lot of that frequency, blue means not a so much of that frequency. The red diagonal streaks across the graphic are the volcanic whistles, described below, where the frequency goes up (or down) over time. You can just hear the first one when you listen to the audio from the link in the text below.

So by doing that we learned three main things about eruptions at Kilauea:

1) Fissure eruptions and dike intrusions at Kilauea really do generate ground movement that is different from explosive eruptions.

2) These fissure eruptions and dike intrusions show two phases of ground movement:

Phase I (purple in the graphic above): The first phase is made up of lots of short earthquakes, close to the dike intrusion and the eruption. These earthquakes are probably related to breaking the rocks when the magma pushes open the crack. 

Phase II (blue in the graphic above): The second phase starts a few hours after the first phase. It doesn't have as many short earthquakes, but instead shows continuous (small) movement of the ground for a few days. This continuous movement is what we call "volcanic tremor". Phase II happens quite far away from the eruption and the dike. 

3) The second phase has something called "frequency gliding" (the diagonal streaks in the graphic above). It means that the frequencies of the waves slowly change over time, a bit like a kettle on the stove that starts whistling at a higher and higher tone when the water is boiling.

To give you an idea what I'm talking about I've taken some of the ground movement and sped it up by a lot. You can play the movement that happened over roughly 1 day in just over 1 minute. That way we can actually HEAR the ground move. Click here to listen to the earthquake activity during the 2011 eruption at Kilauea. At first you just hear some noise like the wind. That's before anything is happening. All of a sudden (around 7 seconds in) you start hearing a lot of clicking sounds, maybe like gun shots or like rain drops on a metal roof. Those are the little earthquakes during Phase I. Then it gets a bit quieter again, and then you start hearing something continuous, like a boiling kettle (around 45 seconds). That's the volcanic tremor from Phase II. If you listen really carefully you can even imagine that you're hearing the frequency gliding, i.e., the whistling getting higher and higher.

You may remember a study in 2013 from Redoubt Volcano up in Alaska. Redoubt was also whistling, for around 1-3 minutes before some of the explosions that happened there in 2009. They called it "screams". It turned out that the screams where actually little earthquakes getting closer and closer together in time, until you can't distinguish them anymore and they're just one continuous scream. 

The screaming or whistling at Kilauea is quite different: It's really slow and lasts for many hours. Nobody has seen gliding that lasts this long anywhere before. Also, the little earthquakes that you can hear in the beginning actually SLOW DOWN before the whistling starts, so the explanation from Redoubt doesn't work here. Many other models that explain this type of behaviour can't produce whistling that would last for several hours, so we spent some time exploring what could generate a signal like that. In the end we decided that the Kilauea tremor and whistling may be related to bubbles in the magma: We think that it's possible that gas bubbles in the magma reservoir beneath the western crater Halema`uma`u can form "bubble clouds", or areas where lots of bubbles collect in one place. These bubble clouds can start swinging, or oscillating, if there is magma flow or something else that can start the oscillation. This swinging is transferred into the ground. When the magma flow changes (for example when a crack breaks open somewhere else in the system, like the dikes in 2007 and 2011) the frequency of the bubble cloud tone can change, and produce the whistling that we observe. 

To know whether that is actually what was happening at Kilauea we would need some more info, for example a detailed study of where exactly the whistling was coming from on the volcano. However, it was still really interesting to see how by comparing the earthquake recordings from several different eruptions we were able to identify similarities and differences, and how that - in combination with other observations during those eruptions - made it quite tricky to come up with possible explanations for what we observed. Many studies focus on just one eruption, but we showed that we can learn a lot by looking at the bigger picture.

If you're still reading this you really must have a lot of spare time, so feel free to check out the journal article that we wrote about all this.

A little digression: Large earthquakes, Alaska in 1964, and why people in Vancouver should be prepared

Time for a little digression. Let's talk about earthquakes! I've recently come back from a conference in Anchorage, Alaska, the Annual Meeting of the Seismological Society of America. I've only ever been to general geophysics/geosciences, or volcano conferences, so this one was quite the change. Since I study a specific type of earthquakes related to volcanoes I'm a bit in between volcanology and seismology, so it made sense to go.

Around 600 or so seismologists met up to talk about earthquakes and related stuff for three days. Overall it was a great conference. The "small" number of attendees was great - it was really easy to meet lots of people with very similar interests. I also liked the fact that they provided breakfast, lunch, and dinners (mostly). One reason for that is - of course, me being a student - the free food aspect, but there is something else: When you find a table to eat you may opt to find people you know, or you can go to a random table, introduce yourself, and start some interesting science talk. Bigger meetings like AGU are great to catch up with friends in different fields, but generally tend to be more anonymous.

A highlight of the conference was the post-conference field trip. Maybe around 1/3 to 1/2 of the conference attendees got on a bunch of busses to head down south towards the Kenai Peninsula. After leaving Anchorage, we stopped in Whittier, this interesting, tiny Alaskan town. We talked about the effects of the 1964 Alaska earthquakes, one of the biggest earthquakes ever recorded.

Around 5:30 in the afternoon on Good Friday, Mar 27, 1964, an earthquake with a magnitude somewhere around 9.2-9.3 struck just East of Whittier at approximately 25 km depth. The shaking was quite intense for a few minutes, but the real damage came from landslides and a tsunami generated by the earthquake. 

Ruins of a house that was abandoned after the earthquake in 1964, close to Girdwood, Alaska.

There was a heartbreaking account of one family's experience of the earthquake in the Anchorage Daily News a few weeks ago. From a seismology perspective, the earthquake is interesting for one specific subfield: paleoseismology. Paleoseismologists can study the change in ground elevation during the 1964 earthquake. A large area reaching from Kodiak island in the West through Anchorage out to Valdez and further East dropped in elevation during the earthquake because of the new plate configuration. Trees in the region that were slightly above sea level before now found their roots in the salt water, and died within a short time. They can be seen as eerie ghost forests until today. 

Ghost forest close to Girdwood, Alaska.

With the trees, a bunch of grass and shrubs ended up in saltwater. They were quickly covered by sand and silt washed up by the tides, and were preserved. During the field trip, we accessed one of the marsh areas. Our field trip guide Peter Haeussler showed us that when you remove the top layer of silt at the edge of the marsh during low tide, you can see a brown peat horizon. That's the grass from the 1964 earthquake!

Peat horizon from the 1964 earthquake. The brown is grass and shrubs that died after they ended up in saltwater after the earthquake, the grey on top is the silt that quickly covered everything.

A piece of grass that died when it was covered in saltwater after the ground dropped in elevation after the 1964 earthquake.

If you dig down deeper you can find more horizons like that, telling tales from previous large earthquakes in the area. Fossils in those peat horizons can be dated, and we thus know approximately at what intervals large earthquakes occur. Offshore BC and the Pacific Northwest, for example, people were speculating whether large earthquakes can occur at all (there aren't many small ones like e.g. in Alaska or New Zealand). Once paleoseismology became established, people found evidence of large earthquakes offshore the West coast of North America. That's how we know! And because we know now, everybody should consider having an emergency kit in the house. Because what we DON'T know is when the next big one is gonna strike.