Role of Infrasound in the USGS Eruption Response

Video Transcript
Download Video
Right-click and save to download

Detailed Description

Infrasound is an important tools for volcanologists, allowing them to detect eruptions, locate the source of explosions, and understand important parameters of volcanic events. John Lyons discusses how the technique is used by the USGS, especially in Alaska, and how we are working to help study infrasound around the world.
 

Details

Date Taken:

Length: 00:36:16

Location Taken: AK, US

Video Credits

USAID, University of Alaska Fairbanks Geophysical Institute, State of Alaska Division of Geological and Geophysical Surveys
 

Transcript

JOHN LYONS:    Stephanie asked me to talk about how we mostly at AVO,

JOHN LYONS:    but increasingly, in the USGS,

JOHN LYONS:    we are using infrasound as a monitoring tool.

JOHN LYONS:    So I put together this talk,

JOHN LYONS:    and it's not going to show a lot of cool science results,

JOHN LYONS:    but it's going to be more of the nuts and bolts of how we use

JOHN LYONS:    infrasound as a monitoring and research tool in the USGS.

JOHN LYONS:    I should say that there's a big group of us in the USGS,

JOHN LYONS:    particularly in Alaska, that are working in infrasound,

JOHN LYONS:    and that includes David Fee who is at AVO,

JOHN LYONS:    but also the University of Alaska, Fairbanks,

JOHN LYONS:    Matt Haney at AVO,

JOHN LYONS:    Hans Schwaiger at AVO,

JOHN LYONS:    West Thelen, who's at the Cascades Volcano Observatory,

JOHN LYONS:    and Aaron Wech at AVO.

JOHN LYONS:    Just some basics about infrasound.

JOHN LYONS:    I think a lot of people are familiar,

JOHN LYONS:    but I'm going to try to cover all the bases here for

JOHN LYONS:    those maybe not so familiar with infrasound.

JOHN LYONS:    It's a low frequency wave that propagates in the atmosphere.

JOHN LYONS:    Similar to seismic, we look at very simple things like the amplitude of the signal,

JOHN LYONS:    the duration of the signal,

JOHN LYONS:    the frequency of the signal when we're monitoring infrasound.

JOHN LYONS:    Usually, the energy is between about 0.5 and 20 hertz.

JOHN LYONS:    Twenty hertz is the upper limit.

JOHN LYONS:    That's the cut-off between infrasound or sound that

JOHN LYONS:    humans can't hear and sound that we can hear.

JOHN LYONS:    Certainly, larger eruptions can produce more low-frequency energy.

JOHN LYONS:    If you have a very large Plinian or Sub-Plinian eruption you can have energy

JOHN LYONS:    down to 10 seconds or even lower.

JOHN LYONS:    But this is the typical volcano band, 0.5- 20 hertz.

JOHN LYONS:    The propagation speed of

JOHN LYONS:    infrasound waves in the atmosphere is usually around 340 meters per second.

JOHN LYONS:    That's slows down as you get into the colder atmosphere,

JOHN LYONS:    but that's the typical wave speed.

JOHN LYONS:    That equates to about, the wave can travel 20 kilometers in about a minute.

JOHN LYONS:    So it's about an order of magnitude slower than seismic propagation.

JOHN LYONS:    But because the atmosphere is such a simple,

JOHN LYONS:    homogeneous medium compared to seismic,

JOHN LYONS:    that sound can propagate long, long distances.

JOHN LYONS:    In some cases, hundreds or even thousands of kilometers and we take advantage of that,

JOHN LYONS:    when using infrasound as a monitoring tool.

JOHN LYONS:    This may be obvious, but infrasound isn't affected by

JOHN LYONS:    clouds or night time.

JOHN LYONS:    It's a really good 24-7 monitoring tool.

JOHN LYONS:    Of course, infrasound is affected by other things like storms,

JOHN LYONS:    wave action in the ocean,

JOHN LYONS:    wind interacting with mountains.

JOHN LYONS:    Wind is the biggest noise source that we fight

JOHN LYONS:    in trying to understand what's a volcanic signal and what's a noise.

JOHN LYONS:    In terms of the volcano infrasound and what can it tell us,

JOHN LYONS:    I put this little scale on the left from things that we're doing

JOHN LYONS:    now and we're doing really well to things that we're working on it.

JOHN LYONS:    Fundamentally, what infrasound can tell us is it can

JOHN LYONS:    discriminate between whether or not something is coming out of the vent,

JOHN LYONS:    has an explosion occurred or not?

JOHN LYONS:    This is the most basic monitoring tool.

JOHN LYONS:    But then, increasing in complexity,

JOHN LYONS:    it can tell us when an eruption started,

JOHN LYONS:    whether it's still going on,

JOHN LYONS:    the duration of an eruption.

JOHN LYONS:    These are important for things like input parameters into transport modeling and fallout.

JOHN LYONS:    It helps us quantify source parameters like mass flow rate.

JOHN LYONS:    It can help identify changes in the intensity or the style of an eruption,

JOHN LYONS:    particularly in the midst of an eruption.

JOHN LYONS:    So it is really a real-time monitoring tool.

JOHN LYONS:    More and more, it's being integrated with seismic which

JOHN LYONS:    provides improved eruption source and wave propagation information.

JOHN LYONS:    As other remote sensing tools are coming online and are available and more quickly,

JOHN LYONS:    satellite remote sensing and lightning,

JOHN LYONS:    these can be integrated with infrasound to

JOHN LYONS:    provide greater insights into what's actually going on at the volcano.

JOHN LYONS:    A good example of this was during the 2016/2017 eruption of Bogoslof,

JOHN LYONS:    which produced a lot of lightning.

JOHN LYONS:    Then more recently, the Kelud and Krakatau eruptions in Indonesia.

JOHN LYONS:    Now getting down into the list of the things that we're working on or things that

JOHN LYONS:    we want to work on, eruption forecasting.

JOHN LYONS:    Typically infrasound has been a detection tool not a forecasting tool,

JOHN LYONS:    but at open-vent volcanoes there have now been a few case studies that have shown that.

JOHN LYONS:    It does seem to be possible to detect subtle changes,

JOHN LYONS:    and this is typically associated with degassing or the movement of magma in

JOHN LYONS:    an open conduit that will allow you to

JOHN LYONS:    do relatively short term forecasting using infrasound.

JOHN LYONS:    Then finally, detecting, tracking,

JOHN LYONS:    and characterizing surficial flows so lahars,

JOHN LYONS:    landslides, pyroclastic flows, and even lava flows.

JOHN LYONS:    This is becoming increasingly feasible with the instrumentation and,

JOHN LYONS:    the algorithms that process infrasound.

JOHN LYONS:    This is the list of things that

JOHN LYONS:    we can do and that we want to be able to do with infrasound.

JOHN LYONS:    In the United States,

JOHN LYONS:    there are now four different volcano observatories

JOHN LYONS:    or four different areas where infrasound is being used.

JOHN LYONS:    It's widely used, of course,

JOHN LYONS:    in Alaska at the Alaskan Volcano Observatory.

JOHN LYONS:    It's also being used in the Commonwealth of the Northern Mariana Islands.

JOHN LYONS:    We have an array in the CNMI on Saipan Island.

JOHN LYONS:    There are also arrays at

JOHN LYONS:    the Cascades Volcano Observatory and at the Hawaiian Volcano Observatory.

JOHN LYONS:    So it's been pretty widely used now across the USGS.

JOHN LYONS:    I'm mostly going to focus on Alaska examples today.

JOHN LYONS:    This is just a map of

JOHN LYONS:    all the eruptions or all the volcanoes with confirmed eruptions since the 1700s,

JOHN LYONS:    those are the red triangles and then Holocene volcanoes,

JOHN LYONS:    in yellow triangles.

JOHN LYONS:    It's about 60, 65 volcanoes that span about 1,500 kilometers along the arc.

JOHN LYONS:    A lot of these are islands that are remote.

JOHN LYONS:    You can see here, AVO is up here in Anchorage so this is

JOHN LYONS:    really remote monitoring for most of these volcanoes and we don't

JOHN LYONS:    have seismic networks on many of these volcanoes.

JOHN LYONS:    So we rely on remote monitoring techniques.

JOHN LYONS:    More and more infrasound is becoming one of

JOHN LYONS:    our main techniques for monitoring these volcanoes.

JOHN LYONS:    Of course in Alaska, I probably don't have to tell all of you this.

JOHN LYONS:    But the primary hazard is ash and aviation.

JOHN LYONS:    There aren't many communities that are close enough to

JOHN LYONS:    active volcanoes that things like pyroclastic flows or lava flows are really a hazard.

JOHN LYONS:    Ash fall is a hazard in some communities,

JOHN LYONS:    but the major hazard is ash at flight levels.

JOHN LYONS:    This is a pre-COVID number,

JOHN LYONS:    but prior to the COVID-19 pandemic,

JOHN LYONS:    there were usually about 50,000 passengers per day transiting the North Pacific.

JOHN LYONS:    I don't know what that number is now,

JOHN LYONS:    but I'd be curious to know,

JOHN LYONS:    it's probably, I don't know, 10,000, maybe less.

JOHN LYONS:    Typically, there are 10 to 12 days per year where ash reaches flight levels in Alaska.

JOHN LYONS:    That's just the average,

JOHN LYONS:    I think, over the past 30 years.

JOHN LYONS:    What we want to do is be able to detect with infrasound or

JOHN LYONS:    any method when is a big eruption occurring and is ash getting to flight levels.

JOHN LYONS:    One of our main tools is the infrasound arrays that we have spread out over the arc.

JOHN LYONS:    Here I'm showing, and the green stars are current AVO infrasound arrays.

JOHN LYONS:    We have six arrays spread out over the arc,

JOHN LYONS:    and we have plans for two more.

JOHN LYONS:    We have plans for one in the far Western Aleutians,

JOHN LYONS:    out on Amchitka Island,

JOHN LYONS:    however, because of COVID we're not able to do field work that far out west this year.

JOHN LYONS:    So instead, we're hoping to install this array in

JOHN LYONS:    the next month or so on the Kenai Peninsula that would cover

JOHN LYONS:    these Cook Inlet volcanoes that are closest to Anchorage.

JOHN LYONS:    What does an AVO array look like?

JOHN LYONS:    This is an example,

JOHN LYONS:    of a Google Earth map of  of the six elements in an array on Adak Island.

JOHN LYONS:    This is currently our Westernmost island.

JOHN LYONS:    It's about 1000 kilometers-900 kilometers from Anchorage.

JOHN LYONS:    As you can see it, it does have air service.

JOHN LYONS:    We are trying to put what we're calling

JOHN LYONS:    regional infrasound arrays in places that are easy to get to with commercial aviation.

JOHN LYONS:    A place where we don't have to hire a ship or a boat to access.

JOHN LYONS:    Currently, the arrays in Alaska all have between four and six elements.

JOHN LYONS:    So between four and six sensors that make up one array,

JOHN LYONS:    and the spacing between those sensors is typically 50 to about 120 meters.

JOHN LYONS:    In Alaska, there's not much vegetation.

JOHN LYONS:    As you can see in this photo in the Aleutians,

JOHN LYONS:    it's typically just grasses,

JOHN LYONS:    and we want to control the wind noise.

JOHN LYONS:    We have begun using these domes, as we call them.

JOHN LYONS:    It's basically a mesh aluminum dome that reduces the wind noise.

JOHN LYONS:    It sits over the sensor and reduces the wind noise.

JOHN LYONS:    If you can see here, this is an example of 

JOHN LYONS:    a Power Spectral Density plot for

JOHN LYONS:    a sensor here in black that's sitting outside the wind dome.

JOHN LYONS:    Then in blue, this is the sensor inside the wind dome.

JOHN LYONS:    This shows an,

JOHN LYONS:    8 dB decrease in just the background wind noise from the wind dome.

JOHN LYONS:    This is basically a wind filter over the sensor.

JOHN LYONS:    We are now outfitting all of our arrays in Alaska with these wind domes.

JOHN LYONS:    But if you have thick vegetation or a forest,

JOHN LYONS:    then you can put the array in the forest.

JOHN LYONS:    It can act very similarly like a wind filter.

JOHN LYONS:    As I mentioned, infrasound is now a primary tool

JOHN LYONS:    for rapid detection of expose eruptions in Alaska.

JOHN LYONS:    We send all of our infrasound data to IRIS so it's publicly available.

JOHN LYONS:    A big question is always like,

JOHN LYONS:    "Why do we need an array?

JOHN LYONS:    Why can't we just install one or put out one sensor?

JOHN LYONS:    It's cheaper, it's easier.

JOHN LYONS:    An array is difficult, it takes up a lot of space, you have to run cables."

JOHN LYONS:    The advantage of an array is that it provides a direction to the source of sound.

JOHN LYONS:    In addition, it provides the wave speed.

JOHN LYONS:    This is an example of an array.

JOHN LYONS:    The way an array works is that,

JOHN LYONS:    as a sound wave passes from their source,

JOHN LYONS:    as a sound wave passes the elements of the array,

JOHN LYONS:    there's a small time difference in arrival between each element.

JOHN LYONS:    What we do is we cross-correlate and get the time difference between all the elements.

JOHN LYONS:    Then we invert those time differences to get

JOHN LYONS:    a slowness vector for a plane wave crossing the array.

JOHN LYONS:    Then from that slowness vector,

JOHN LYONS:    we can calculate the back-azimuth or the direction that that sound came from,

JOHN LYONS:    and the velocity that that wave is sweeping across the array path.

JOHN LYONS:    That’s useful information for determining whether a signal is of interest or is noise,

JOHN LYONS:    and there's a good codebase to do those things automatically.

JOHN LYONS:    This is why we use arrays because we can get that information which is critical for

JOHN LYONS:    separating a real source from noise automatically,

JOHN LYONS:    and that's what we want to do.

JOHN LYONS:    Then this becomes the basis for things like our infrasound alarms.

JOHN LYONS:    When we were designing arrays in Alaska,

JOHN LYONS:    we can talk about local versus regional,

JOHN LYONS:    but in Alaska, most of our arrays are regional which means

JOHN LYONS:    they're more than 15 or 20 kilometers from the volcanoes.

JOHN LYONS:    When you start to get outside of 15 or 20 kilometers from the volcano,

JOHN LYONS:    because as sound propagates away from the source,

JOHN LYONS:    it tends to be refracted upward in the atmosphere.

JOHN LYONS:    That's because the atmosphere is colder as you go upwards.

JOHN LYONS:    The sound is refracted upwards,

JOHN LYONS:    and that can lead to shadow zones around or outside of about 20 kilometers.

JOHN LYONS:    I think Hans Schwaiger is on this call.

JOHN LYONS:    Hans has come up with this software that models

JOHN LYONS:    the regional infrasound propagation  during different times of the year.

JOHN LYONS:    Because depending on the time of the year,

JOHN LYONS:    the atmospheric winds change and that changes how sound propagates

JOHN LYONS:    at distances of 20-250 kilometers or more.

JOHN LYONS:    We model that, and we can understand whether

JOHN LYONS:    a proposed location for an array is good during different times of the year,

JOHN LYONS:    whether we expect infrasound arrivals

JOHN LYONS:    at certain locations during different times of the year or not.

JOHN LYONS:    This is an example of one of these forward models of

JOHN LYONS:    infrasound propagation for a source so you can pick a volcano.

JOHN LYONS:    The code gets the atmosphere weather conditions

JOHN LYONS:    from zero kilometers up to 140 kilometers in the atmosphere,

JOHN LYONS:    and then generates this 2D image.

JOHN LYONS:    Basically, what it's showing here is how much transmission loss,

JOHN LYONS:    or how much signal will be lost due to propagation from this source at this time.

JOHN LYONS:    Areas that are in the warmer colors are more likely to get an infrasound detection,

JOHN LYONS:    cooler areas less likely.

JOHN LYONS:    In this case, these pink circles are AVO infrasound arrays here,

JOHN LYONS:    here, here, here, and here.

JOHN LYONS:    The arrays to the West of the volcano and

JOHN LYONS:    Southwest will be very unlikely to get a detection if there was an eruption at this time.

JOHN LYONS:    While these ones to the East and to the North will be more likely to detect

JOHN LYONS:    that eruption. In addition.

JOHN LYONS:    this is a tool called AVO-G2S,

JOHN LYONS:    and Hans automatically runs these for all of our restless volcanoes in Alaska.

JOHN LYONS:    If an eruption happens,

JOHN LYONS:    we can quickly look,

JOHN LYONS:    or if a volcano becomes restless,

JOHN LYONS:    we can look and see whether it's likely or unlikely that

JOHN LYONS:    a certain array will detect  infrasound or an explosive event if it happens.

JOHN LYONS:    Then locally, there are problems with actually being too close to the volcano.

JOHN LYONS:    That's usually not a problem in Alaska.

JOHN LYONS:    Then topography, you have to consider that if you have

JOHN LYONS:    steep topography between your array and your source,

JOHN LYONS:    it may block that sound.

JOHN LYONS:    So that's just something to consider if you're putting arrays locally.

JOHN LYONS:    Yesterday, Hans was nice enough to run two of these forward models for Reventador.

JOHN LYONS:    This was if there was an eruption at Reventador,

JOHN LYONS:    yesterday, what would the propagation look like?

JOHN LYONS:    So the top plot is at zero.

JOHN LYONS:    This is a single frequency plot for a 0.5 Hz source.

JOHN LYONS:    This is at 0.5 hertz,

JOHN LYONS:    and this is at 0.1 hertz.

JOHN LYONS:    I'm guessing there are strong winds out of

JOHN LYONS:    the East because you're getting good propagation to the West,

JOHN LYONS:    but really bad propagation to the East.

JOHN LYONS:    If you had an array in this area,

JOHN LYONS:    it can be very unlikely to pick up sound or infrasound from an eruption of Reventador.

JOHN LYONS:    But if you had an array to the South or to the West,

JOHN LYONS:    it would be more likely.

JOHN LYONS:    Then if the eruption was lower frequency,

JOHN LYONS:    it would be more likely to radiate sound in all directions.

JOHN LYONS:    Any questions about that?

STEPHANIE:    John,  I have a question?

STEPHANIE:    What characteristics of the eruption would get it into

STEPHANIE:    that lower frequency category most commonly?

JOHN LYONS:    It's just the size, the energetics.

JOHN LYONS:    So little Strombolian eruptions,

JOHN LYONS:    they're usually not producing 10-second energy.

JOHN LYONS:    But a sub-Plinian or a Plinian eruption would produce that much lower frequency energy.

STEPHANIE:    What about like a Vulcanian Redoubt-style explosion?

JOHN LYONS:    If it was a large Vulcanian eruption, yeah,

JOHN LYONS:    certainly it produces lots of low frequency energy,

JOHN LYONS:    but open vent, more Strombolian in style, probably wouldn't.

JAY:    John, I have a question.

JOHN LYONS:    Yeah.

JAY:    Can you use those forward models for the atmosphere to take

JAY:    sound pressure recorded at regional distances so that might

JAY:    reduce it down to a local pressure that

JAY:    get at eruption dynamics if you're just recording

JAY:    regionally like hundreds of kilometers away?

JOHN LYONS:    Yeah. That's a good question. The problem is that

JOHN LYONS:    the atmospheric models are still pretty coarse,

JOHN LYONS:    so it's hard to

JOHN LYONS:    get how exactly that atmosphere affects the signal.

JOHN LYONS:    This is something we struggled with

JOHN LYONS:    a lot during the writing of the Bogoslof special issue.

JOHN LYONS:    Because the global weather models are really good,

JOHN LYONS:    but the spacing is coarse.

JOHN LYONS:    The regional models are not so good.

JOHN LYONS:    So I'd be hesitant to try to apply

JOHN LYONS:    this the actual effect on

JOHN LYONS:    the pressure as far as making a correction for the wind or for the propagation.

JOHN LYONS:    That's where we want to get to, certainly.

JOHN LYONS:    But I think the atmospheric specifications aren't quite good enough yet.

JAY:    Okay. Thanks.

JOHN LYONS:    Yeah. Okay. Continuing on array design,

JOHN LYONS:    there's always this question of, well,

JOHN LYONS:    how many sensors can I get away with?

JOHN LYONS:    Or how many sensors do I need?

JOHN LYONS:    Of course, for an array,

JOHN LYONS:    the minimum to get a direction is three.

JOHN LYONS:    But there are huge error problems with a three element array,

JOHN LYONS:    so the minimum that we suggest is four.

JOHN LYONS:    But if you have five or six,

JOHN LYONS:    you really reduce the effects of spatial aliasing.

JOHN LYONS:    If you're looking over multiple frequencies,

JOHN LYONS:    it really improves the array response,

JOHN LYONS:    and here's an example from the Okmok array in Alaska.

JOHN LYONS:    The red elements were the original four element infra sound array.

JOHN LYONS:    Then a couple of years ago,

JOHN LYONS:    we decided to expand it to a six element array,

JOHN LYONS:    then these yellow elements for the two additions.

JOHN LYONS:    We have some codes that do the model

JOHN LYONS:    the array response based on the location and the frequencies that you're interested in.

JOHN LYONS:    So you can see here, we're looking at this four-element array.

JOHN LYONS:    The left rose, this is azimuthal uncertainty.

JOHN LYONS:    It's six-and-a-half or seven degrees of uncertainty for the four element array.

JOHN LYONS:    Then a pretty wide uncertainty in the trace velocity,

JOHN LYONS:    this is in kilometers per second.

JOHN LYONS:    However, when you add those two elements and now you have a six element array,

JOHN LYONS:    you can see it really greatly reduces the azimuthal uncertainty in the array,

JOHN LYONS:    as well as the trace velocity uncertainty.

JOHN LYONS:    In addition to improving the azimuthal uncertainty and the trace velocity uncertainty,

JOHN LYONS:     adding more elements reduces noise or improves the signal to noise,

JOHN LYONS:    and it adds redundancy.

JOHN LYONS:    In Alaska, the most expensive thing isn't the array or installing the array,

JOHN LYONS:    it's getting to the array typically.

JOHN LYONS:    If we have a six element array,

JOHN LYONS:    we can lose one or two elements and still be able to get it back azimuth,

JOHN LYONS:    and we might not be able to visit that array during winter,

JOHN LYONS:    so we might have to wait several months,

JOHN LYONS:    so it's really important for us to have that redundancy built in.

JOHN LYONS:    The other benefit of arrays is that we have

JOHN LYONS:    these really pretty mature codes now that

JOHN LYONS:    automatically process the data and provide these plots that anyone in

JOHN LYONS:    the observatory can access and look at what's going on with any volcano.

JOHN LYONS:    This is just a screenshot of the array processing output

JOHN LYONS:    of the least-squares beamforming array processing code.

JOHN LYONS:    This runs automatically every 10 minutes.

JOHN LYONS:    This code is freely available. Anyone can use it.

JOHN LYONS:    It's a Python code.

JOHN LYONS:    The array processing outputs,

JOHN LYONS:     a waveform,

JOHN LYONS:    this MCCM which is the mean cross-correlation maximum,

JOHN LYONS:    so it's the average of

JOHN LYONS:    all the maximum cross-correlation values between the array element pairs.

JOHN LYONS:    These low values tell you that it's probably just some noise happening.

JOHN LYONS:    Then when you get these high values,

JOHN LYONS:    that tells you that you have a well correlated signal crossing the array.

JOHN LYONS:    This code also outputs the trace velocity.

JOHN LYONS:    So when you have trace velocities in this gray box between

JOHN LYONS:    280 and about 430 meters per second,

JOHN LYONS:    those are realistic infrasound propagation velocities.

JOHN LYONS:    If you see velocities outside of that,

JOHN LYONS:    you know that that's probably not a signal of interests.

JOHN LYONS:    Then the bottom plot is showing us the back azimuth from 0-360.

JOHN LYONS:    What we do is for each array,

JOHN LYONS:    we have volcanoes of interest,

JOHN LYONS:    and we can plot those.

JOHN LYONS:    The back azimuth from the array to

JOHN LYONS:    that volcano of interests, and that's what we're showing here.

JOHN LYONS:    This is Makushin, Cleveland, Okmok and Bogoslof.

JOHN LYONS:    In this case, there's some background noise and then we have this signal of interest,

JOHN LYONS:    it's well correlated, it has a reasonable trace velocity,

JOHN LYONS:    and it's more or less coming from the direction of Bogoslof volcano.

JOHN LYONS:    Then we can be reasonably confident that yes,

JOHN LYONS:    Bogoslof volcano is producing this signal.

JOHN LYONS:    This is updated every 10 minutes.

JOHN LYONS:    We found that this tool is pretty intuitive.

JOHN LYONS:    After some basic training  essentially everyone in the observatory understands how to use this tool.

JOHN LYONS:    I can go to quickly access the data and have a look and

JOHN LYONS:    see if something is showing some sign of unrest or is erupting.

JOHN LYONS:    That same array processing code is also the basis for automated alarms.

JOHN LYONS:    This code doesn't send out alarms,

JOHN LYONS:    it's just running in putting this information onto a web page.

JOHN LYONS:    Then we have another version of this code.

JOHN LYONS:    It's the same code essentially,

JOHN LYONS:    but it's running in shorter time window.

JOHN LYONS:    So we have a certain set of parameters that we've set up.

JOHN LYONS:    The azimuth, that's how many degrees off of the actual volcano the signal is.

JOHN LYONS:    There's some range of velocities that we set up and then some pressure threshold.

JOHN LYONS:    If all those trigger positively,

JOHN LYONS:    then this code sends out an email and

JOHN LYONS:    text alert to whoever is signed up to get those messages.

JOHN LYONS:    As far as situational awareness,

JOHN LYONS:    this is a really important part of

JOHN LYONS:    our alarm structure for monitoring volcanoes in Alaska.

JOHN LYONS:    Now, everyone is remote,

JOHN LYONS:    so everyone's at home,

JOHN LYONS:    but we can get these notifications on our phone,

JOHN LYONS:    and then it tells you to go look at

JOHN LYONS:    the computer and see what's going on at the volcano.

JOHN LYONS:    Again, these are open source Python codes that are pretty easy

JOHN LYONS:    to set up and get running once you have array data.

JOHN LYONS:    Now, more and more as what we're 

JOHN LYONS:    updating all of our stations in Alaska from analog to digital.

JOHN LYONS:    As we do that, we're adding single infra sound sensors that are co-located with seismic.

JOHN LYONS:    In some cases, we're adding just one or two sensors per network.

JOHN LYONS:    But in volcanoes that are particularly active,

JOHN LYONS:    we're adding 3-6 sensors that become a local network around those volcanoes.

JOHN LYONS:    Because this network configuration requires

JOHN LYONS:    some different tools than array,  so we're developing tools, for instance,

JOHN LYONS:    to be able to locate the sources of sound in networks automatically.

JOHN LYONS:    This is an example from Shishaldin volcano.

JOHN LYONS:    This uses a method called reverse time migration.

JOHN LYONS:    It's basically just back propagating the source to get the location.

JOHN LYONS:    In this case, the green dot is the vent and

JOHN LYONS:    the star is the location from these three stations.

JOHN LYONS:    This is a more of a research tool at this point.

JOHN LYONS:    but we're working towards having automated infrasound

JOHN LYONS:    event location as part of the AVO automated processing codes.

JOHN LYONS:    Then with just a single co-located seismic and infrasound sensor,

JOHN LYONS:    you can do things like look at

JOHN LYONS:    the coherence of the signal between the infrasound and seismic.

JOHN LYONS:    This is an example again from Shishaldin volcano.

JOHN LYONS:    This plot is showing the coherence at different frequencies.

JOHN LYONS:    When the coherence is high,

JOHN LYONS:    you know that signal is likely a real signal and not just some noise.

JOHN LYONS:    Sometimes it can be difficult to tell a real signal.

JOHN LYONS:    These are pretty obvious, but on a single infrasound sensor,

JOHN LYONS:    it can be really difficult to tell a signal from just a burst of wind.

JOHN LYONS:    This coherence method is very useful for identifying

JOHN LYONS:    signal from noise if you just have a single seismic and infrasound sensor.

JOHN LYONS:    Then through VDAP, we're starting to expand the use of the infrasound internationally.

JOHN LYONS:    I think VDAP's now providing equipment,

JOHN LYONS:    the software that the USGS is using,

JOHN LYONS:    some training, and then analytical support.

JOHN LYONS:    The first big example of this was in 2018,

JOHN LYONS:    after the major eruption of Fuego volcano,

JOHN LYONS:    when we installed multiple arrays around different lahar channels,

JOHN LYONS:    or drainages to try and detect and track lahars,

JOHN LYONS:    and that work is ongoing.

JOHN LYONS:    As I mentioned, that's on the cutting edge of what we want to be able to do.

JOHN LYONS:    We also installed a six-element array that's been really

JOHN LYONS:    useful to INSIVUMEH for tracking eruptive activity at Fuego,

JOHN LYONS:    and INSIVUMEH in Guatemala is using the USGS i-Pensive array processing software operationally now,

JOHN LYONS:    and they're getting some pretty interesting results, pretty exciting.

JOHN LYONS:    In Peru, we also visited Peru in 2018,

JOHN LYONS:    and the Geophysical institute there is using I-Pensive,

JOHN LYONS:    the I-Pensive software for array processing.

JOHN LYONS:    We had planned to install an array at Ubinas this year,

JOHN LYONS:    but that of course got put on hold because of COVID,

JOHN LYONS:    but I think there's strong interest in Peru for installing more infrasound,

JOHN LYONS:    and it sounds like in Ecuador as well, that there's interest.

JOHN LYONS:    Of course, even the IG has infrasound for some time.

JOHN LYONS:    So just to summarize, now,

JOHN LYONS:    infrasound has really become a mature tool equal to seismic for monitoring.

JOHN LYONS:    There's been really major growth across the USGS in

JOHN LYONS:    the last 10 years or so in the number of stations,

JOHN LYONS:    the quality of the sensors,

JOHN LYONS:    and most of the data now is going to IRIS,

JOHN LYONS:    so other people can access that data and look at it.

JOHN LYONS:    Yeah, lots of opportunities to collaborate,

JOHN LYONS:    and then there's this growing open source code base that really

JOHN LYONS:    makes analyzing these events much easier.

JOHN LYONS:    In the past, it was pretty difficult.

JOHN LYONS:    Everyone had their own code,

JOHN LYONS:    but now that these codes have become easily shareable,

JOHN LYONS:    they're all Python-based, easy to share,

JOHN LYONS:    and people respond when you have questions,

JOHN LYONS:    and they're really pretty easy to set up.