 Hello, I'm Scott Cameron from Beezer and I want to welcome you to the webinar on the Cascadia MegaQuake Earthquake Early Warning and Tsunami Modeling. This is the final webinar in a three-part series on the MegaQuake. Our speakers for today are doctors Richard Allen of the University of California Berkeley and Dr. Diego Melgar of the University of Oregon. The webinar series addresses a key question. Is the West Coast ready for a 9.0 magnitude earthquake followed by a large tsunami? Data collected over the last 30 years show that multiple giant earthquakes and associated local tsunamis have struck the Pacific Northwest for at least the past 10,000 years. The 800 mile long Cascadia Subduction Zone, which extends from Northern California, Oregon, Washington, and to southern British Columbia, is the main source of these earthquakes and the accompanying tsunamis. This three-part webinar series looks at the science and engineering associated with the earthquake source, the hazards, current strategies to mitigate loss of life, and emerging opportunities in early warning and reducing uncertainty. Now just a few quick things before we get started. First, the audio for today's event will be streamed through your computer speaker's or phone. We will be taking questions through the Q&A box located on the bottom toolbar of the Zoom screen. Simply type your question in the box at any time and click send. The moderator will get to as many questions as possible in the final 15 minutes of the webinar. Second, this webinar is being recorded. Please understand that any questions you submit may be read aloud and included in our recording. A link to the recording as well as a copy of the slides will be posted on our website within the next week or so. And third, if you have any technical issues during the event, please try restarting the login process. Please also note this is being recorded and the video and audio will be posted. And with that, I'd like to turn things over to our first speaker, Dr. Richard Allen. Great. Well, thank you, Scott. Hopefully you can now see my screen. So it's great. I'm very much happy to be participating in this seminar series. We've been thinking a lot about Cascadia and the science of Cascadia, the hazards of Cascadia. And so I'm delighted to kind of talk a little bit about earthquake early warning and how earthquake early warning can really help mitigate the hazards in Cascadia. So I've been tasked with talking about ShakeAlert today. And so ShakeAlert is an earthquake early warning system, an incipient earthquake early warning system, if you like, for the US West Coast. So covering California, Oregon and Washington, which of course includes Cascadia. And so I'm going to give an overview of what earthquake early warning is about, how earthquake early warning works and sort of where we are with this project. And then I'm going to hand it over to Diego, who's going to kind of extend that to talking about the really big earthquakes, which of course is what we're worried about in Cascadia, and also talking about the potential to do local tsunami warning as well. So I'm at the UC Berty Seismology Lab, but really I'm representing a much larger group that's been working on ShakeAlert. And you can see the institutions involved. This is by no means even all of the institutions involved. I'm focusing on the sort of key project participants here. ShakeAlert is a USGS project. It will be the USGS issuing alerts later this year. And then of course the universities that run the seismic networks have been very involved in the development of the concept and now the implementation. And that's Caltech, UC Berty, University of Oregon and the University of Washington. And I also want to acknowledge the Gordon Betty Moore Foundation, who really made an investment in the research that has led to ShakeAlert. And now I'm pleased to say that the both federal and state governments are what are really now taking on the funding piece. So I'm going to give an overview and the bottom line here is that ShakeAlert's coming this year. The plan is to start issuing warnings up and down the West Coast later this year, hence the line here coming to a phone near you. And I'll explain what that means in a moment. Okay, so I want to back up a little bit and start by talking about alerts and thinking about alerts and explaining why alerts are important and why earthquake alerts are also important. So I think in the modern era where we all have communications in our pockets, we've come to really expect alerts for a whole variety of different purposes. Hurricane alerts of course have been around for a long time. And I don't think we could imagine not getting alerts and warnings in the run up to hurricanes. Of course, those alerts have more time associated with them by which I mean days or hours as a hurricane is approaching a region. But we've come to expect these alerts and we've come to understand how to respond to them. Same for tornadoes. Tornado warnings and watches have been around for many years. Tornado warnings are a much shorter time scale. Now we're talking about minutes of warnings. And the other thing that this image really illustrates is of course you wish you were tornado warning to a region. But then the actual specific houses in this case that the tornado affects is only some subset of that. So that's something to bear in mind. We push out warnings to wide regions that could be affected by an event. And then some portion only some portion of that region is then typically affected. Bringing this a little closer to home for me in Northern California at the end of last year we had some very serious fires that I think everybody knows about that really went through entire neighborhoods in a matter of hours one night resulted in many fatalities and resulted in many people asking why did we not get a warning? Why did we not get a warning in the hours and minutes running up to this disaster that this was underway? And there's a lot of soul searching about why is it as a society we're not able to get these warnings out to people so they can get themselves out of harm's way. Of course fires then continued in January in Southern California and were then followed by serious mudslides. And again there was a lack of warnings for these mudslides which led to fatalities. And people are responding to that. In fact right now today thousands of people have evacuated in Southern California because of the storms that we've just had coming through in the last day or two in order to get out of a way. So my point is that we're used to getting alerts. We are actually expecting more and more alerts more and more accurate information and people are used to responding to them and understand the consequences of not responding to these alerts. So why not earthquake alerts? Here's a typical image I think for what we can expect from an earthquake in the United States. We're very fortunate in the U.S. to have building codes that mean that we don't expect many buildings or all buildings to collapse. Some of course may. But this is perhaps a more typical representation of what we should be expecting after an earthquake. And that is that the building is still standing but the interior contents is been thrown around, ceiling tiles are falling down, bookcases have fallen over. And so when it comes to getting an alert, getting an alert to protect yourself from this kind of damage from the injuries being in this room is what we're really talking about. Again closer to home for me, the Napa earthquake a couple of years ago had just one fatality that this is a fireplace that collapsed into a living room. And again having a few seconds of alert will allow you to get out of harm's way, move to a safe zone. And so that's kind of what we're thinking about when we're talking about earthquake early warning. And then just to finish this is, I just was stunned by this image. I can't help but include it. This was again in the aftermath of the Napa earthquake. Needless to say there are many, many aftershocks and this guy really shouldn't have his head in this crack given that there's a very high probability of an aftershock when this picture was taken on the day of the Napa earthquake. And this is really just to illustrate that there are many specific examples where having a few seconds of warning to protect yourself really is very valuable. So that's what we're talking about earthquake alerts. And the point is that this is possible. This is already being done in various parts around the world. And to illustrate that I want to show this video from Mexico City. The siren that you can hear in the background in this video is actually the early warning system that puts out this message across Mexico City. That's why all of these people are in the street because they have a warning. In this video obviously illustrates the value of having had that warning, of having had a little bit of time to actually get out of the building. And in fact, just while this is an unreinforced masonry type building, and there are not so many of these buildings in the US, this sort of also illustrates that the actual damage is actually significantly into the earthquake. The shaking had actually been going on for some time by the time that this building collapsed. But so the point here is that earthquake early warning is possible. Other countries are doing it and it's useful. Okay, so let's back up. And what is earthquake early warning and why is it possible to do earthquake early warning? So for most people, they think of earthquakes as being an instantaneous process. The earthquake is the moment at which they feel shaking. And of course, that's not the case. For earthquakes, particularly large magnitude earthquakes, there's a significant amount of time between when the earthquake starts and when people feel shaking. So this is an animation. It's obviously a cartoon, but it's running in real time. And it's going to show a magnitude eight coming down the San Andreas Fault. That broad brown line is the actual rupture coming down the fault. The circles are the seismic energy. So the yellow circles are the P waves. And so for people in the Bay Area, in this particular scenario, they would just start to feel the P waves. And these P waves, you would feel it for sure. But everybody would look at one another and probably assume that this is a relatively small earthquake that's close by and that that will stop momentarily. It's really not until the S wave arrives that you would start to expect to see any kind of damage for this sort of earthquake. And so the S wave, obviously, right now, the red circle is just beginning to propagate across the Bay Area. So now there would be really very strong shaking. And you might expect there to be to be damage. But the strongest shaking and most damage won't actually occur until the fault rupture itself passes by San Francisco. That's still another 30 seconds away. And so this is just to illustrate that earthquakes, big earthquakes, they take a significant amount of time from when they start until they finish or until potentially the shaking is felt in your particular location. This, of course, is for Northern California, but we could spin this map around upside down. And this could be a magnitude eight coming up the San Andreas Fault towards Los Angeles. Or rather than going south from the Mendocina Triple Junction, we could go north. And this could be a magnitude nine rupture on the Cascadia Subduction Zone. It doesn't matter. The point is that these events take time. And the concept behind early warning is to recognize an earthquake when it starts as rapidly as possible and then provide a warning to people before they start to feel the shaking. So that's the concept. So who could actually make use of early warning? There's a whole range of users. But for me, I think that the most valuable use or at least the first use that we want to be thinking about for early warning is about personal protection, is what do I do to protect myself during the course of an earthquake? And so the key thing to be doing there is to be reducing the likelihood of being injured. The image on the right is the one I showed at the beginning from the Northridge earthquake. But when we look at both the Northridge and the Loma Prieta earthquake, just as being two examples of relatively serious earthquakes in the US, in the Loma Prieta earthquake, more than 50% of injuries were linked to falls, people falling. And in Northridge, more than 50% were due to non-structural hazards. So bookcases, things like that falling on top of you. So the point here being that is if everybody received just a few seconds of warning, if everybody then dropped, took cover and hold on to do exactly the same thing that you should do in an earthquake anyway, then we could potentially reduce the number of injuries in an earthquake by more than 50%. That's a significant impact in terms of reducing the effects of an earthquake. In the case of Northridge, it's estimated that the cost of just the injuries is somewhere between two and three billion dollars. So for the people who want to do the cost benefit analysis of this, there's a huge saving to be made by doing early warning. And for me, I think this kind of reaction, individual response to protect themselves in an earthquake is perhaps the most elite, most important, a real use of early warning and probably one of the first applications of early warning we would want to see. But then there's a whole other range of applications related to automated control, slowing and stopping trains. The trains in the Bay Area have actually been using ShakeAlert, the research version of ShakeAlert since 2012 to slow and stop trains. But other kinds of automated response include stopping elevators, isolating hazardous machinery and chemicals, data security issues so that you don't lose massive amounts of data sets and just general situational awareness. So there's many potential applications for earthquake early warning. So how do you get the warning? I mean, the answer to that, I think, is we want to push out warnings in as many ways as possible. The first way that everybody thinks of is through using smartphones or cell phones, which the majority of people have in their pockets or in their bag, pretty much all the time. And so what might that look like? Well, this is an app, this is an app that we created here at the Berkeley Seismu Lab a few years ago, just to demonstrate the capabilities of doing it. So this is a real app that gets warnings from ShakeAlert. This version of the app gives you a countdown until you might expect the shaking. It also gives you a sense of the strength of shaking that you might expect. So this is one way that you could receive the alert, but you could also receive it on your computer. This is an image of what's called the user display that ShakeAlert has built to display the warnings on a desktop computer. But as we move to a public system, this alert should of course be pushed out through a whole variety of approaches. TV, radio, your home security system could get the warning, office, PA systems, etc. And exactly what the nature of the warning should be, the messaging itself is also something that is still under discussion, under debate. But recent lessons in Mexico are sort of beginning to suggest that the message should really be as simple as possible. It should just be earthquake, drop cover and hold on. That kind of message rather than this more detailed information perhaps that I'm showing here. And so finally, I hope that this is going to work. I want to demonstrate that the technology exists. And rather than me telling you it exists, I want to have Rachel Maddow tell you that it exists with this short little clip from the Rachel Maddow show a couple of years ago. Digital communications travel at the speed of light. Thanks to things like fiber optic cable, we can move information literally in a flash and that is good, just in the abstract. But that is, it turns out potentially life-saving. If the information that you are moving at the speed of light is notification that an earthquake is about to happen. Earthquakes happen in a specific place in the ground, right? Earthquakes have an epicenter. But you don't just feel an earthquake at the epicenter. The shuttering waves of motion in the earth from earthquake, they emanate out from the epicenter of the quake, traveling at the speed of sand. So if you had the kind of motion sensors that detected earthquakes, if you had seismometers along fault lines all over earthquake prone regions, when there was an earthquake, the seismometers nearest to the epicenter, they could register the earthquake has happened, right, fuel the shake, and then they could send a digital signal at the speed of light, notifying communities nearby that this traveling at the speed of sound, tremor, this motion of the earth is about to arrive. This is not a way of predicting earthquakes before they happen. It's a way of basically warning people that an earthquake has just happened and that they are about to feel its effects. Brace yourself. But this thing works. Scientists at UC Berkeley say that their shake alert to earthquake early warning project, it set off an alert. It did go off this weekend during this weekend's large quake in Napa, California. It was about to be felt in Berkeley and this is the alert that told them so. Watch this. Earthquake, earthquake, light, shaking, expected in three seconds. I got a 10 second warning that they were about to feel quake and 10 seconds is not much time, right? But this this brace yourself warning system, it works thanks to the simple fact that the speed of light is faster than the speed of sound and earthquakes only move at the speed of sound. With more sensors in more places, they presumably deliver even more warning time California's hoping to have that bigger system in place in the next few years if they can get it funded and finished. From the LA Times today, once fully developed, the system could give downtown Los Angeles 40 to 50 seconds of warning that the big one was headed from the San Andreas Falls, giving time for elevators to stop at the next floor and open up for firefighters to open up garage doors for high speed trains to slow down to avoid derailment and for surgeons to take the scalpel out of a patient. Giving your surgeon enough time to get the scalpel out of you before the giant earthquake starts shaking the operating room. That idea of a tectonically shaking a scalpel inside your body somewhere, that is something you can never unknow and I'm sorry but the idea that this warning technology, that it works, that it worked this weekend in fact. It's not just on a drawing board somewhere, it is in effect and it could be expanded and is being expanded. That is an excellent thing. They've got a system like this in Japan, they've got it in Mexico and some other places around the globe. California could get their whole statewide system done in a few years. God bless the geeks. They will save us all. Now it's time for the last verb with Lawrence O'Donnell. Hi Lawrence. Rachel, I just want them to make the earthquake warning voice a little less scary. So this video is two years old at this point. So this has been possible for some time. Needless to say it is complicated. We are getting very close though and the good news is that this talks about California but we're not talking about California, we're talking about the entire west coast for earthquake early warning. So today, ShakeAlert is spanning the west coast as a whole by which I mean California, Oregon and Washington. ShakeAlert is running and detecting earthquakes and pushing out alerts to test users as we speak using over 800 seismic stations that run up and down the west coast. This data is streaming continuously into four processing centers and about 75 institutions currently get the alerts from this test system. So this is why we're ready to start talking about the being public warning or limited public warning later this year because we know that this is possible. We know that we have a system that is generating alerts and it's time to move forward and I just kind of want to again emphasize that this is being done through this broad collaboration with the USGS as the lead agency, Berkeley, University of Washington, Caltech and Oregon and of course also with state partners as we move forward in all three states. It requires a lot of people to get coordinated to make this happen and that is exactly what's happening which is great to see. So in terms of building ShakeAlert, what is it that we need to do to get to this limited public rollout? Well the cost estimate that was put together a few years ago by the group for what it would cost is about $16 million per year plus about $38 million of infrastructure upgrade and the good news is that there's good news and then there's even more good news. The good news is that we're pretty much 50% funded so last year Congress allocated $10 million and then in terms of the infrastructure there was $10 million for California, Oregon also contributed some funding and then in addition another $10 million is in the budget for California this year but the really good news is as of a few hours ago when President Trump signed the FY18 budget in fact that includes a significant increase for ShakeAlert going from that $10 million last year up to $23 million this year and that's an increase for the operating costs but it's also another additional $10 million for the infrastructure to add the additional stations so this is as of literally two hours ago which is why it's not on this slide but we're getting we're not all the way there but we're getting pretty close to what the target was in terms of this funding and our ability to operate the system is sort of similarly far along in terms of the algorithms I would say the algorithms are pretty much 90% of the way there at least they're 90% of the way in terms of being ready to issue alerts based on the simple algorithms the point source algorithms there's still plenty more research to be done to improve the algorithms for the really big earthquake and that's what Diego is going to be focusing on but then the other piece is the sensors we're about 50% of the way there with the sensors this additional funding that that was finalized this morning will actually really help move us forward to building out the the rest of those sensors but despite the fact that we're only sort of 50% there in terms of building out the system we're still moving towards this limited public rollout later this year October is the the target date at this point and there's a lot of discussion going on within the shake alert project right now as to what the scope of the limited public rollout should be as I already mentioned there are already a group of users out there that are automatically responding to these alerts but being one example the metro system in LA being another there are up and down up and down Cascadia as well as California of course there are a group of providers alert providers that are getting engaged with the project in order to broadcast the the warning to to more users this includes early warning labs regroup and sky alert so this is a way of taking that kernel of an alert and pushing it out to lots of users and then the really exciting prospect is the idea that wireless emergency alerts this is the amber alert system that we all get on our phones may also be able to issue the alerts come this fall exactly how fast they will be able to do that is not clear and it may not be as fast as we would like but this really opens up the possibilities of reaching a very large group of people later this year we'll have to see how that pans out to give one example that I think those of us involved in the shake alert project to particularly excited about is to use smartphone apps to deliver the alert to school teachers this is the idea here is that in schools it's a relatively controlled environment we can train students to react in an appropriate fashion to the alerts they already do earthquake drills of course and so by putting an app on teachers phones we can potentially reach a large group of people pretty quickly again I wanted to give a sense of of where we are and what might be possible specifically deciding who gets the alerts and when these alerts get rolled out still has to be decided and of course that's a decision that the USGS has to make because they're the the responsible agency for actually issuing the alerts but just to sort of wrap up we're kind of ready to do this just to illustrate that I'm going to just take one specific example it happens to be very local to me it was right beneath Berkeley but this is the magnitude 4.4 earthquake just a month or so ago I guess a couple of months ago now of course this isn't the kind of earthquake that would do any damage but this is an earthquake right beneath the metropolitan environment and the warning shake alert pushed out a warning for this earthquake in just a few seconds there is there is the likelihood for these events that there will be a region very close to the epicenter that gets no warning and that's what's illustrated by the red circle on the in the diagram on the right but this also demonstrates just how quickly the system is now reacting and is pushing out the warning so I think this is just another demonstration that shake alert is ready for this test and for this this limited public rollout and so that system the system that pushed out the alert for the magnitude 4.4 is a an algorithm that's based on point source which is a great place for us to get started however as we know cascadia is is ready for a very large earthquake as is the San Andreas fault and so in order to do better than that we need to be able to treat large magnitude earthquakes differently and so that's the point at which I'm going to stop and I'm going to hand over to Diego to start talking about that thanks Richard I'm sharing my screen right now okay so that's right so what we're concerned about in the cascadia subduction zone are these very large magnitude events and this requires a different treatment than what we can currently do with what Richard described as the point source algorithms so what I'll talk about today is a lot more focused still on research and less on operational realities but we're making good progress the research for tsunami early warning has mostly been carried out by these west coast universities but we really have been working closely together with both tsunami warning centers the pacific tsunami warning center in Honolulu and the national tsunami warning center in Palmer Alaska who have really been great at teaching us how they think about tsunami warning what they need how they respond what responsibilities do they have so that we can tailor our algorithms and research towards stuff that is actually useful to them and of course all of this under the auspices of NOAA and NASA who's funded a lot of the GPS effort that I'll show you here today so in kelp in Oregon in in washington and cascadia in general we know that large events have happened in the past and we know this with quite a high degree of certainty especially because of places like this this is the nesco and ghost forest in coastal oregon where we know that every single one of these cedar and spruce trees died um on the exact same year and we know that that's because the tsunami came ashore the land subsided during a magnitude nine earthquake and the trees cannot live in the salty water and eventually died so we know because of observations like these and others that these large events happen roughly every 300 years and we know that the last one we know it to the day happened on january 26 of the year 1700 now this is our cascadia subduction zone where i will focus for the rest of my talk here is oregon in washington uh the northern california border is here portlands around here seattle this is a puget sound area and this is seattle of course these big earthquakes happen because of plate tectonics because we have the juan de fuca plate right here moving towards north america roughly at two inches per year basically the speed at which your fingernails grow and crashing into the north american continent and as it does so every now and then it will make large earthquakes very large earthquakes the size of the piece of fault that ruptures during these magnitude nine events can be as big as 700 miles now that's a very very long distance so our first challenge for issuing tsunami warning alerts is to be able to accurately measure a behemoth a giant of the size to measure it in real time and to measure it accurately the second challenge is to do it quickly we know because of the speeds at which tsunamis typically propagate we know exactly how much time we will have to issue a warning we know that the tsunami waves will likely originate closest to the trench this is this blue line the deepest part of the ocean offshore cascadia and we know the distances from the trench to the coast they range between 40 miles roughly in southern oregon to 80 miles here in the olympic peninsula that means that the first tsunami waves from one of these large events will arrive in five to ten minutes so that's the second challenge after measuring this gigantic event and doing the best that we possibly can we have to issue a forecast of what we expect the tsunami wave heights to be in under five minutes for it to be useful to the people that are immediately in harm's way now the third challenge which is somewhat related is that these events are very infrequent the last one was 317 years ago so how do we know that we're doing well unlike shake alerts that can be exercised regularly with these small magnitude events we haven't had a big event in cascadia in a really long time so the third challenge is coming up with a proper assessment of whether our algorithms and our research are functioning correctly and providing the best possible and useful tsunami warning alerts so this is what we're talking about here for the cascadia subduction zone this is a picture of the tohoku tsunami that happened in japan in 2011 this is a beautiful model put together by nois pml group and it shows the tsunami amplitudes all across the pacific basin the tsunami warning centers both in honolulu and alaska are experts at this kind of warning where they can issue warning from an event that starts anywhere in the world they issue warning to people downstream across the pacific where we have hours of lee time they've been doing this for 50 60 years now with great success but where the frontier lies is in this white box this region immediately adjacent to the earthquake where we only have minutes of lee time is there where we still have a lot of work and a lot of improvements to do i'll remind you that even though um the tsunami warning centers have been around for a while japan is the only country in the world that currently today has a local tsunami warning system that was built for that very express purpose of warning the people immediately next to a large earthquake that's the situation right now worldwide so i'd like to explore a little bit how that system performed in japan and see if there's any lessons that that we can take from that what happened in 2011 in japan was that their system even though the earthquake magnitude was nine we now know with the benefit of hindsight it was a magnitude nine system the actual magnitude computed by the real-time system never grew above a magnitude eight this is a condition known as magnitude saturation basically we cannot tell the large from the very large we can't tell them apart with just seconds to minutes and as a result this is the warning that came out of the public local tsunami warning system in japan where red means major tsunami orange means a significant tsunami and yellow means the tsunami heights is estimated to be about half a meter this is what happened in three minutes now this is what the tsunami warning should have looked like and this is the map that came out 13 hours after the earthquake this is now long after the first waves have arrived everywhere across the japanese islands it's because of this magnitude saturation that this map which was the real tsunami was not issued in three minutes this condition of magnitude saturation is a real problem for us especially in cascadia because we will have these very large events so using our the experiences from our from our japanese colleagues we tried to research why and how could we fix this kind of problem to understand why these large events are tricky we need to think about the real sizes so here are the these rectangles represent the fault areas the sizes of fault that broke in some of these famous california earthquakes here's the napa events that richard just spoke about the chunk of earth that broke in that event was roughly 20 miles across if you go up all the way to the mexican earthquake which some people in southern california might remember in 2010 that was about a hundred miles long and in between here's the loma prieta earthquake the world series earthquake of 1989 magnitude 6.9 was roughly 40 miles across i've scaled them down here so here are those california events and here's what subduction zone events like the ones in cascadia here's what they look like here's how they compare to the california events they're truly massive that japan earthquake is almost 400 miles across the earthquake in indonesia and sumatra that caused the big tsunami that killed more than 200 000 people magnitude 9.2 was almost 600 almost 800 miles across so these big events are truly truly large and as a result seismometers the rulers that we use to measure earthquakes they have a really hard time measuring things this big it's possible but it's very very difficult so what we proposed and what we started working on and what seems to be a very reasonable path forward is to use other kinds of measurement devices of sensing technologies to see if a big earthquake is is going to get big and it turns out that one of the best solutions for this problem is gps so this is high quality gps not like the gps on your phone here's the gps antenna measuring signals to gps satellites in the sky solar panels to power it in a satellite uplink to beam the data back to a processing center this gps is different so you are all used to gps on your phones because you when you open up google maps to find a place for dinner you ask the phone to tell you where you are that phone gps can tell you where you are roughly within 15 feet inside this blue circle a kind of scientific gps that we're using for warning technologies is much more precise it can tell you where you are or the position of the antenna to roughly within an inch this is about the size of a quarter coin so it's very different it's also more expensive than what you have on your phone but it turns out to be extremely useful so japan during that 2011 earthquake actually had a very dense network of these high quality gps stations but it wasn't at the time being used for any kind of warning however we can study the data to see how it could have been used had it been available in real time to the to the warning groups over there this data shows what the how the gps stations move during that magnitude nine earthquake here on the left we're watching the horizontal motion of the entire island of hanshu in japan these coastal sites are moving almost five meters they're lurching five meters towards the ocean during the earthquake here on the right hand side you have the vertical motion of the gps stations and you can see that these coastal sites subsided by about a meter now what's important is that you don't need to be a genius seismologist to figure out that because of the way these arrows are distributed the most likely area to have ruptured during the earthquake is more or less there the hyperseners the red star so this kind of analysis where we take these patterns of arrows to figure out how big the earthquake is is what we really need to be doing in cascadia and elsewhere and we know for fact that this works because we've retrospectively looked at other big earthquakes around the planet to see this kind of gps approach with work what i'm showing you here are the gps magnitude calculations if we take these stations for many earthquakes in japan here's an earthquake in chile over here's an earthquake in costa rica for example the dashed lines are the true magnitudes of the events and the blue lines are how we predict the magnet what we predict the magnitude to be using these gps stations for that japan case by about a hundred seconds after the start of the earthquake that number is seconds we would have known that it was a magnitude nine so we completely avoid the issue of saturation in chile for this magnitude 8.8 earthquake by about 50 seconds we would have known it was an 8.8 another example here the nicoya earthquake in costa rica by about 20 seconds we would have known it was a magnitude 7.6 so from these kinds of tests we know that this approach of using gps works but gps is more than just a magnitude ruler it actually can tell you what kind of earthquake is unfolding here i've calculated for you what the gps vectors would look like if you had stations everywhere on the surface of the earth for what we call a strike slip earthquake this is a lateral faulting event or one block of the fault moves laterally with respect to the other side of the fault this is what we get in california with a san andreas fault for example the vectors would have this particular pattern on the surface of the earth this is the horizontal motion of the gps stations if we now look at a similar size magnitude event same magnitude magnitude 7 but now it's a thrust earthquake where one block of the fault moves above the other block of the fault this is the kind of event that we get in cascadia this is what a magnitude nine in cascadia would do the pattern of the arrows is very very different so just by examining these patterns we would be able to say not just the magnitude but what kind of event we are contemplating or is unfolding in real time and that turns out to be very important because both of these kinds of events make very different vertical deformations so this strike slip event if you ask how it moves the surface of the earth up and down you'll get a pattern like this one where the vertical deformation is very small less than five centimeters and only concentrated on these small narrow areas that same magnitude 7 event if it's a thrust earthquake can actually have a substantial vertical deformation of the surface of the earth now if you put the ocean over this vertical deformation this is what makes a tsunami this moving up and moving down of the water column will generate the destructive tsunami waves so it's really important to know not just the magnitude but the kind of earthquake that is going on and gps can do that for us now we know that this is more than an academic exercise just this year on in january in 2018 we had a magnitude 7.8 earthquake in alaska that happened right here very close to the trench if you're in a tsunami warning center and an earthquake happens there you get really worried because we know that these earthquakes have made very large tsunamis in the past it turns out that this earthquake was one of those lateral strike slip events so it really didn't make a big tsunami because it mostly moves the earth side to side not up and down so that made less than 30 centimeters of tsunami impacts in the islands closer to the earthquake same magnitude event magnitude 7.8 but in 2010 in indonesia that event because it was a thrust earthquake and because it was very shallow that made a 19 meter tsunami much much bigger that had substantial effects on the islands here in indonesia and also on the main islands of sumatra so we need to know the kind of earthquake not just the magnitude now in the western us i think we're in pretty good shape this is the gps network there's actually not every single gps station most of them this is what it looks like today so we we're actually poised to take these kinds of algorithms that i just described that can identify magnitude and kind of faulting we can use them now today to issue the kinds of local warnings that we need there's actually also stations on vancouver island from the canadian geological survey but they're not pictured here so we're in good shape so for challenge number one which is to measure accurately a gigantic event in real time i think that is more or less solved now we need to complete the the road and going from research to operations the challenge the second challenge which is forecasting the tsunami amplitude in that quick time once we know the magnitude and the style of faulting of the the earthquake we're also doing great there so this is a tsunami model from again from the pacific tsunami warning center they have a code called rift rift needs to know the magnitude and the kind of earthquake that you're dealing with and it can very quickly estimate what the tsunami will be and this is for a scenario earthquake in the cascadia subduction zone and these are the amplitudes that rift is predicting all across washington oregon bridge columbia california and down into mexico you can see the scale of the tsunami over here on the left so these models can actually run quickly once we know what the earthquake is doing now how good are they how good are these kinds of forecasts yes we can make them but do they compare to what the actual tsunami does in real life well that's actually an open question that we're working on i show you here an example of an earthquake in chile this is that magnitude 8.8 earthquake in chile it made a big tsunami this black line here shows the amplitude of the tsunami everywhere along the coast for the chilean earthquake it's our best reconstruction of what actually happened in real life during the tsunami over here on the right is what the forecast would have been in two to three minutes after the start of the earthquake using these kinds of gps methods you can see that the pattern of the black lines is not exactly right but overall the peak amplitudes are fairly well matched the other thing that we have done here is we've colored every county they're called regions in chile every county by its corresponding warning level and what ends up happening is that even if you don't get the details quite right for every single wiggle here of the forecast amplitude the levels of warning are still okay because we can we only need forecast forecasts across an entire region not necessarily at the level of hundreds of meters for example but this isn't always the case in that earthquake i think we do okay here's an earthquake where things are a little trickier this is another earthquake in northern chile magnitude 8.3 again this black line is our best reconstruction of what actually happened during the real tsunami and here's that black line for what we can actually do with these rapid gps models you see that we underestimate the tsunami in this region compared to what happened in reality now the colors or the level of warning that each region would receive they're not so far off this region right here still gets red which is the correct level of warning but overall the warning is underestimated in other parts of the country for this particular event so gps driven models of tsunami inundation will be fast and in my opinion they'll probably be good enough but some level of uncertainty will always remain with these kinds of models there were always we will always have some we will have missed some details of the earthquake source that we don't get quite right however i think that's okay so i think that more or less solves challenge number two which is to forecast the tsunami obviously we can do more work to make it better but i think we're on a good path now the final challenge for for this is challenge number three which is because these events are frequent how do we test our systems and and this is what we do we create scenario events this is a scenario event for the cascades abduction zone you can watch the event as it ruptures that's what this guy moving here is showing and we generate scenario gps data you can see these arrows as they unfold we generate thousands of these scenarios and we put them into our real-time system and we ask how well do we do well that's shown down here the real magnitude of the scenario event is this dashed line the red line is what the magnitude calculation is for a seismic only system it exhibits that saturation that we're now familiar with but the blue line is how well the gps system would have done for this particular event and you see that by about 70 seconds we get to the correct magnitude which is good news um for tsunami warning we do this for 1300 events and this is work done by christine rule so what you're seeing here are 1300 events the real magnitude scenario magnitude and the magnitude from the rapid warning system and you can see that overall you want to be close to this black line which means you're calculating the magnitude correctly and we're in good shape we're usually within plus or minus 0.3 magnitude units of the simulated magnitude so that is good news for cascading now as i said we will always miss some details this is one of those scenario events earthquakes are are messy they have some parts of the fault move more than other parts of the fault and that's what is shown here you will concentrate motion in some parts over other parts this is what the predicted earthquake should look like from the real time system so what we can do in real time is low resolution and blocky and we can't get all the details quite right and as a result our tsunami forecast will also be uncertain and that is something that will be difficult to overcome so those are the three challenges and we have good solutions for them and i'll finish and i'll wrap up with this i think overall the outlook is quite positive with continued effort for cascadia we will have warnings for tsunamis in under five minutes all the key components are there both gps and tsunami modeling and we're working closely with the tsunami warning centers to make the transfer from research to operations so we're almost done but not quite because of these details these complexities real earthquakes we're an earthquake of one single magnitude can make tsunamis that range from 30 centimeters to 19 meters looking at tsunamis from on land with gps will always be inherently uncertain because we're not measuring the actual tsunami itself if we're uncomfortable with that uncertainty the only real solution is to go offshore and that's what japan has done japan has installed this beautiful network of offshore sensors offshore hanshu and these guys they measure the tsunami directly so all that ambiguity with what the earthquake is doing goes away because we're measuring the tsunami and that's what we're interested in these are real-time cabled it's the fastest most reliable way to issue local warnings it's also incidentally very useful for basic science but it's very expensive for japan it was 500 million dollars to build this out and roughly five million dollars a year to run it this is a proposal for that what that would look like in cascadia put forth by a group at the university of washington a cable comes offshore and goes up and down cascadia with nodes of stations for everybody to use for research and for warning now is there a sufficiently relevant societal impact to warrant this expenditure we're likely looking at something in the neighborhood of one billion dollars to build something like this out to my mind the answer is yes but that's obviously a very complex discussion which was which we still need to have so we'll finish there thank you i want to thank both of our speakers you guys did a great job and i think we're going to get a slide up here now that gives you some guidance about how to submit questions we have a few minutes for questions i'm going to start out with a couple that i are that we have received earlier um and then we'll take them from the audience uh richard first questions for you what is you made tremendous progress with shake alert but what is the remaining major challenge you still face to getting to the point where you can go public with your alerts on a wide scale yeah that's a great question there's two pieces i would say one is of course that the system is not fully built as i was sort of laying out the in terms of the seismic stations only 50 percent of the stations have been put out so far we need to add the other half however as i also showed the system is running very well in regions that have the stations that are pretty closely pretty dense already which is primarily the urban areas on up and down the west coast so i mean i'm very excited about the possibility of the limited public roll out later this year i think the greatest challenge frankly is fear of screwing up and we as a seismology community really have to be ready to kind of take that extra step as a community is of scientists it's not like engineers engineers have always had to design bridges or buildings um that that can withstand earthquakes and they used to that as seismologists i think we're less ready for that we're less used to that and i think that there's a lot of concern um amongst seismologists about are we ready to do this we might make a mistake the system might make a mistake and and people are very concerned about that and for good reason it's good to be to recognize that but i think that we really have to get past this we have to start delivering the alerts the system is working pretty well what we've learned from mexico is that the the receivers of the information the public they understand that there are technical limitations to these systems and they recognize that the system is not perfect but they still see it as being valuable that's what we've learned from mexico so we should use this to boy us as a as a community the seismology community to move forward we obviously need to make sure that there's good education of the users but then we need to just start start doing this and and actually get this running so i think that that's the greatest challenge is is taking that step and and and just getting started thanks geego a question for you you've shown us an exciting new approach to uh quantifying uh tsunami predictions actually why an earthquake is occurring how does this approach particularly with a forecast of wave heights compare to the conventional approach that we've been relying on that's used to create the flood inundation maps that communities use for their their emergency planning it's it's very similar and it uses shares many of the same technologies to make to make these hazard maps we basically propose scenarios or we look at earthquakes from the past and we make very painstaking models of the source with as many details as possible we simulate the propagation and we do this hundreds of times maybe and we come up with these hazard maps we're using many of the same technologies except we're doing it as quickly as possible with imperfect models with big uncertainties so the technologies are shared um but the information is not always shared for example a level of response that a community should have to warning should likely be tailored to what the expected hazard is and i'm thinking of places like crescent city here in california they know that even the smallest tsunami is amplified in the crescent city area so they should take action at lower levels of forecasted wave heights than other parts of the coast for example so the really hazards and warning are a continuum we need to be working before the hazard we need to be working during the hazard and after the hazard and what we're discussing here is only the during portion of that hazard but again work needs to be done both before and after here's a question came in early on the really for both of you how clearly the subduction zone extends further north than than the u.s into into vancouver island area how coordinated are we with our colleagues in canada in extending something like shake alert or the tsunami warning system that dago talked about to cover the totality of the zone at risk so i can answer the seismic piece first perhaps so as you saw on my map you may have noticed the map when i showed the seismic stations that are feeding data into shake alert that station coverage actually extends up into canada and obviously the geographic sorry the political boundary is not the relevant boundary when it comes to the hazards and that's why data from the canadian stations is also streaming into shake alert and so what that allows is for shake alert to actually be able to detect earthquakes just north of the border and include that information and i believe that there is a sort of similar kind of discussions about using the warnings using the information that comes out of shake alert and to issue warnings north of the border because the usgs is the alerting agency for earthquakes here in in the u.s and so they they will issue alerts in the u.s but it will be up to the canadian government about issuing alerts north of the border but in terms of the science and the data already we're crossing the border pretty seamlessly i think i think that's also true of the gps network the collaboration with the canadians is is good it can always be better but it's good i think the real challenge for gps or the next step is that this kind of technology can be used by the tsunami warning centers on a global scale there's actually gps networks in many many parts of the world but it's a very real challenge to get countries to open up their gps data to the tsunami warning centers and it's strange because they already do it for the seismic data so we still need to do some lobbying and some teaching and education and work with other communities to get them to open up and understand that there's a mutual benefits to sharing the data thanks i have a couple questions here one for each of you uh probably came from folks who live in the uh puget sound in willamette valley areas richard how how much warning time would we have for a cascadia breakout mega quake for the folks who live in puget sound and the portland willamette valley area so the amount of warning for earthquake early warning is a function of the distance that you are from the epicenter so if the earthquake starts a long way away such as for when we're talking about puget sound for example if the earthquake starts down at the southern end of the cascadia subduction zone it's going to take about five minutes for that rupture to come all the way up um the subduction zone now one of the issues one of the science questions that we are still working on is how quickly can we really determine um how strongly shaking is going to be for you in puget sound if we see an earthquake starting to the south so we can issue a warning immediately to say that there is a significant earthquake and this may uh mean that there is just moderate shaking for you or it may mean there's very strong shaking and in that case there could be minutes worth of warning if we want to issue alerts um saying that there's going to be very strong shaking the warning may be less but so in terms of people thinking about warning times we're talking about seconds tens of seconds best case scenario is a few minutes but that we should be thinking about how we can react with warnings of a few seconds and and to build on that are uh to to Diego Diego some questions about the potential impacts of the tsunamis extending particularly in uh the Straits of Wanda Fuka into Puget Sound potentially impacting those communities in there any thoughts on that yeah that's ongoing work and mapping out the hazards of what will happen during the next big event and it's actually quite interesting the the amplitudes are probably not going to be incredibly high tsunamis have a hard time getting into these shallow narrow channels but you could expect some really strong currents sometimes for several days um after the tsunami so you wouldn't I don't think you wouldn't see some of these like 30 meter inundations that we see in places like Japan and Indonesia within Puget Sound but you would see some you know meter level waves and definitely incredibly strong currents for quite a long while well one more question for again for both of you clearly the costs in some of the uh uh programs that are being talked about shake alert you know well on its way but it's still not not cheap and if we go to uh some of the technologies Diego talked about for the offshore quite expensive uh what new technologies are out there that might help bring some of these costs down that could uh help make uh move this along well I mean I first of all I would say that um I'm not sure that this is particularly expensive I mean the shake alert piece that the sort of 16 million dollars per year to operate for example it's about twice what we currently spend or what we spent on seismic network monitoring previously previous to earthquake early warnings so we're talking about sort of doubling the operational costs of the seismic networks in order to provide warnings for everybody so I would challenge the idea that that's not particularly cheap still that it's true that there are other technologies um we are here at Berkeley for example we're starting to explore the use of smartphones to detect earthquakes now there's no way that the cell phone detectors will ever replace the quality of the traditional seismic networks that shake alert currently uses needless to say people are carrying them around but we do think that we can use detection from smartphones to detect earthquakes and then the logical next step once you can detect earthquakes is can you provide warnings and that's something that we're exploring we think that there is value for it um particularly of course in places where there are no traditional seismic networks but right now there's no question that's a research effort to be clear that's research big bright line shake alert is a proven technology shake alert can actually deliver warnings to people and and that's why we're moving forward with shake alert well thanks thanks thanks a lot I unfortunately I think that's all the time we're going to have for for the questions today really appreciate the time you spent with us as a reminder to the audience the first two parts of this series are currently available on the Beezer website and this third webinar will be posted within the next week or so so check it out and with that I'd like to thank our two speakers Dr. Drs. Allen and Melger for their presentations and thank you all again for participating have a great day