WEBVTT 00:00:00.000 --> 00:00:05.000 [noise] 00:00:05.000 --> 00:00:20.000 Welcome back everyone! [WooHooHoo] We have another exciting session to share with you now. Earthquake Early Warning around the world and back to northern California and to take us on this journey through time space and 00:00:20.000 --> 00:00:27.000 warning we have the wonderful Sara McBride and Jessie Sanders take it away moderators. 00:00:27.000 --> 00:00:41.000 Thank you so much, Sarah. Our first speaker today is Mitsuyuki Toshiba from the Japanese Meteorological Agency (JMA) and he will be speaking on a review of nationwide Earthquake Early Warning in Japan, 00:00:41.000 --> 00:00:52.000 15 years of operation by JMA. 00:00:52.000 --> 00:01:04.000 Thank you very much for inviting me to this important meeting. So, in Japan, nationwide Earthquake Early Warning is being operated by Japan Meteorological Agency, JMA, since October, 2007. 00:01:04.000 --> 00:01:12.000 So JMA is a national government organization in charge of national government organizations of the weather forecast, climate change, earthquake observation, tsunami 00:01:12.000 --> 00:01:13.000 early warning, volcanic observation, and earthquake early warning. So, JMA 00:01:13.000 --> 00:01:24.000 issues warnings nationwide and issues to general public using TV, video, radio, smartphone broadcasting. 00:01:24.000 --> 00:01:30.000 I will show an example. 00:01:30.000 --> 00:01:32.000 This is the April... I 00:01:32.000 --> 00:01:40.000 this is, April 14, 2016, magnitude 6.2 event. This is the 2016 Kumamoto earthquake event. Origin time is at 9:26 p.m. 00:01:40.000 --> 00:01:53.000 Because of 9 o'clock in NHK public broadcast TV channel broadcasted news program and here, I will show that the air-time ShakeMap using data estimation. 00:01:53.000 --> 00:02:23.000 Okay, shall we go. [video in Japanese] 00:02:28.000 --> 00:02:41.000 [video in Japanese] 00:02:41.000 --> 00:03:11.000 [video in Japanese] 00:03:37.000 --> 00:03:53.000 Here I'd like to expand our JMA seismic intensity scale. 00:03:53.000 --> 00:03:57.000 JMA assessment intensity ranging from 0 to 7. Unseismically tested 5 00:03:57.000 --> 00:04:00.000 and 6 are divided by 2. That is [indiscernible]. 00:04:00.000 --> 00:04:14.000 So, total 10-degrees are used to represent strength of strong ground shaking. So this is the JMA scale, and MM scale, and MSK scale. 00:04:14.000 --> 00:04:20.000 So there are two category of JMA EEW one is called "Forecast" the other is called "Warning." Forecast is the message for advance users. 00:04:20.000 --> 00:04:32.000 Advanced user means railway company, lifelong companies, and the data company, something like that. 00:04:32.000 --> 00:04:43.000 On the "Forecast." The forecast is in case seismic data is 3+ or larger predicted and the radio system, 4 or larger are specified in the forecast message. On the warning. 00:04:43.000 --> 00:04:52.000 "Warning" is a service for general public. So, seismic tested low or larger is a predicted "Warning" message is issued. 00:04:52.000 --> 00:05:00.000 So this warning message broadcast by TV, radio, smartphone, mail. So here I'd like to show the important sound. Okay. 00:05:00.000 --> 00:05:10.000 [video audio] This means TV and radio receiving earthquake early warning message. 00:05:10.000 --> 00:05:15.000 [video audio] One more important sound. 00:05:15.000 --> 00:05:23.000 This means not all receiving earthquake early warning message. So to see it, a new number of forecasts and warning. 00:05:23.000 --> 00:05:39.000 So regarding forecast, usually about to 10 or 20 warnings issued in a year and major [indiscernible] co-card such as Tohokaskik and the Kolmotoaski, the number of earthquakes increased. 00:05:39.000 --> 00:05:42.000 So this is the history of the JMA EEW. 00:05:42.000 --> 00:05:51.000 Seismic observation and the development of algorithm from 2010, ocean bottom seismometers and the 00:05:51.000 --> 00:05:57.000 deep borehole reservation are incorporated and, algorithm IPF, so the [indiscernible] are introduced. 00:05:57.000 --> 00:06:04.000 So next to right, I will show up the Tohoku earthquakes and the blueprint of IPF on the [indiscernible]. 00:06:04.000 --> 00:06:11.000 So this animation shows the performance of the JMA EEW, during monitoring M9 Tohoku earthquake. 00:06:11.000 --> 00:06:17.000 So here I will show the last time from origin time, pass to Forecast, and pass to Warning. 00:06:17.000 --> 00:06:35.000 So for this event, totally, 15 messages issued and here shows the largest estimate of moment. So, [indiscernible] forecast ratio, which means seismic data predicted, and the pink range a mere warning region, which means that seismic 5> or larger is predicted. 00:06:35.000 --> 00:06:42.000 So I will show the waveform of without G4 stations. Okay, shall we go? 00:06:42.000 --> 00:06:47.000 Earthquake go-card and the P-wave, Hesuvik, Vladimir Nuzza source. 00:06:47.000 --> 00:06:57.000 On the past station, [indiscernible] disaster period and the past 4 gas tank warning issued. 00:06:57.000 --> 00:07:01.000 [video audio] Telephone, TV, broadcasting earthquake early warning message. 00:07:01.000 --> 00:07:10.000 Still weak shaking. Past strong shaking. Past the point. Appeared from now on so, JMA EEW worked well. 00:07:10.000 --> 00:07:19.000 Message is about 50 seconds before strong shaking. So this is the procedure assurance, monitor estimation 7.7. 00:07:19.000 --> 00:07:28.000 [indiscernible] non-assurance should estimated 7.9. 00:07:28.000 --> 00:07:36.000 [indiscernible] magnitude, 8.0. 00:07:36.000 --> 00:07:41.000 14 assurance magnitude 8.1. 00:07:41.000 --> 00:07:53.000 For 15, this is a last message. So far, so good. But problem [indiscernible] from now on, forecasting around here, [indiscernible], back to ground 0, which means seismic tested. 00:07:53.000 --> 00:08:03.000 Predicted but actually that the point appeared like this actually seismic 6 up 6 lower observe it. 00:08:03.000 --> 00:08:09.000 So under prediction or counter region, this is due to the large extent of source rupture of the magnitude 9 event. 00:08:09.000 --> 00:08:20.000 This is a past program, and the second [indiscernible] during aftershock period. This show the seismic activity of March 4th, so this is no more station. 00:08:20.000 --> 00:08:27.000 And much nicer. Watch this. 00:08:27.000 --> 00:08:35.000 [video played] Mainshock. Aftershock. Aftershock activity, quite, quite active. So sometimes, multiple aftershock occur simultaneously. 00:08:35.000 --> 00:08:50.000 In this case, system is interpreted as the multiple simultaneous event to be a large single event, which due to over estimation, magnitude and [indiscernible]. Interestingly, the problem is our important project and IPA Method was incorporated into JMA EEW in 2016. 00:08:50.000 --> 00:09:03.000 But before IPA Method, using the arrival time difference with P-wave, hypocenter, determined. Using Hypocenter and using amplitude 00:09:03.000 --> 00:09:14.000 magnitude estimated, so two-step approach. IPF Methods 00:09:14.000 --> 00:09:18.000 [indiscernible] also used or high standard examination, by doing so, we can avoid the mislocation of a hypocenter. 00:09:18.000 --> 00:09:29.000 Here, I'm pretty sure it's very small. And the PLUM method is a simplified version of ground-motion-based method that is what incomplete in the JMA in 2018. 00:09:29.000 --> 00:09:37.000 So it is very simple. So, actual detection of a storm shaking nearby station in this case, warning issued. 00:09:37.000 --> 00:09:48.000 In this case, the station, storm shaking, and warning issued. So actually, 10 seconds duration, strong ground shaking, alive, but [indiscernible] 00:09:48.000 --> 00:10:01.000 So no missed alarms and we don't need to wait 3 seconds or [indiscernible] which needs to happen to earthquake early warning. 00:10:01.000 --> 00:10:12.000 I will show this right, and this is real-time data and processes for operations of JMA EEW. 00:10:12.000 --> 00:10:22.000 In September, last year, four months ago arrived but [indiscernible]. 00:10:22.000 --> 00:10:24.000 So high note data is also integrated to IPF message and we course, GMPE-extended IPF method. 00:10:24.000 --> 00:10:41.000 So IPF-extended IPF Method for source-based method and from method where we preview the best method of the ground-motion-based method, working partly at the present. 00:10:41.000 --> 00:10:52.000 Okay, that's all. Thank you very much. 00:10:52.000 --> 00:11:08.000 Thank you very much. Our next speaker is Alan Husker from Caltech who will be talking about earthquake early warning in central Mexico, "The Seismic Alert System." 00:11:08.000 --> 00:11:19.000 Hello, my name is Alan Husker. I'm going to talk today about the earthquake early warning system of Mexico, which is called the Sistema de Alerta Sismica Mexicano (SASMEX) 00:11:19.000 --> 00:11:30.000 and a lot of the information comes from Armando Cuéllar Martinez thesis from 2018 when I was a professor at UNAM and I was on his thesis committee. 00:11:30.000 --> 00:11:37.000 It should be noted that there's a lot more seismic activity in Mexico than in the U.S., in the contiguous U.S. at least 00:11:37.000 --> 00:11:50.000 and so that allows for a lot of training sets, have bigger earthquakes a lot of them, and so Mexico on average has a magnitude 7 or above about every 1.6 years 00:11:50.000 --> 00:11:59.000 and in the contiguous U.S. it's about every 15 years and so they just have a lot more earthquakes, 92, 5's and 6's are very common along the coast. 00:11:59.000 --> 00:12:04.000 So the overall majority of the earthquakes are coming from the subduction zone. There's some 00:12:04.000 --> 00:12:18.000 further inland including the September 19, 2017, which caused a lot of damage in Mexico City. Just to give you an idea over time, this is all earthquakes from 1900 to the present 00:12:18.000 --> 00:12:25.000 and actually now past 2020 so there's been some more; magnitude 7's and above. 00:12:25.000 --> 00:12:35.000 This comes from an article that I wrote for a journal that's similar to Scientific America in Mexico. 00:12:35.000 --> 00:12:47.000 So part of the reason that this is so relevant is this is Mexico City. It was built on top of a lake bed and now there's about 20 million people living in the zone and so this is what the lake used to look like; 00:12:47.000 --> 00:12:54.000 it's been drained for the most part. There's a little bit left down here in Xochimilco, but for the most part it's just the city on top of it. 00:12:54.000 --> 00:13:13.000 And like I said, there's about 20 million people living in the zone. To give you an idea of the amplification site factors; down here, this station here comes from this gray zone which is kind of the rocky zone and then this comes from the middle of what used to be the lake bed, which is right here, which is close to downtown. 00:13:13.000 --> 00:13:19.000 And you can see the difference in amplification; and there's a 5km difference. This is actually further away from the earthquake. 00:13:19.000 --> 00:13:23.000 This is from the 1985 earthquake that was a magnitude 8.1 00:13:23.000 --> 00:13:40.000 you can see the difference in amplification between the two, and also the extended CODA is quite a bit different, for this station that's in the middle of the basin compared to one that's in the hard rock site. 00:13:40.000 --> 00:13:47.000 And, just a kind of further evidence of how important this is in 1936. 00:13:47.000 --> 00:13:57.000 This is just one example that there was this well head that was put into place and this used to be the surface at that time and so the ground sank and this photo is actually a few years old. 00:13:57.000 --> 00:14:02.000 This means there's about 8 years of subsidence that's occurred since 1936 00:14:02.000 --> 00:14:10.000 to the present, and so it's just a large amount of sedimentation and huge side effects that are there. 00:14:10.000 --> 00:14:16.000 And the side effects actually have changed for the years as well, which is interesting. But, to the side, just again, kind of reiterating the side effects and how big they are in Mexico City. 00:14:16.000 --> 00:14:27.000 You can see the variations in PGA here across this image. They vary enormous amounts. This is PGA. 00:14:27.000 --> 00:14:41.000 And over here we have spectral amplitudes and at 1 second you can see 1 second amplification here is getting it to be above 600 times what it would be and look at hard rock locations. 00:14:41.000 --> 00:14:49.000 SASMEX has an interesting history, so it was installed in 1991 along the coast. 00:14:49.000 --> 00:14:57.000 There was only 12 stations kind of in this area here, and at the time, well, it was based on technology that was available at the time. 00:14:57.000 --> 00:15:06.000 So no cell phones, it was all point to point, radio hops, and so the black dots are the seismometers, 00:15:06.000 --> 00:15:17.000 and the triangles are all those radio hops and so hopped up to Mexico City up here was the final hot point and then went to alert the city from there with radios. 00:15:17.000 --> 00:15:25.000 These are the radio that's emit the alert and so they have these posted around the city, particularly in schools. 00:15:25.000 --> 00:15:31.000 It wasn't very well funded when it was first installed. And also what was interesting it was basically a private company that runs it, 00:15:31.000 --> 00:15:40.000 CIRES is the name of the company, which is called the Instrumentation and Seismic Record Center in English. 00:15:40.000 --> 00:15:51.00 Which is a nonprofit NGO, but it was originally set up with the help of professors at UNAM and then a little bit kind of forgotten about for many years and had some funding problems 00:15:51.000 --> 00:15:57.000 that was retaken up after about 10 years or so and then it got better funding; and then, built out 00:15:57.000 --> 00:16:14.000 across southern Mexico. What that also did since it's a nonprofit NGO, they were kind of left to make the system on their own and 00:16:14.000 --> 00:16:30.000 they kind of had little oversight, which means they can make decisions and build out things but then there has not been a lot of questioning of the system, but people seem happy with it overall. 00:16:30.000 --> 00:16:43.000 One interesting thing is their pay structure they don't get money for directly from the federal government they get it from the different cities and so Oaxaca, actually the city stopped paying them for a while and they shut off the alarms in Mexico City for a little bit and there was actually an earthquake that occurred during the time so it was not alerted. 00:16:43.000 --> 00:16:52.000 This was about 10 years ago. 00:16:52.000 --> 00:16:58.000 There's kind of an interesting alert strategy and so they alert for anything that's above magnitude 5. 00:16:58.000 --> 00:17:08.000 With 250 km of the city and then they have different thresholds if it's further away from the city. 00:17:08.000 --> 00:17:19.000 Let me get into the algorithm real quick. Basically, there's two steps to the algorithm and this is there's different versions of the algorithm but this is for anything coming from the coast. 00:17:19.000 --> 00:17:26.000 Start implemented other algorithms but they're based on the same idea for earthquakes coming more inland. 00:17:26.000 --> 00:17:40.000 So the trigger basically comes from his VO or H parameter. They take basically the square of the verticals versus the square of the one vertical versus the square of the horizontals. 00:17:40.000 --> 00:17:46.000 And then get this perimeter V over H and so that it's high enough than it triggers. 00:17:46.000 --> 00:18:00.000 Then they take this summed square of the waveform, and that's "A," which is the sum square of the waveform, and then "M" is the final value at the end of the area underneath the line here and they take these two values 00:18:00.000 --> 00:18:14.000 and they've come up with this "fit to magnitude" so "M" is the log of the value here and "A" is the log of the area under the curve here, 00:18:14.000 --> 00:18:24.000 and you can go into more detail, and get this from Cuellar et al., 2017, but basically between each one of these lines is magnitude 4.6, 4.8, 5.0, 00:18:24.000 --> 00:18:30.000 they fit to these two values and it's a simple least squares fit. 00:18:30.000 --> 00:18:44.000 They've also done some more machine learning, but basically it's still a very similar fit, and so if it's to one or side of the other of this line of these two parameters, basically it's when the alarm will go off. 00:18:44.000 --> 00:18:53.000 The way they evaluate the algorithm is a bit different than the way we evaluate the way the algorithm is evaluated by ShakeAlert. 00:18:53.000 --> 00:19:01.000 And so, first of all, they have different algorithms I should mention, they have gone for faster ones. 00:19:01.000 --> 00:19:11.000 So this one is how they look at two times the TS minus TP time. So if you look at this over here, this is 2 times the TS minus TP time. 00:19:11.000 --> 00:19:16.000 Now they have other algorithms that are TS minus TP, which is a lot faster. 00:19:16.000 --> 00:19:22.000 And then they have TP, which is the P wave, arrival time plus 3 seconds, and so it becomes very fast. 00:19:22.000 --> 00:19:36.000 And so here's some different examples of when alerts could be issued using this faster algorithm. The evaluation here, so along the bottom in these two cases, this is the TP plus 3 and this is TS minus TP. 00:19:36.000 --> 00:19:48.000 But this is the way they evaluate this and so they have their magnitude versus MW and you can see in this case of magnitude 8.1 here, this is considered a success for the ShakeAlert system. 00:19:48.000 --> 00:19:55.000 It would not be considered a success, but they would consider it a success because it was alerted when it was above magnitude 5. 00:19:55.000 --> 00:20:05.000 But it's very far from the actual magnitude 8.1 earthquake and again I just throw up this values here to see that. 00:20:05.000 --> 00:20:17.000 Some final thoughts on SASMex. It's a very simple complete system with long-warning times, the location of earthquakes with special condition in Mexico City, being so far away, it emphasizes speed over accuracy. 00:20:17.000 --> 00:20:21.000 There's no cell phones involved, it's only alerts over a loudspeaker system. 00:20:21.000 --> 00:20:22.000 The alerts are not analyzed with the same categories of ShakeAlert or the rest of the Google community. 00:20:22.000 --> 00:20:39.000 For example, there's no evaluation of overloading for timeliness and any time there's an alert issued it's considered a success. 00:20:39.000 --> 00:20:54.000 Anyway, so that's something to think about. One other thing that's interesting is I would say that from my experience working there, one of the important things we consider, the aftershocks, what happens for learning aftershocks. 00:20:54.000 --> 00:21:01.000 People respond to the aftershock alerts much more than they do any other alert because now they've had the main shock. 00:21:01.000 --> 00:21:05.000 And that can lead to injury and possible death. And so it's something to consider what should happen in the aftershock sequence. 00:21:05.000 --> 00:21:14.000 And I don't think there's been enough research on that other than my anecdotal evidence right now. 00:21:14.000 --> 00:21:18.000 And I'll leave you with that. Thank you very much. 00:21:18.000 --> 00:21:23.000 Thank you, Allen. Our next speaker is Francesco Finazzi from the University of Bergamo. 00:21:23.000 --> 00:21:29.000 Francesco unfortunately could not join Zoom today due to time zone differences, but we're very thankful that he was able to provide 00:21:29.000 --> 00:21:35.000 his talk recording. He will be speaking about "Crowd-source smartphone-based Earthquake Early Warning— 00:21:35.000 --> 00:21:44.000 The Earthquake Network citizen-science initiative." 00:21:44.000 --> 00:21:58.000 Hello everyone, my name is Francesco Finazzi. I'm a professor of statistics at the University of Bergamo in Italy and I am the founder of the Earthquake Network Citizen Science Initiative 00:21:58.000 --> 00:22:10.000 That I'm about to discuss. First, I want to thank Sarah for giving me the opportunity to present my research in this workshop and as you can see from the title my research is about crowd-sources 00:22:10.000 --> 00:22:26.000 smarthphone-based earthquake early warning systems. Earthquake Network (EQN) was the first citizen science initiative to use crowd-source smartphones for the real-time detection of earthquakes. 00:22:26.000 --> 00:22:31.000 Probably I was not the first, with this idea, but EQN was the first operational system of this kind. 00:22:31.000 --> 00:22:50.000 And in fact, since 2013 the system is sending real-time alerts to people who are part of the initiative. 00:22:50.000 --> 00:22:58.000 Why smartphones? Smartphones have what it takes to detect earthquakes in real time and also to receive the alert. 00:22:58.000 --> 00:23:11.000 In fact, they have an accelerometer, operational system, and the Internet connection. The single smartphone is possibly the worst instrument to detect and measure an earthquake, but many smartphones in a network make a reliable detection system 00:23:11.000 --> 00:23:29.000 and this is why EQN works. Here we have an example of the alert. So when the server detects an earthquake and an alert is sent 00:23:29.000 --> 00:23:37.000 to all the smartphones in the area. On this smartphones you will see a map with the preliminary 00:23:37.000 --> 00:23:48.000 estimate of the epicenter, the location of the S wave, the countdown and the expected level of shaking at the location 00:23:48.000 --> 00:23:55.000 of the smartphones. Here in this picture, you can see an example of an Q and smartphone network. 00:23:55.000 --> 00:24:12.000 Each dot is a smartphone, which is monitoring for earthquakes and here we are in Santiago in Chile so where the initiative is popular, the number of smartphones in the network can be very large. 00:24:12.000 --> 00:24:23.000 I like to assess the value of the network. So currently we have around 2 million active users. So if I had to implement myself. 00:24:23.000 --> 00:24:39.000 The network, I would have to buy smartphones for €400 million euros. The Internet service would cost me €200 million euros per year and the energy to charge the smartphones would cost me €8 million euros per year. 00:24:39.000 --> 00:24:50.000 In practice, the actual cost is less than €20,000 per year. And I believe this is the power of crowd-sourcing. 00:24:50.000 --> 00:24:58.000 NQN is not the only active smartphone network. 00:24:58.000 --> 00:25:01.000 We also have MyShake by the Berkeley Seismological Laboratory. 00:25:01.000 --> 00:25:05.000 And then we have the Android earthquake alert system by Google. 00:25:05.000 --> 00:25:14.000 The Google system is an opt-out system. So it means that Google doesn't require to install an app on the smartphones. 00:25:14.000 --> 00:25:21.000 So they are able to exploit all the Android smartphones. The coverage potential is global but currently the service is not active everywhere and the alert is limited to the Android smartphones. 00:25:21.000 --> 00:25:35.000 EQN also targets iOS and [indiscernible] smartphones. 00:25:35.000 --> 00:25:45.000 To give an idea of the potential of the EQN system here we have the EQN detection of the Turkish-Syrian event 00:25:45.000 --> 00:25:56.000 of the last February. So in this map, we can see the epicenter and then each dot here is a smartphone that detected 00:25:56.000 --> 00:26:08.000 the event while squares are smartphones that didn't detect or didn't still detect the event at the time of the EQN detection. 00:26:08.000 --> 00:26:23.000 Here the detection delay was 10 seconds, so after 10 seconds from origin time, EQN sent an alert to people living in the area and we did an EQN install. 00:26:23.000 --> 00:26:46.000 Here you can also see the location of the P and S waves, and here in particular, the geometry of the network was not optimal because the epicenter was in-between two cities so the delay was 10 seconds but in general the delay is less than 5 seconds. 00:26:46.000 --> 00:27:01.000 We also wanted to compute the warning time, not with respect to the arrival of a P or S wave, but we consider the threshold of peak-ground acceleration, 00:27:01.000 --> 00:27:13.000 so in this case 12% g and then we computed at the location of our stations, the warning time with respect to the exceedance of that threshold. 00:27:13.000 --> 00:27:27.000 So for instance in Antakya the 12% g threshold was exceeded after around 60 seconds from the EQN alert and this is the warning time. 00:27:27.000 --> 00:27:33.000 Now also for cities closer to the epicenter like this city, which is only 21km from the epicenter 00:27:33.000 --> 00:27:49.000 the exceedance was close to the [indisernible] detection, but still in this city the warning was a bit higher than 50 seconds before the very strong shaking. 00:27:49.000 --> 00:27:58.000 Using this information, we computed the warning time distribution with respect to the EQN users. 00:27:58.000 --> 00:28:07.000 So here we have the warning time distribution for users. That we're exposed to intensity 9, 10, and 11. 00:28:07.000 --> 00:28:19.000 So here you can see that the warning time was up to 60 seconds. We did the same for intensity 8 and then we computed the warning time distribution for the entire population. 00:28:19.000 --> 00:28:41.000 So assuming that the alert was issued to the entire population, not only to EQN users. And here we discover that if the alert was issued to the entire population, we could have alerted 1 million people with an alert with warning time higher than 40 seconds. 00:28:41.000 --> 00:28:48.000 And nearly the same for intensity 8. EQN is also such a network and in particular, using the app, you can send the felt reports. 00:28:48.000 --> 00:29:09.000 And you can discuss the felt event in chat rooms which are available on the app. Eventually users receive a survey to fill out in particular after a strong earthquake. 00:29:09.000 --> 00:29:28.000 Using the survey, we can understand that what's the behavior of the app so we can understand for instance, what was the percentage of users who received an accurate early warning or a late warning or even a false warning for instance. 00:29:28.000 --> 00:29:35.000 More importantly, we use the surveys to understand the reactions of the account users after an early warning. 00:29:35.000 --> 00:29:46.000 And in particular, we discover that only 25% percent of the users who received an early warning reacted in the best way. 00:29:46.000 --> 00:29:54.000 And this is a problem because, earthquake early warning can be useless without proper and regular training. 00:29:54.000 --> 00:30:03.000 I must say that even myself, when I received the warning from my app, I don't react because I'm not trained to do it. 00:30:03.000 --> 00:30:15.000 And this is, I believe, even more important for opt-out systems like the Google system, because if you don't know that your smartphone has a service like this, probably you will not react. 00:30:15.000 --> 00:30:31.000 When you receive the first warning. So I believe that future research should also focus on the sociological and psychological aspects of early warning and not only to the technical problems. 00:30:31.000 --> 00:30:43.000 So this was my last slide. Here you have a list of references and if you have questions or comments please send me an email. 00:30:43.000 --> 00:30:47.000 Thanks for the attention and bye. 00:30:47.000 --> 00:30:52.000 Thank you very much. And then once again, Francesco is not able to join us 00:30:52.000 --> 00:31:11.000 in-person, so if you have any specific questions for him please use his email. Our next speaker is Elizabeth Cochran from the U.S. Geological Survey and she will be talking about ShakeAlert, an earthquake early warning system for the West Coast of the United States. 00:31:11.000 --> 00:31:22.000 First, my thanks to the conveners for the invitation to give an overview of the Shake Alert earthquake early warning system that we have here in the west coast of the United States. 00:31:22.000 --> 00:31:32.000 The goal of the ShakeAlert earthquake early warning system is to mitigate the impact of earthquakes by providing advanced warning for moderate to strong ground shaking 00:31:32.000 --> 00:31:45.000 and we use this by using a network based earthquake early warning system with a distribution of stations across an area that are used to quickly detect and characterize an ongoing earthquake. 00:31:45.000 --> 00:31:51.000 ShakeAlert was built on top of the Advanced National Seismic System. There was already an existing seismic network and we've since built out that network to really support the early warning 00:31:51.000 --> 00:32:13.000 system. ShakeAlert began in 2006 with algorithm development and by 2019 we had public rollout in California followed in 2021 by public rollouts in Oregon and Washington. 00:32:13.000 --> 00:32:21.000 This slide gives a broad system architecture for ShakeAlert where we have data coming in from our seismic stations 00:32:21.000 --> 00:32:28.000 that are sent through our servers, which are located in Seattle, Berkeley, Menlo Park, and Pasadena. 00:32:28.000 --> 00:32:45.000 That's where we do the detection and characterization of the earthquake. These solutions are then combined and there's an alerting layer which determines whether or not an alert should be released based on various magnitude thresholds. 00:32:45.000 --> 00:32:54.000 And alerts are picked up by our subscribers and distributed forward to users or to automated systems. 00:32:54.000 --> 00:32:59.000 We're gonna walk through some of these steps by looking at the magnitude 7.1. Ridgecrest earthquake. 00:32:59.000 --> 00:33:08.000 So on the left here I'm showing 2 plots on the top is the magnitude estimate for the Ridgecrest earthquake through time. 00:33:08.000 --> 00:33:23.000 After earthquake origin and the bottom plot is the location error. The different symbols indicate the different algorithm solutions, Epic and Finder, and then the combined solution aggregator solution in the black squares. 00:33:23.000 --> 00:33:32.000 So in this case, you can see that the earthquake is relatively rapidly detected as a magnitude 5.5 initially 00:33:32.000 --> 00:33:45.000 that magnitude estimate goes up to 6.4 and then levels off this is in part due to some telemetry delays from some of the stations so that data were incomplete. 00:33:45.000 --> 00:34:04.000 In addition to those magnitude and location estimates, Finder also provides a line source. So this is an estimated extent of the fault rupture, which is really important for getting accurate estimates of near-field ground motions close to where the rupture is happening. 00:34:04.000 --> 00:34:18.000 So the Finder and Epic Solutions are then combined in the alert layer. This layer creates the alert messages which include both the earthquake location and magnitude but also polygons of expected shaking. 00:34:18.000 --> 00:34:31.000 So if you're interested in shaking of MMI 5 you just look at whether or not you are inside of that MMI 5 predicted shaking polygon. 00:34:31.000 --> 00:34:49.000 Once an event is detected and we create our polygons, we then have to decide. Who those alerts are sent to in general we use magnitude both a magnitude threshold and an intensity threshold to decide when and where to distribute alerts. 00:34:49.000 --> 00:35:02.000 The magnitude threshold that's used for most cell phone apps is magnitude 4.5 and we use a slightly higher threshold for wireless emergency alerts of magnitude 5. 00:35:02.000 --> 00:35:10.000 So let's look at some statistics, since public rollout, so from October of 2019 through September of 2023. 00:35:10.000 --> 00:35:25.000 This is based on some work by Angie Lux and others that's currently in review. So during that time period there were 53 total magnitude 4.5 earthquakes in the NSS catalog. 00:35:25.000 --> 00:35:35.000 Of these ShakeAlert detected about 77% of those correctly. An additional 10% were what I call a "missed alert." 00:35:35.000 --> 00:35:44.000 This means that ShakeAlert did detect an event, but estimated the magnitude to be less than magnitude 4.5 and so these would not have been distributed using that alerting threshold of 00:35:44.000 --> 00:36:02.000 magnitude 4.5 or greater. Additionally, there were about 13% missed events, these are events where ShakeAlert didn't have a detection within 100 km or 30 seconds origin time, 00:36:02.000 --> 00:36:11.000 and these tend to be offshore events or edge of network events where station densities are lower, 00:36:11.000 --> 00:36:26.000 and may not have the significant ground shaking within the area of interest. We can also look at the performance from the perspective of the ShakeAlert magnitude 4.5 and larger detections. 00:36:26.000 --> 00:36:38.000 There were 95 of these total. Of these, 41 events were correctly estimated by Shake Alert to be magnitude 4.5 or larger, 00:36:38.000 --> 00:36:41.000 as the catalog event was as well. About 50% of these are what I call a "necessary alerts." 00:36:41.000 --> 00:36:55.000 So this is where ShakeAlert estimated the magnitude to be 4.5 or larger but they were associated with earthquakes smaller than that. 00:36:55.000 --> 00:37:05.000 In the end, the ground motions are probably similar between what ShakeAlert predicted and what the true ground motion were, 00:37:05.000 --> 00:37:11.000 so this probably isn't a huge deal in terms of alerting folks. And then there are about 7% "false alerts." 00:37:11.000 --> 00:37:23.000 These are ShakeAlert magnitude 4.5 and larger detections with no matching in ANSS event within a 100 km or 30 seconds. 00:37:23.000 --> 00:37:38.000 This can happen either from noise on stations or sometimes more often kind of a mischaracterized event where for example an offshore event is estimated to be farther offshore than it should be. 00:37:38.000 --> 00:37:45.000 So this often happens in kind of areas where the station density is sparser. We're gonna quickly look at a couple of more earthquakes in more detail. 00:37:45.000 --> 00:37:55.000 This is the 2021 magnitudes, 5.7 and 6.2 doublet. 00:37:55.000 --> 00:38:04.000 So the magnitude 5.7 earthquake occurred just offshore of Petrolia followed 11 seconds later by a magnitude 6.2 event. 00:38:04.000 --> 00:38:08.000 ShakeAlert successfully detected this event. On the left you can see as a base map the ShakeMap, which shows the observed distribution of shaking. 00:38:08.000 --> 00:38:22.000 And then the octagon show the ShakeAlert alerting regions, which nicely cover the area affected. 00:38:22.000 --> 00:38:31.000 Additionally, in pink is shown the plum earthquake early warning, a ground motion based method that was also being tested. 00:38:31.000 --> 00:38:59.000 On the right is showing the observed ground motions in the triangles and the two curves, the lower curve the solid curve is based on the maximum alert magnitude from ShakeAlert which was M5.7, which closely matches the dash curve which is the ground motion expected based on the catalog magnitude. 00:38:59.000 --> 00:39:03.000 The second event that we're going to look at caused ShakeAlert a bit more trouble. 00:39:03.000 --> 00:39:07.000 So this was a magnitude 6 Antelope Valley earthquake in 2021. 00:39:07.000 --> 00:39:22.000 This earthquake occurred right near the California-Nevada border in an area of sparse station spacing. 00:39:22.000 --> 00:39:29.000 And was proceeded by a small foreshock because of these ShakeAlert estimated the M6 earthquake to instead be two magnitude 4.8 earthquakes, one centered near mammoth and one near Stockton. 00:39:29.000 --> 00:39:50.000 The estimated shaking polygons not really covering the affected area. On the right again, I'm showing a plot from Lux et al., which shows the predicted shaking estimated from the ShakeAlert magnitude 4.8 00:39:50.000 --> 00:40:06.000 in the solid curve and based on the catalog magnitude of 6 and the dashed curve you can see that with this underestimation at the magnitude ShakeAlert did not correctly estimate the ground motions for this event. 00:40:06.000 --> 00:40:14.000 Looking forward, there's a lot of ongoing research being conducted to improve this system. So this includes looking at different data sources 00:40:14.000 --> 00:40:22.000 such as DAS-based observations made on fiber optic cable, which could be useful for detecting offshore events. 00:40:22.000 --> 00:40:28.000 Additionally, there are other methods being looked at for improving our earthquake detection and characterization. 00:40:28.000 --> 00:40:48.000 So this includes GNSS based methods that can improve magnitude estimates for large events as well as ground motion based methods which don't estimate the source or location of an earthquake simply identifying where ground shaking might be strongest. 00:40:48.000 --> 00:40:57.000 And also methods to improve the ground motion predictions in areas where there might be significant site response. 00:40:57.000 --> 00:41:20.000 And there's also a really large amount of research going on related to social science, community engagement, and sort of ensuring that the earthquake early warning messages are used to their full extent. 00:41:20.000 --> 00:41:27.000 Okay, I think we reached the. End of the presentation. Thank you so much, Elizabeth. Our next speaker is Frannie Edwards from the San Jose State University, Manetta Transportation Institute 00:41:27.000 --> 00:41:42.000 and the talk is about earthquake early warning systems and emergency management. 00:41:42.000 --> 00:41:56.000 Good afternoon. Thank you for letting us talk with you this afternoon about some topics related to earthquake early warning system challenges for emergency managers. 00:41:56.000 --> 00:42:09.000 These challenges fall into several areas. The first and greatest concern is the issue of trust in government. Many times people do not believe messaging that comes from a government agency 00:42:09.000 --> 00:42:21.000 and they will spend time that they need for responding to the message confirming it with friends, relatives, radio, television, Dennis Mileti called this "milling" 00:42:21.000 --> 00:42:32.000 and it's important for us to establish our relationship with our communities, this allows them to respond quickly to emergency messaging such as an earthquake early warning system. 00:42:32.000 --> 00:42:36.000 We also have communities in which many of our members are either non-English speaking or have English as a second language 00:42:36.000 --> 00:42:49.000 and in either case receiving an emergency warning message may be confusing or difficult for them to understand. So we need to make a special effort to do outreach to those communities 00:42:49.000 --> 00:43:02.000 and help them understand what the message means and how they should respond. Another group of great concern is the homeless. 00:43:02.000 --> 00:43:15.000 Many of them live in creeks or river beds that are particularly vulnerable to natural hazards and yet they may not necessarily have a device on which they can receive an emergency warning message. 00:43:15.000 --> 00:43:21.000 We also need cooperation from the private sector. As we know, 85% of critical infrastructure is owned by the private sector. 00:43:21.000 --> 00:43:33.000 And we need their cooperation in using this earthquake early warning for the greatest benefit to save lives and protect the community. 00:43:33.000 --> 00:43:47.000 For example, we could encourage them to attach the earthquake early warning system to hazardous material control rooms so that the operators would quickly get a siren or a notice that they need to stop hazardous activities. 00:43:47.000 --> 00:43:55.000 Similarly in hospital surgical suites, an alarm or a light could warn the surgeon to take his hands out of the patient. 00:43:55.000 --> 00:43:57.000 Nursing homes have a very vulnerable population. Most of whom have limited mobility and they may be sitting in wheelchairs 00:43:57.000 --> 00:44:12.000 or at tables where things could fall on them. They're not capable of doing Drop and Cover, so we need to be concerned with preparing ahead of time 00:44:12.000 --> 00:44:33.000 how they can protect themselves in the event of an earthquake early morning. Ambulances similarly need to have a notice that goes off in the vehicle so that the driver can pull to the side and the paramedics can adjust the treatments that they're giving so that they wouldn't be life-threatening in the event of shaking. 00:44:33.000 --> 00:44:38.000 We also look for cooperation from our schools. For many years, schools have been doing Drop and Cover drills 00:44:38.000 --> 00:44:48.000 and so students that are sitting at desks or at tables have a ready method for protecting themselves when the earthquake early warning siren or light goes on, 00:44:48.000 --> 00:45:12.000 but they need to be taught that and practice it. It would be most helpful if the schools would also invest in retrofits so that tall thin furnishings and things like file cabinets can be secured ahead of time and the file cabinets can be kept in a locked position so that the drawers don't fly out during the shaking as happened throughout Los Angeles in the Northridge earthquake. 00:45:12.000 --> 00:45:20.000 These are a matter of budgeting. Where can they find the funding to do these retrofits to make the students safer? 00:45:20.000 --> 00:45:34.000 We also look for cooperation from public agencies. For example, fire departments can install an item that allows the fire doors to be automatically raised when the earthquake early warning system goes off. 00:45:34.000 --> 00:45:43.000 The picture on my slide shows Palm Springs Airport where they have installed such a device on this big door in their fire station. 00:45:43.000 --> 00:45:55.000 Sewer plants have controls that mix hazardous chemicals and they need the same kind of warning system so that the operator or an instant 00:45:55.000 --> 00:46:06.000 intercept can cause the sewer plant operations to pause while the shaking is going on. Similarly, public transit systems can benefit from an alarm. 00:46:06.000 --> 00:46:14.000 Sun transit in Palm Springs has installed alarms in their buses and their dispatch centers, 00:46:14.000 --> 00:46:26.000 so that the bus drivers get a dual notice to pull over to the side as soon as it's safe and stop the bus so that the passengers and the driver will be safe during the shaking. 00:46:26.000 --> 00:46:33.000 But again, the installation of these intercepts and interlocks are a matter of budget needs 00:46:33.000 --> 00:46:40.000 and public agencies in California have very limited budgets due to proposition 13. So working with public officials to create policy that makes such funding a budgetary priority 00:46:40.000 --> 00:46:59.000 is a really important part of our job. We also need funding for alarm installations at various locations throughout the community, including interfaces with these systems as we've described. 00:46:59.000 --> 00:47:12.000 We need better information on cell phone apps provided to the whole public and in addition to information we need training on how people who own Android phones can access an already installed program. 00:47:12.000 --> 00:47:26.000 For people with iPhones, they need instructions on how to go to the Apple Store and download the application that they would need to be able to get these notifications immediately. 00:47:26.000 --> 00:47:36.000 Emergency management's focus is on the community. So we start with individual preparedness when possible and when families can afford it. 00:47:36.000 --> 00:47:56.000 So on my slide at the top you see a go-kit put together by Lockheed Martin for their employees to have both at work, at their desks, in their cars on the way home and at home so that the employee, and if at home the family can have some basic supplies to begin to respond safely to an earthquake event 00:47:56.000 --> 00:48:06.000 when they hear the earthquake early warning. We also need planned post event assistance for those who cannot afford to put together their own kits 00:48:06.000 --> 00:48:13.000 and this includes planning with the American Red Cross for shelter opening in the community and where we would get the supplies. 00:48:13.000 --> 00:48:31.000 How we would find the supplies to make those shelters livable. The most important part of emergency management planning is to remind people that they need to have their prescription drugs and a supply of water to take them with available in their emergency kit, 00:48:31.000 --> 00:48:36.000 in their pocket or purse when they're at work and whenever they're out of their home. 00:48:36.000 --> 00:48:53.000 The Red Cross can replace prescription drugs that have been lost, but it won't happen immediately because the pharmacies in the community will be victims as well, experiencing shaking that may result in spillage or confusing items that are on the shelf. 00:48:53.000 --> 00:49:03.000 So the Red Cross will have to go to an unaffected community to get these prescription refills and that can take hours, even days. 00:49:03.000 --> 00:49:17.000 Alternate power for medical home devices is another concern. Providers can provide the equipment, but the user has to be sure to keep the batteries charged or to have replacement batteries available. 00:49:17.000 --> 00:49:23.000 How do we educate people on the importance of these actions? 00:49:23.000 --> 00:49:30.000 The earthquake early warning system brings many benefits to the community, if they're properly educated and prepared. 00:49:30.000 --> 00:49:43.000 We need as emergency managers to provide incentives for people to get their cell phone app installed. Training and information about downloading and accessing what's on the phone. 00:49:43.000 --> 00:49:54.000 We need an incentive for people to prepare their homes, to secure furniture, to have family Drop and Cover drills so that everyone knows where is safe. 00:49:54.000 --> 00:50:02.000 When you're at school and do a Drop and Cover drill you're usually sitting at a desk or at a table and by getting underneath that piece of furniture you can be safe. 00:50:02.000 --> 00:50:09.000 But what about at home? Are there things overhead that could fall down and hit you? Is there large furniture that could fall over? 00:50:09.000 --> 00:50:19.000 So these things need to be secured ahead of time. We need incentives for schools to prepare, securing furniture, and holding regular Drop and Cover drills. 00:50:19.000 --> 00:50:30.000 The children's safety should be an adequate incentive, but with so many programmatic requirements coming from state and federal government, they may see Drop and Cover drills for earthquakes 00:50:30.000 --> 00:50:47.000 as a less high priority because they see them as not so likely to happen, but the earthquake early warning system being present in the classroom might be that incentive, it might remind them that Drop and Cover drills are an important follow-on to hearing that alarm. 00:50:47.000 --> 00:51:00.000 We need incentives for public agencies in the private sector to install automatic interfaces that allow them to have the alarm system immediately open the fire department 00:51:00.000 --> 00:51:11.000 garage doors or to stop hazardous materials operations, or to alarm and warn the surgeon to quickly remove his hands from the patient. 00:51:11.000 --> 00:51:22.000 While these things seem obvious, it's a financial obligation that may be difficult to be met in a rural hospital or for a small manufacturing operation. 00:51:22.000 --> 00:51:37.000 Because as we always say in public administration, unfortunately, it all comes down to money, and this is where the earthquake early morning system can be an incentive for emergency managers to sell to their communities 00:51:37.000 --> 00:51:48.000 knowing that when an earthquake comes, they will be warned and they can take action. Thank you. 00:51:48.000 --> 00:52:05.000 Thank you very much. Our final speaker for this session before the discussion is Anne Bostrom from the University of Washington who will be talking about "The recent evolution of earthquake early warning expectations and needs on the U.S. West Coast." 00:52:05.000 --> 00:52:25.000 I'm Anne Bostron. It's my pleasure today to share preliminary findings from the second earthquake early warning survey for the U.S. West Coast developed in collaboration with the indomitable Sara McBride, Julie Becker, Jim Goltz, Bob de Groot, Lori Peek, and Brian Terbush, Maximilian Dixon, and the rest of the wonderful social science working group and I analyzed with John Downes, a colleague of mine 00:52:25.000 --> 00:52:38.000 at the University of Washington. In the first round of this survey, we fielded a compact seven-minute Survey of over a thousand adults in each state through North Samaras speak probability-based panel and loses non-probability panel calibrated using North-true-North to collect responses representing the adult populations of these states. 00:52:38.000 --> 00:52:51.000 In the second round in April, May of 2023, we did something very similar delayed a bit due to funding administration contracting. 00:52:51.000 --> 00:53:12.000 The survey took longer because we added a few questions and we increased the sample sizes. Today I will highlight just a few of our findings regarding pre-Pacific Northwest rollout experiences and perceptions of earthquake risks and earthquake early warning and how they have changed between 2021 and 2023. 00:53:12.000 --> 00:53:27.000 The Survey objectives including assessing earthquake experience, perceived risk from earthquakes perceived usefulness of ShakeAlert for a number of actions, emergency preparedness and earthquake preparedness perceptions, preferences for alerting thresholds, content, 00:53:27.000 --> 00:53:35.000 and alertly time and tolerance for false and missed alerts. We also asked about people's socio-economic characteristics. 00:53:35.000 --> 00:53:49.000 I will only speak about a few of these today. I will also skip the methods details here just to note that the total sample was over 3,000 across the three states and we weighed all of the analyses that we are reporting here today. 00:53:49.000 --> 00:54:07.000 The fast majority of respondents across both survey weighs reported having personally experienced an earthquake. As you may recall from the first earthquake survey, the earthquake early warning survey that we reported, these experiences are modally a mild earthquake shaking. 00:54:07.000 --> 00:54:19.000 However, the share of respondents would personally experience an earthquake in Oregon and Washington shifted to a slightly lower percentage in 2023 and we're not exactly sure why this could be due to shifts in public population 00:54:19.000 --> 00:54:30.000 demographics over that time period since it includes COVID, but it also could be a survey artifact since we did change the sampling slightly to achieve a larger sample. 00:54:30.000 --> 00:54:38.000 However, the waiting approaches should have accomplished a matching of population representative numbers. 00:54:38.000 --> 00:54:47.000 Since February, 2021 shown here in gold for old. Awareness has increased considerably. 00:54:47.000 --> 00:54:58.000 Particularly in Oregon and Washington where rollout happened after the baseline survey. So awareness in Oregon and Washington went from 11% to over 30% 00:54:58.000 --> 00:55:06.000 approaching awareness in California, which is at 39%. An estimated 39% in 2023. 00:55:06.000 --> 00:55:14.000 Washington and Oregon are very similar in many of their profiles in the responses. So here we show California compared to Washington. 00:55:14.000 --> 00:55:18.000 Here you can receive the responses again in gold from 2021 and in 2023 in blue for new. 00:55:18.000 --> 00:55:27.000 A majority in all three states understand that ShakeAlert sends a warning that an earthquake is approaching that ShakeAlert sends a warning that an earthquake is approaching, which is correct. 00:55:27.000 --> 00:55:30.000 However, there are slight declines for California across several of the questions we asked about ShakeAlert though they are small. 00:55:30.000 --> 00:55:38.000 For example, the percentage reported in California that ShakeAlert is operated by the U.S. Geological Survey decreased from an estimated 32% to an estimated 28%. 00:55:38.000 --> 00:55:54.000 This is not a large decrease and it rose in Washington and Oregon or remained constant between 26% and 33%. 00:55:54.000 --> 00:56:11.000 These are still relatively small proportions and we see that almost an equal proportion believes the misconception that ShakeAlert predicts earthquakes with this remaining constant California at 31% and increasing from 37% to 40% in Washington. 00:56:11.000 --> 00:56:18.000 When asked to what extent participants agree or disagree with the statements shown under their horizontal access here. 00:56:18.000 --> 00:56:32.000 In all states, participants agreed that ShakeAlert is helpful. These include that it would enable them to get some useful information about the earthquake even if they didn't feel the shaking, that it would enable them to help other people nearby. 00:56:32.000 --> 00:56:34.000 That it enable them to physically protect themselves from an earthquake and that it would enable them to mentally prepare themselves for shaking. 00:56:34.000 --> 00:56:50.000 Note that responses hover around 4, which is equal to agree. And are relatively constant. However, we see slight but significant drops on every statement 00:56:50.000 --> 00:56:58.000 in California from 3.9, 3.8 from 2.7 to 3.6 etc. On all four. 00:56:58.000 --> 00:57:07.000 Over 60% in all three states reported in 2023 that they would tolerate false alerts. 00:57:07.000 --> 00:57:18.000 This is similar to what we saw in 2021 as you can see but with an increase in Washington and Oregon and Oregon and the slight decline in California. 00:57:18.000 --> 00:57:27.000 This is largely due as you can see here to the associated declines and the percentages reporting don't know's in Oregon and Washington. 00:57:27.000 --> 00:57:37.000 There were also increases in all three states of the proportions reporting that they had seen or heard information about how to respond to an earthquake early warning. 00:57:37.000 --> 00:57:47.000 Going from 37 or 40% up to over 40%. So 45% California, 42% in Oregon and 45% in Washington. 00:57:47.000 --> 00:58:01.000 This is good news. On other items we see, again, drops in California. So from 70% to 63% reported that they had seen or heard information about how to respond to earthquake shaking in California compared to significant increases in Oregon and Washington. 00:58:01.000 --> 00:58:14.000 Reporting that they had seen or heard this type of information. We also saw a drop in California on preparedness 00:58:14.000 --> 00:58:23.000 such as practicing, responding to an earthquake drill or participating in training or exercises to better respond and increases 00:58:23.000 --> 00:58:29.000 though insignificant increases in Oregon and Washington on these items. 00:58:29.000 --> 00:58:36.000 Most respondents when asked about their preferences for alerting channels prefer cell phone or smartphone alerts. 00:58:36.000 --> 00:58:42.000 We saw this in 2021 and it remains high in 2023 by state. 00:58:42.000 --> 00:58:50.000 However, there are also again, oops, sorry, declines in California. Across all of these channels 00:58:50.000 --> 00:58:55.000 from 92 to 87% for cell phone or smart alerts in California and similar drops in every single channel. 00:58:55.000 --> 00:59:07.000 The two other channels not reported here including home, telephone, and family of friends and they were preferred by fewer than 20%. 00:59:07.000 --> 00:59:17.000 Majorities were interested in all of these types of alert information when asked if you received an earthquake alert through your phone, what information would you want in that alert 00:59:17.000 --> 00:59:25.000 and that included shaking intensity. Guidance on what action to take, information on additional hazards, a countdown timer and potential impacts from the shaking. 00:59:25.000 --> 00:59:48.000 I'm showing here the percentages for 2023. And you can see that there are significant increases for Oregon and Washington in some places and significant decreases again for California on shaking intensity on information on guidance on what action to take, and on information in additional hazards and even account down timer. 00:59:48.000 --> 01:00:14.000 We also see that interest in post-alert messaging remains fairly high. When asked if you received an earthquake alert through your phone, what information would you want to receive later immediately after the earthquake in 2023 we see similar profile of responses to what we saw in 2021 across all three state over 70% would like further information about the earthquake that has occurred, so it's location and magnitude and information on 01:00:14.000 --> 01:00:22.000 additional hazards such as there could be a tsunami in coastal areas. Interest in these types of information is higher in Oregon and Washington than is in California and there seems to be lower interests in all types of information 01:00:22.000 --> 01:00:35.000 in California than in the other two states, similar to what we saw on the other questions. 01:00:35.000 --> 01:00:55.000 We again asked which level is the lowest level of shaking you think would be useful to be warned for, and in 2023 we say again something very similar to what we saw in 2021 that the normal response is that people would like to be alerted for mild shaking with over 40% preferring that 40% or higher in California, Oregon and Washington. 01:00:55.000 --> 01:01:04.000 But we again see that about a quarter of respondents or higher would like lower levels of alerting and are interested in being alerted for weak shaking even, for example, 01:01:04.000 --> 01:01:08.000 and there are a few respondents who would prefer higher levels thresholds for alert including MMI 5 and 6 and above. 01:01:08.000 --> 01:01:25.00 So there are a strong interest in low alerting thresholds. I know the slight difference in California profile compared to the other two states. 01:01:25.000 --> 01:01:32.000 So to summarize the takeaways. Awareness of ShakeAlert has risen from 2020 to one to 2023 ib all states. 01:01:32.000 --> 01:01:41.000 Awareness of earthquake early warning has increased in Oregon and Washington dramatically. False alert tolerance remains high and has increased in Oregon in Washington. 01:01:41.000 --> 01:01:45.000 Trends and alerting preferences remain relatively stable. An overwhelming majority of response prefer cell phone alerts compared to other channels. 01:01:45.000 --> 01:01:55.000 There's interest in post-alert messaging and in lower alerting thresholds as there was in 2021. 01:01:55.000 --> 01:02:10.000 However, there are in California slight drops in the perceived helpfulness of ShakeAlert and in desired alerting information that we wonder whether these might be signs of early alerting fatigue so to speak. 01:02:10.000 --> 01:02:15.000 This remains to be investigated, thoroughly and further. Thank you for your attention. These are preliminary results and we gratefully acknowledge support for this. 01:02:15.000 --> 01:02:28.000 Research to the National Science Foundation supported through USGS. Thank you. 01:02:28.000 --> 01:02:35.000 Thank you very much to Anne and all of our speakers. So round of applause to everybody. So now we will be moving on to our discussion section. 01:02:35.000 --> 01:02:53.000 We do have several questions that we have prepared ahead of time, but if you have your own questions, please feel free to put it in the chat or even better to raise your hand. 01:02:53.000 --> 01:02:58.000 So Sara McBride will be leading our discussion. Go ahead, Sara. 01:02:58.000 --> 01:03:07.000 Thank you, Jessie. I just want to thank all of our wonderful speakers today for doing such fantastic presentations around earthquake early warning systems. 01:03:07.000 --> 01:03:16.000 And this one goes to really looking at what the goals are of each individual systems that were presented. 01:03:16.000 --> 01:03:31.000 So if you're representing a system like Elizabeth or Mitsuyuki or Allen. Could you talk about what the goals are of your system? 01:03:31.000 --> 01:03:49.000 Sure, I can start. So thanks for that question, Sara. In general, when I think about earthquake early warning, I think that the goals for an early warning system are to provide advanced warning for moderate to strong ground shaking. 01:03:49.000 --> 01:04:10.000 So how that's actually spelled out for ShakeAlert is a little different so the priorities that were recently defined for ShakeAlert are to provide accurate ground motion estimates within plus or minus one MMI units for 01:04:10.000 --> 01:04:22.000 moderate to large earthquakes with some emphasis on the larger earthquakes. And then additionally, kind of separate goal is to provide at least 10 seconds of warning 01:04:22.000 --> 01:04:27.000 for areas of stronger shaking. 01:04:27.000 --> 01:04:30.000 Thanks, Elizabeth. Allen? 01:04:30.000 --> 01:04:38.000 Sure, so for Mexico the goal, the original goal is specifically they thought there was going to be an earthquake in the Guerrero Gap and they really wanted to alert for that specific earthquake. 01:04:38.000 --> 01:04:49.000 It was the original goal and because it's on the coast and Mexico City is inland, it gives about a minute of warning time. 01:04:49.000 --> 01:04:58.000 Since then there hasn't been any earthquake, there's been plenty of earthquakes other places and they've expanded the system to other areas as well with kind of the overall goal being the same. 01:04:58.000 --> 01:05:15.000 They only kind of care about the big ones, however they understand that in order to get the big one you have to alert before it's grown to be fully like a magnitude 8 and the wave is already propagating towards the city so it needs to be alerted before then. 01:05:15.000 --> 01:05:29.000 Before you know the rupture time gets to be close to a minute at that point anyways and so yes, they alert on a lower magnitude threshold, which is magnitude 5 01:05:29.000 --> 01:05:36.000 with that in mind. They're also starting to alert now for earthquakes that are closer. 01:05:36.000 --> 01:05:41.000 This is largely because of the earthquakes that are closer. This is largely because of the 2017 earthquake that was closer, 01:05:41.000 --> 01:05:46.000 that was a deeper earthquake within the subduction zone so now they have faster alerts for those. 01:05:46.000 --> 01:05:53.000 One of the things I didn't mention, which is kind of a side note, that the population is well aware that they have a minute. 01:05:53.000 --> 01:06:14.000 And so there's a study done, recently; well, actually it's been a couple years now, but where they found that people were actually taking that into account because they knew that oh have a minute and so they took time to put their shoes on or grab their pets or things like this before taking actions. 01:06:14.000 --> 01:06:17.000 Thanks, Allen. Mitsuyuki would you like to talk to us about the goals of your system? 01:06:17.000 --> 01:06:27.000 Yeah, of course our final goal is giving people the time to secure their safety. 01:06:27.000 --> 01:06:33.000 So from our ground motion view point to provide the people the warning as to strong ground shaking. 01:06:33.000 --> 01:07:01.000 And, it's [indiscernible] plus minus one degree on JMA scale. So of course it's depending on the distance of the epicenter, but our final goal is to provide people before strong shaking with accuracy less than one plus minus one on JMS scale. 01:07:01.000 --> 01:07:05.000 Thank you. 01:07:05.000 --> 01:07:17.000 That's really clear and really helpful. Frannie, I know that you're part of the ShakeAlert System, but from a technical user perspective, what would you say is the goal for your system? 01:07:17.000 --> 01:07:31.000 Well, as an emergency manager, of course, my goal is pretty much as our friend from Japan said that we want people to be notified in a timely manner so they can prevent injury to themselves and the members of their household. 01:07:31.000 --> 01:07:42.000 But that's only going to happen when we have adequate training and I thought it was very interesting that the representative from Italy specifically said he got the warning on his phone and he didn't know what to do. 01:07:42.000 --> 01:07:56.000 So I think it's an important nexus that we have to create between the very efficient scientific capability of notifying and people's response to the buzzing when they hear it. 01:07:56.000 --> 01:08:02.000 Yes, absolutely. And Anna, you know, I know you're working on the social science, with me. 01:08:02.000 --> 01:08:13.000 But do you think the goals of ShakeAlert are well reflected in the understanding of people in the surveys, the two surveys that you've now led. 01:08:13.000 --> 01:08:17.000 Well, that's a good question. I would ask you the same thing. I think so. 01:08:17.000 --> 01:08:18.000 People understand 01:08:18.000 --> 01:08:22.000 we are there to help them? However, as in some data that I didn't report, they want longer warning times, 01:08:22.000 --> 01:08:36.000 they think they need longer warning times that are available to any of them, and they think that they need those longer warning times in order to be able to do anything. 01:08:36.000 --> 01:08:39.000 So there's some some misunderstanding of that and we also see that there's quite a large percentage who think that the system predicts earthquakes. 01:08:39.000 --> 01:08:53.000 So there are some mismatches between what the system does and what they think it does, and that could potentially lead to some problems as we've seen in the past. 01:08:53.000 --> 01:09:15.000 So, yeah, so on the other hand, there is this tension as Frannie just highlighted and also Allen did that people also have a desire for a lot of information and they may not be able to respond to earthquakes if they aren't practiced at it, if they don't have drilling. 01:09:15.000 --> 01:09:27.000 So there's a tension between this, the need to understand and have a regular reminder of what an earthquake is and only be alerted for something that's really so large that you're large likely to get harmed. 01:09:27.000 --> 01:09:36.000 So I don't know how to resolve that, but one potential path towards this is to do what they've done in Japan and have two different levels of alerting 01:09:36.000 --> 01:09:46.000 for example. Another might be to make an educational app available that alerted it lower thresholds, for example, that would have a slightly different function. 01:09:46.000 --> 01:09:55.000 Yes, and it speaks to some of our research around how important, you know, ShakeOut and Drop, Cover, and Hold On drills are so critical and important. 01:09:55.000 --> 01:10:05.000 So the next question I have is a little bit of attention, right. You have mentioned, so thank you for setting up this question about which is more important in terms of meeting our goals 01:10:05.000 --> 01:10:09.000 the accuracy or timeliness? 01:10:09.000 --> 01:10:16.000 And I'll throw that to Elizabeth. You're welcome. 01:10:16.000 --> 01:10:17.000 I am. 01:10:17.000 --> 01:10:25.000 You throw me the hard questions. So, I would say you have to consider both, right? 01:10:25.000 --> 01:10:30.000 So what we want to do is make sure that we're providing people with warning when they need it, right? 01:10:30.000 --> 01:10:47.000 So our focus needs to be on considering, you know, the stronger ground shaking and then evaluating when we need to provide people with alerts to make sure we don't miss any alerts for those times of strong ground shaking. 01:10:47.000 --> 01:10:59.000 And then also considering as part of that, you know, say you want to be warned or you need to be warned for MMI 6, 01:10:59.000 --> 01:11:06.000 you're gonna get a longer warning if you decide to take action when we think the ground motion is lower than that. 01:11:06.000 --> 01:11:16.000 So there's this, as you said, tension between kind of the accuracy, which times generally the accuracy of our system 01:11:16.000 --> 01:11:23.000 is higher during an earthquake, right? We have more information coming in. We can better characterize that event 01:11:23.000 --> 01:11:38.000 with this need to give people advanced warning. So it is that tension and it's not an easy question or idea to explain to the general public. 01:11:38.000 --> 01:11:42.000 Thanks, Elizabeth. 01:11:42.000 --> 01:11:43.000 Yeah, go ahead, Allen. 01:11:43.000 --> 01:11:56.000 Can I add something from the Mexican perspective? The timeliness is definitely more important, but also the question of what is accuracy is different for a seismologist than it is for the public. 01:11:56.000 --> 01:12:05.000 In general, the public there seems to think it's accurate as long as there was an earthquake or an alert that was issued and an earthquake occurred. 01:12:05.000 --> 01:12:14.000 If an earthquake occurs but there is no alert they consider that inaccurate. That was kind of like the general public's perception of it. 01:12:14.000 --> 01:12:22.000 And so I think the Mexican system tried to follow that as much as possible. I don't know that's the same as what ShakeAlert is doing, but I'm representing Mexico today. 01:12:22.000 --> 01:12:35.000 I think that's really, really useful. I want to get Frannie's take on this in terms of from an emergency management perspective, what your preferences might be. 01:12:35.000 --> 01:12:43.000 Well, I think we're much more concerned with timeliness. When ShakeAlert was being designed, this was a constant 01:12:43.000 --> 01:12:48.000 debate between the scientists and the social scientists. It doesn't really matter to us exactly where the rupture occurred or exactly what the moment 01:12:48.000 --> 01:13:03.000 magnitude is. What matters to us is that an earthquake happened. In a populated area and the result is that people might have been injured or that they could have taken protective measures. 01:13:03.000 --> 01:13:11.000 So timeliness is really critical and I understand the scientific side of needing to know what really happened, but 01:13:11.000 --> 01:13:33.000 for me as an emergency manager, that's something you guys can deal with later on. What I need to know at the moment of the shaking is that enough people were notified early enough in that they were able to get away from truly hazardous situations and at least Drop and Cover and underneath something protective. 01:13:33.000 --> 01:13:41.000 Yes. Mitsuyuki can you give us some of your thoughts? 01:13:41.000 --> 01:13:50.000 Okay, so denoting time, it's a very good point to bring up, in case of a strong shaking and moderate shaking, or while shaking. 01:13:50.000 --> 01:13:52.000 So, for example, weak shaking. So, deducing a time from 20 seconds to 50 seconds is not so important. 01:13:52.000 --> 01:14:09.000 But regarding strong shaking, for example, reducing Hmm. Yeah. In case of even 5 seconds, 01:14:09.000 --> 01:14:24.000 even 3 seconds before strong shaking that's very important, so I think it depends on people in case they are in strong shaking area or weak shaking area. 01:14:24.000 --> 01:14:34.000 So, anyway the purpose of earthquake early warning a lot of warning should be before strong shaking, 01:14:34.000 --> 01:14:58.000 so that is one of the important point. So in case of a very distant earthquake, so yeah, 5 seconds area it's not so important but in the case in a strong shaking very near high central region that is an important thing and so in case of very precise prediction so seismic interesting, plus minus one it's probably, I think 01:14:58.000 --> 01:15:02.000 that total of many people. 01:15:02.000 --> 01:15:12.000 Thank you. Anne, I wanted to ask you, and this was sort of going through my mind when you were presenting earlier around timeliness and how much time people thought that they needed. 01:15:12.000 --> 01:15:13.000 And do you think that's maybe an indicator of preferences within various public's that we're trying to serve with ShakeAlert. 01:15:13.000 --> 01:15:25.000 People obviously 01:15:25.000 --> 01:15:34.000 want more warning than they have. People do have a sense that they can't do much 01:15:34.000 --> 01:15:46.000 with 10 seconds, though those people will experience them could probably do quite a bit with 10 seconds, and people also overestimate how much they can do with each of the time intervals that we offered, as you may recall. 01:15:46.000 --> 01:15:55.000 And so, yeah, it probably does reflect preferences, but it also reflects a lack of experience with earthquakes where they really need to do something. 01:15:55.000 --> 01:16:10.000 As we saw the same thing that has been seen in other surveys that people tend to freeze rather than Drop, Cover, and Hold On and they're doing this sort of sense making thing that Frannie referred to and that is not necessarily a protective action depending on where they are at the time of shaking. 01:16:10.000 --> 01:16:22.000 So, it's a tricky notion and I think the important thing here is what everybody's been repeating that they have some warning before the strong shaking. 01:16:22.000 --> 01:16:32.000 And ideally they would know what to do with that warning. The categorization of what happened, which several of you referred to and you asked questions about, 01:16:32.000 --> 01:16:43.000 I would want to refer people to the paper that you are a lead author on post alert, messaging where we talked about the different ways you can think about what's happened with an alert and I think 01:16:43.000 --> 01:16:51.000 reflect on the differentials between how seismologists for example think about those categories and how people in the public who are trying to use that information think about those categories. 01:16:51.000 --> 01:17:07.000 There is quite a big discrepancy between them and thinking about that in terms of how we give people information after an alert to help to update their understanding of the system is really important. 01:17:07.000 --> 01:17:18.000 Thanks, Anne. That actually leads really perfectly to my next question, Jesse is laughing because she knows what I'm gonna ask next, which is thinking about the development of your system, 01:17:18.000 --> 01:17:42.000 so ShakeAlert or the Japanese system or the Mexico system. Thinking about these different systems, who was involved in developing those systems and was it, do you think looking back were the right people involved in developing it, and was anyone missing in the development of these systems and then thinking over time 01:17:42.000 --> 01:17:51.000 you know, have the people changed who started developing the system to know who are actively running the system. So all of those questions. 01:17:51.000 --> 01:18:01.000 I'm gonna head over to Allen first, you know, starting with around the early development and reflecting on that. 01:18:01.000 --> 01:18:05.000 The early development of the Mexican system. 01:18:05.000 --> 01:18:06.000 Yes. 01:18:06.000 --> 01:18:14.000 So I wasn't really around at the time. It started in 1991 or 1992. 01:18:14.000 --> 01:18:27.000 So it's a really old system. So it was all. Very much based on the technology in that time and it's stayed with that technology which is something to think about maybe that's going to happen with the ShakeAlert system as well. 01:18:27.000 --> 01:18:35.000 And you get locked in perhaps to the time when your technology came on board. Oh, it might just be a reflection of the economy. 01:18:35.000 --> 01:18:44.000 But yeah, they do point to radios and it's all, you know, alert systems, no cell phones because it's kind of baked into the system. 01:18:44.000 --> 01:18:47.000 Does that answer your question. 01:18:47.000 --> 01:18:54.000 Yeah, you know, kind of, I mean, it is hard because yeah, you weren't around when it first got developed and you know 01:18:54.000 --> 01:19:00.000 little bit complicated I think for you to answer. Elizabeth you've been in the ShakeAlert realm for a really long time now 01:19:00.000 --> 01:19:15.000 and what do you think about who was involved in the development of ShakeAlert and was the right people and, how is it going now? 01:19:15.000 --> 01:19:22.000 Yeah, so I think, so initially when the system, when I was around, when the system was started in 2006. 01:19:22.000 --> 01:19:34.000 It was really an effort led by universities to develop algorithms for looking at kind of the technology of how we can quickly detect and characterize earthquakes. 01:19:34.000 --> 01:19:46.000 And then over time that sort of built into a pilot system. I would say early in the development probably what was missing was, the social science. 01:19:46.000 --> 01:19:58.000 So Sara is aware that, the social science research really didn't get started until probably within the last 5 years. 01:19:58.000 --> 01:20:12.000 And so I think that was a big piece is really understanding what users wanted on the system as well as matching what the technological capabilities of the system were. 01:20:12.000 --> 01:20:29.000 So I think that was probably the biggest, yeah, early on and then generally, you know, I think from sort of the developer perspective, the system really started out with this idea that we were gonna provide a message 01:20:29.000 --> 01:20:38.000 and then other people were gonna take that message and use it, and so we weren't necessarily the people issuing alerts 01:20:38.000 --> 01:20:55.000 we were leaving that decision to others and I think there's enough complexity with sort of a well weaken, detecting and rapidly characterize earthquakes to provide these alerts that we maybe 01:20:55.000 --> 01:21:09.000 needed to have more thought early on the kind of what our abilities to provide rapid alerts and what the uncertainties were. 01:21:09.000 --> 01:21:24.000 Thanks, Elizabeth. Mitsuyuki, can you talk to us a little bit about the development of your system and whether you think the right group of people were involved in that initial development. 01:21:24.000 --> 01:21:32.000 JMA started in 2007, so at the time, yeah, they wanted a group, so group didn't want to miss it. 01:21:32.000 --> 01:21:44.000 So at too fast to 2007 so maybe [indiscernible] but it's, 15 years, JMA, group tried to integrate the message. 01:21:44.000 --> 01:21:58.000 So, last year, 2023, yeah, the source hypocenter manage determination integrated into extended IP message. 01:21:58.000 --> 01:22:07.000 And also, probably after the lesson from the Tohoku earthquake we developed strong ground motion message. 01:22:07.000 --> 01:22:19.000 So at the present is two algorithms. One is source-based algorithms and ground motion based message working priority. 01:22:19.000 --> 01:22:27.000 So of course, I'm a person of the JMA and also cooperation with the university group. 01:22:27.000 --> 01:22:36.000 Yes some effort to develop the method is continuing 01:22:36.000 --> 01:22:51.000 Right. Thanks. Frannie, I wanna throw this over to you as well. What you think about the development of your side of the system and how it's gone for you. 01:22:51.000 --> 01:23:01.000 That's a good question. I think emergency managers in most communities have so many jobs that they have to do, 01:23:01.000 --> 01:23:06.000 but I'm not sure they've kept up very well with the earthquake early warning system 01:23:06.000 --> 01:23:26.000 or how it's developed. In 2015, my colleague Dan Goodrich and I did some research for the Mineta Transportation Institute on how the Japanese rail lines were using earthquake early warning on high speed rail because we were doing research on behalf of California high speed rail 01:23:26.000 --> 01:23:29.000 to look at how they could integrate their system to make the trains operate safely and so that was the human group that we were looking at, 01:23:29.000 --> 01:23:45.000 Were the train riders, and in doing that research, I just covered that we're not very many emergency managers that were keeping up 01:23:45.000 --> 01:23:56.000 with the development of California's system. They knew Berkeley was doing something. That was pretty much the response I got when I asked my colleagues or USGS is doing something. 01:23:56.000 --> 01:24:02.000 So I think it just speaks to the fact that we come at this from two very different perspectives. 01:24:02.000 --> 01:24:11.000 All the emergency managers want to do is protect people and property. Scientists are looking at better understanding the phenomena. 01:24:11.000 --> 01:24:23.000 And it's not that the two things are in conflict, but the emergency managers who are the ultimate consumers of this information to present to their communities 01:24:23.000 --> 01:24:31.000 are just not keeping up to be honest, they're not as involved as they might be. 01:24:31.000 --> 01:24:47.000 And I think the people in the Bay Area, because they're close to USGS and because of Anne Wein's tremendous work with the Hayward fault research and the amount of outreach that she and her team have done has made a big difference there. 01:24:47.000 --> 01:24:57.000 But if you go statewide and you talk to people that are in Santa Barbara or San Diego and you talk about earthquake early warning they kind of roll their eyes and talk about something else. 01:24:57.000 --> 01:25:07.000 Yeah, that's a really valuable insight, Frannie. You know, and in terms of understanding how emergency managers are talking about this. 01:25:07.000 --> 01:25:15.000 Within their communities and how they're making use of it. So really appreciate your perspective. And is there anything? 01:25:15.000 --> 01:25:16.000 What? 01:25:16.000 --> 01:25:35.000 You need to go to other parts of the state. 01:25:35.000 --> 01:25:36.000 Yeah. 01:25:36.000 --> 01:25:39.000 Yeah, I mean, I wish we had more Anne's period that would make me happy. Speaking of my favorite Anne Bostrom, just quickly curious if there's anything like from your perspective as having been a researcher because you did some really great earthquake early warning research in Washington state, 01:25:39.000 --> 01:25:50.000 what in 2014 that paper came out, you know, like you've been doing this for a while before ShakeAlert was publicly available warning system. 01:25:50.000 --> 01:25:52.000 What are your thoughts on this? 01:25:52.000 --> 01:26:02.000 Yeah, that was a result of being involved in a BM9 project. Of course, that was, and John Vidali was a force 01:26:02.000 --> 01:26:11.000 for that project and also in public outreach and education and was very interested in the whole warning aspect of it 01:26:11.000 --> 01:26:26.000 so that promoted that research. In that process, as you know, state emergency managers in Washington state, but I think also in Oregon to some extent, but especially in Washington have been very good partners on this work and have also promoted it. 01:26:26.000 --> 01:26:33.000 I don't know the extent to which they have communicated with local emergency managers about earthquake early warning, it'd be interesting to 01:26:33.000 --> 01:26:38.000 follow up and talk with them about it. 01:26:38.000 --> 01:26:47.000 The same kind of experience that Fran is talking about that we have tsunami with earthquake early warning. 01:26:47.000 --> 01:26:48.000 No, sorry. 01:26:48.000 --> 01:26:54.000 Oh, my Internet sounds terrible. Yeah, I just wanna say one thing. 01:26:54.000 --> 01:26:56.000 Go ahead and. 01:26:56.000 --> 01:27:04.000 Oh, I was just going to say there were the TRInet studies in the 1990 and then there was a kind of a gap. 01:27:04.000 --> 01:27:15.000 And so it's interesting to me that the TRInet studies weren't sort of taken up as the technology was pushed forward, but kind of got dropped 01:27:15.000 --> 01:27:19.000 and it would be interesting to hear Jim Goltz take on that and if anybody else was involved in that could comment on it. 01:27:19.000 --> 01:27:31.000 Yeah, it is really interesting that we have had earthquake early warning social science research since 1999 that was published i think in 1999 and then 2003 was another paper that Jim published. 01:27:31.000 --> 01:27:43.000 And so it's gone and sort of fits and starts in terms of the development of social science around earthquake early warning in the United States, but I don't want to distract us too much around that. 01:27:43.000 --> 01:27:49.000 I want to talk about some external, and that we've done some internal reflecting on our system and the goals. 01:27:49.000 --> 01:28:01.000 And you know how we perceive success amongst ourselves but what do you think are the key challenges with communicating our alert strategies. 01:28:01.000 --> 01:28:07.000 And goals like either and we've kind of touched on this a little bit but either in general or more specific to your system. 01:28:07.000 --> 01:28:15.000 And particularly around communicating alert performance. I'm gonna hand this to, let's see here 01:28:15.000 --> 01:28:22.000 Ann do you wanna start out with this? I always keep leaving you to the end. So. 01:28:22.000 --> 01:28:26.000 I got distracted I was answering something in the chat. Sorry. Overall performance. 01:28:26.000 --> 01:28:29.000 That's alright. What do you think are the key challenges of communicating 01:28:29.000 --> 01:28:34.000 alert performance and strategies? 01:28:34.000 --> 01:28:40.000 I think the key challenges are actually sort of terminology and that categorization that I talked about. So how do you think about what's good and what's not so good? 01:28:40.000 --> 01:28:59.000 And that goes back to that discrepancy in terms of what people see as useful or important between scientists and people in different positions like emergency managers or the people who are trying to figure out they should do something. 01:28:59.000 --> 01:29:06.000 Thanks, Anne. Elizabeth, what do you think? What do you think are some of the key communication challenges? 01:29:06.000 --> 01:29:30.000 I think we touch on a few of them already, but it's this idea what are people's expectations, how do we accurately or appropriately set those expectations for when and how people should receive alerts and then of course a big I know you're also a big proponent of this, but it's post-alert messaging I think goes a long way in terms 01:29:30.000 --> 01:29:43.000 of making sure people understand perhaps why they did or didn't receive an alert. And use some of the language and, for example, your post alert paper 01:29:43.000 --> 01:29:58.000 To describe how the system performed. You know, there's obviously a different set of, analyses that we do as scientists but I think that's key and it's really key 01:29:58.000 --> 01:30:05.000 to have people understand when they might or might not receive an alert. So I think we haven't quite 01:30:05.000 --> 01:30:15.000 managed that communication fully with the ShakeAlert system yet. I think there's still a lot of confusion. 01:30:15.000 --> 01:30:23.000 Thanks, Elizabeth. Mitsuyuki, what do you think are some of the key challenges in Japan for communicating 01:30:23.000 --> 01:30:27.000 around earthquake early warning and the goals? 01:30:27.000 --> 01:30:28.000 So, fortunately we have an 01:30:28.000 --> 01:30:40.000 we have a lot of earthquakes. So, we started 2007 people didn't know the meaning of earthquake early warning. 01:30:40.000 --> 01:30:45.000 So, JMA conducted a big campaign, but even so, people didn't know the meaning. 01:30:45.000 --> 01:30:54.000 But it's a frustration, they're automatically changed after Tohoku earthquake in eastern Japan and the Kumamoto earthquake in 2016. 01:30:54.000 --> 01:31:02.000 So at present we don't have to explain what earthquake early warning, people already know the meaning of earthquake early warning. 01:31:02.000 --> 01:31:09.000 So actually experience is a best teacher and the viewpoint of the technology. 01:31:09.000 --> 01:31:16.000 So to send a message is the same as 2007 01:31:16.000 --> 01:31:26.000 Needs a sound and text message. That's all. But to current, we have more, we are now investing it a more, direct. 01:31:26.000 --> 01:31:37.000 message, a more informative message using the current technology. So now JMA is investigating now how to communicate to people more different ways using modern technology. 01:31:37.000 --> 01:31:46.000 In addition to sound, that takes the message. 01:31:46.000 --> 01:31:53.000 Thank you. Alan, what do, what do you think about this question around some of the challenges around communicating around about earthquake early warning. 01:31:53.000 --> 01:32:05.000 So I agree with Mitsuyki that, it's all about experience and so if you feel that experience they're going to know what to do. 01:32:05.000 --> 01:32:08.000 And so in Mexico, this is not a problem just like in Japan, there's plenty of earthquakes. 01:32:08.000 --> 01:32:14.000 This can become more of a problem on the West Coast of the U.S. where there's just not as many earthquakes. 01:32:14.000 --> 01:32:20.000 I will say the other thing though is the way the errors are reported, I kind of mentioned this in my talk 01:32:20.000 --> 01:32:41.000 by Mexico it's viewed very differently than it is from ShakeAlert. I feel sometimes the public responds to things we say and so there's like an earthquake that say might give you 5.1 but we reported a ShakeAlert comes out of the 6 and I go oh sure you said it was a 6 because that's what we're saying that there's some problem with that because 01:32:41.000 --> 01:32:50.000 it was reported too high or whatever. Whereas, you know, there was alert that went out for an earthquake and so that's a good thing and we should just be happy about that in some level. 01:32:50.000 --> 01:33:08.000 And that's what the public really cares about. As opposed to you know some of these errors that we kind of focus on which is important I think for the scientists but just when that gets reported into the public, I think it twists a little bit, makes it sound like a system is not probably working as as it really is. 01:33:08.000 --> 01:33:13.000 Yeah, thanks, Allen. Frannie, what do you think some of the challenges are you've already mentioned the numbers and it's been really, really useful getting your perspectives on this. 01:33:13.000 --> 01:33:22.000 What do you think are some of the challenges around communicating earthquake early warning? 01:33:22.000 --> 01:33:25.000 What do you think are some of the challenges. 01:33:25.000 --> 01:33:32.000 I think the two biggest problems are first of all the public doesn't understand much about earthquakes. 01:33:32.000 --> 01:33:37.000 So when you try to explain it to them from a technical perspective about the fact that you have a fault rupture that occurs one place and you call it a magnitude 6. 01:33:37.000 --> 01:33:51.000 In San Jose during Loma Prieta we felt nothing. In Oakland, their infrastructure fell down. 01:33:51.000 --> 01:33:58.000 Well, San Jose is a lot closer to the fault. Why didn't San Jose feel anything and everything fell down in Oakland. 01:33:58.000 --> 01:34:15.000 And those conversations that I end up having with my public members are really, really difficult because people don't pay attention in school when they get their science lesson in seventh grade by the time they're adults they've forgotten most of what they were taught. 01:34:15.000 --> 01:34:22.000 So I think one of the greatest losses we've had in the state of California was the loss of earthquake preparedness month. 01:34:22.000 --> 01:34:33.000 April used to be earthquake preparedness month and all month long TV stations and radio stations would carry little vignettes, maybe a 2 minute 01:34:33.000 --> 01:34:43.000 item on what you should do, and radio stations would come and talk to us, emergency managers in different communities in the Bay Area, 01:34:43.000 --> 01:34:50.000 And we'd get a chance to give people some little reminders, but none of that's going on anymore. 01:34:50.000 --> 01:35:02.000 And however the messaging is being, put out there it isn't reaching me or the people I know I'm not saying it's not happening, but it's not happening in a way that's 01:35:02.000 --> 01:35:29.000 robust that's making members of the public pay attention and so now you introduce the idea of an earthquake early warning system and people think you're going to give them two minute notice so they can pick up their dog and you know close the TV off and all these things and when you tell them no, you're gonna get some 5, 10, 15 seconds warning, which is going to let you stop what you're doing 01:35:29.000 --> 01:35:37.000 and get someplace safer. To Drop and Cover, protect your eyes, something simple they just shake their heads and walk away. 01:35:37.000 --> 01:35:47.000 So I really think we need to go back to a regular earthquake warning program for the whole public. 01:35:47.000 --> 01:35:55.000 To tell people what to do, how to prepare. As you probably know, the cities have something called CERT, Community Emergency Response Team. 01:35:55.000 --> 01:36:02.000 That's FEMA sponsored. But getting money to put that program into place is now very difficult 01:36:02.000 --> 01:36:09.000 because the EMPG money has all been mixed in with the UASI money and it all goes to police and fire. 01:36:09.000 --> 01:36:16.000 So if we really want to be serious about making earthquake early warning as valuable as it can be 01:36:16.000 --> 01:36:22.000 we need to think about how we're going to fund the public education that allows the listener 01:36:22.000 --> 01:36:27.000 to respond properly to the message. 01:36:27.000 --> 01:36:36.000 Thanks, Frannie. I really like what you're talking about. I do want to give a plug for ShakeAlert.org which has a number of toolkits and some for emergency managers as well. 01:36:36.000 --> 01:36:45.000 So we do have resources out there for emergency managers. However, you know, we can always learn how to disseminate those a bit more effectively as you're suggesting. 01:36:45.000 --> 01:36:55.000 So thanks so much for those insights. You mentioned something about, helping with your pet, your dog. 01:36:55.000 --> 01:37:00.000 And I know I saw Ann laughing because that was actually within, I think, some of the comments that people thought they could do with the amount of time they had, right Ann. 01:37:00.000 --> 01:37:08.000 And from recollection from the survey. 01:37:08.000 --> 01:37:09.000 Yeah. 01:37:09.000 --> 01:37:16.000 Yes, indeed. And there is, I mean, there are, there is a proportion of people who really do focus on protecting others. 01:37:16.000 --> 01:37:26.000 Sorry. high portion, but there's a staple, people are interested in that there's a really interesting conversation going on in the chat about people turning off the alerts for phones. 01:37:26.000 --> 01:37:40.000 Yes, I'm really enjoying this and we only have 5 min left unfortunately. So I can't really go into in depth in the chat, but maybe the last question, which is a question of optimism looking towards the future. 01:37:40.000 --> 01:37:48.000 Maybe this can help help us a little bit with the conversations in the chat and it's really like what types of innovations would you like to see in earthquake early warning systems, either yours or globally within the future. 01:37:48.000 --> 01:38:03.000 And there, you know, what would be like your number one improvement that we could make to the system if you are being aspirational about future technology. 01:38:03.000 --> 01:38:11.000 And I'm gonna hand, I think this over to Allen first. Allen? Yeah, I know. It's like, no, shoot. 01:38:11.000 --> 01:38:12.000 Yeah, you got this, Allen. 01:38:12.000 --> 01:38:16.000 So I'm representing Mexico, right? 01:38:16.000 --> 01:38:17.000 Oh yeah, yes. Well, yeah. 01:38:17.000 --> 01:38:32.000 If I were representing Mexico, what I would say is they have it currently set to magnitude 5 and above, and they have a full minute of time as the wave is progressing, 01:38:32.000 --> 01:38:38.000 so if there could be a way to update alerts as they're coming, I think that would be a great way to improve the system. 01:38:38.000 --> 01:38:46.000 Yeah. 01:38:46.000 --> 01:38:47.000 Oh, sure. 01:38:47.000 --> 01:38:51.000 Elizabeth, for ShakeAlert. What would be a key innovation that you would love to see for the system looking forward? 01:38:51.000 --> 01:39:11.000 For me, a key innovation would be a wider distribution of the alerts. So, alert through as we heard from Japan, there's a really nice system of providing really detailed alerts to TV, radio, and other methods. 01:39:11.000 --> 01:39:25.000 And then additionally, coupled with that is sort of follow-up information, and so I think if you have that kind of flows loop of information, it is really helpful for people to understand. 01:39:25.000 --> 01:39:29.000 How and why and when they're getting alerts. 01:39:29.000 --> 01:39:35.000 Thanks, Elizabeth. Frannie, what would be a key innovation looking forward into the future? 01:39:35.000 --> 01:39:42.000 I think I'm still focused on the public information side. I think the science is robust and developing and there's a lot of NSF grant funding that encourages more research. 01:39:42.000 --> 01:39:58.000 There's no help for the public education side at the ground level. So as you point out, there's excellent materials 01:39:58.000 --> 01:40:05.000 from USGS from state OES from FEMA. But who knows they're there? You know, I do. 01:40:05.000 --> 01:40:11.000 I'm an emergency manager. I go to meetings. But what about the people in my community? How do I get them to know about these things? 01:40:11.000 --> 01:40:29.000 And how could we get funding so that USGS might get that 2 min on the TV couple times in the month of April to let different USGS members explain to the public, here's what you're going to get and this is what you should do. 01:40:29.000 --> 01:40:31.000 I'd like you all to be TV stars. 01:40:31.000 --> 01:40:41.000 Yeah, I think we went into science to not become TV stars, Frannie. I think that was the intention and the goal. 01:40:41.000 --> 01:40:51.000 But yes. Thank you. Mitsuyuki what do you think? What are you looking in the future and Japan's done so many phenomenal things with earthquake early warning, 01:40:51.000 --> 01:40:59.000 it's hard to know what the dream would be for you next. 01:40:59.000 --> 01:41:00.000 Yeah. 01:41:00.000 --> 01:41:06.000 For me, yeah, anyway, for JMA, as I said before sending a message to the technology in 2007. 01:41:06.000 --> 01:41:19.000 That means only sound on the text message, but now using current technology, we can send a message using a picture and also video. 01:41:19.000 --> 01:41:30.000 So, oh, so now JMA is investigating how to send a message using the picture, automation or something. 01:41:30.000 --> 01:41:38.000 And so, so now it's our goal in 2030. 01:41:38.000 --> 01:41:49.000 JMA, improve how to send a message, it's, our goal in 2030. 01:41:49.000 --> 01:41:50.000 Excellent. I'd love that. Putting a picture, putting the Drop, Cover, and Hold On iconography 01:41:50.000 --> 01:42:01.000 it's been a dream of mine for quite a while. Oh, Ann just popped off. Oh no, because she was gonna be the last person I asked. 01:42:01.000 --> 01:42:23.000 Hopefully she comes, oh, maybe she's coming back on as a guest. Yeah, we've been looking at that and Janet Sutton and Michelle Wood have done a really good study on including the iconography and we have messages in the future so you know when the technology is available we'll do that well, you know we'll look at that. 01:42:23.000 --> 01:42:34.000 Did you wanna pop back on and give us like, you know, the last, what you would love to see in the future? 01:42:34.000 --> 01:42:42.000 Oh, thank you. My Internet keeps cutting out. I would love to see if we have a future where the Internet doesn't cut out. 01:42:42.000 --> 01:42:59.000 I would love to see more dynamic alerting. So all the things that people have mentioned and thinking about sort of the flow from pre-alert messages, reminding people that there's... [technical problems] 01:42:59.000 --> 01:43:02.000 Oh no, I think we lost Ann again. 01:43:02.000 --> 01:43:12.000 [indiscernible] 01:43:12.000 --> 01:43:13.000 Oh. 01:43:13.000 --> 01:43:16.000 Yeah. Ann do you wanna pop that into the chat? Yeah, it's a little in and out. 01:43:16.000 --> 01:43:19.000 Can you hear me? 01:43:19.000 --> 01:43:38.000 Do you wanna pop that into the chat, maybe? But I did hear post-alert messaging, which, I, did write a paper and I'm working on a follow-up paper actually on 5-years post-alert messaging for ShakeAlert because you know, post-alert messaging is so critical to successful alerting I feel. 01:43:38.000 --> 01:43:49.000 It's, you know, with Shake Alert, I always joke, you know, we're either right or we say we're sorry in different ways and then explain how the system worked or didn't work to meet people's expectations. 01:43:49.000 --> 01:43:59.000 And that's, the art and the science of post-alert messaging. Jessie, did you, I want to throw it over to you if you had any last comments? 01:43:59.000 --> 01:44:14.000 Once again, thank you all so much to our wonderful speakers in this session. I agree with everything. I think communication is really the key thing especially for the early warning system here. 01:44:14.000 --> 01:44:24.000 I think that's kind of the next, big source of improvement, we can focus on. 01:44:24.000 --> 01:44:32.000 But yeah, it is almost 4:45pm. I believe it is time for us to officially wrap up this session. 01:44:32.000 --> 01:44:41.000 So once again, thank you so much everybody. And then I will pass this on to, Sarah. 01:44:41.000 --> 01:44:59.000 Thank you so much. Thank you to all of our speakers and our wonderful moderators. And without further ado, I'm gonna bring us on to our last presentation of the evening because we have the privilege of having Mitsuyuki Toshiba from Japan Meteorological Agency with us today and because it's a year ending in four, so we didn't even get one day into the year without a 01:44:59.000 --> 01:45:04.000 major earthquake happening we now have a special presentation from Mitsuyuki Toshiba on the 2024 Noto peninsula earthquake. 01:45:04.000 --> 01:45:12.000 So let's just take it away John. 01:45:12.000 --> 01:45:19.000 Thank you very much for giving me the opportunity to explain overview of the 2024 Noto Peninsula earthquake. 01:45:19.000 --> 01:45:21.000 This earthquake occurred January 1st, New Year's Day in the year 2024. 01:45:21.000 --> 01:45:35.000 Magnitude 7.6 event on JMA scale and moment magnitude 7.5. So, 14 time is up for 10 p.m. 01:45:35.000 --> 01:45:45.000 Two-hundred thirty-two people were killed and even now 22 people are missing. Many kind of disasters are reported. 01:45:45.000 --> 01:46:00.000 This picture from our newspaper company. For example, collapse of building, landslide, and this is probably due to tsunami and nuclear park shop as 300 km from procedure area. 01:46:00.000 --> 01:46:05.000 So, not a plate boundary. It's not to create a boundary. 01:46:05.000 --> 01:46:13.000 And no border plate nearby. No volcano nearby. So, 2024 Noto Peninsula earthquake. 01:46:13.000 --> 01:46:21.000 This is the largest intra-plate earthquake in Japan in 100 years. 01:46:21.000 --> 01:46:32.000 So, this shows aftershock activity including aftershock activity. Aftershock area is about 150 km. 01:46:32.000 --> 01:46:40.000 So rupture occurred 150 km. So what was an important characteristic of this event? 01:46:40.000 --> 01:46:46.000 Swarm activity started a couple of years before at HERF. 01:46:46.000 --> 01:46:48.000 Okay. 01:46:48.000 --> 01:47:01.000 Including magnitude 6.5 event in May, about 8 months ago. So some activity started a couple of months years ago and last May, magnitude 6.5 event and this magnitude, 7.6 event. 01:47:01.000 --> 01:47:08.000 [noise] 01:47:08.000 --> 01:47:15.000 After the magnitude's 7.6 event aftershock activity is something like that very active aftershock. 01:47:15.000 --> 01:47:16.000 [noise] 01:47:16.000 --> 01:47:22.000 So looking back to my old data, 2007 magnitude 6.9 event; Mw 6.6 event occurred around the shear. 01:47:22.000 --> 01:47:38.000 We call this event 2007 Noto Peninsula earthquake and you been to it's something like that. 01:47:38.000 --> 01:47:49.000 So rupture area may be overlapped and in 1993 magnitude 6.3 event occurred around the shear. 01:47:49.000 --> 01:48:01.000 And active 4th event reported along northern coastal area of Noto Peninsula, probably this 4th activated it. 01:48:01.000 --> 01:48:07.000 And this shows aftershock activity. 01:48:07.000 --> 01:48:17.000 2024 Noto Peninsula earthquake as compared with other major earthquakes in Japan, for example, 1993 SW off Hokkaido earthquake, 01:48:17.000 --> 01:48:29.000 and 1983 Akita earthquake magnitude 7.7 event and 2016 Kumamoto earthquake, and 1995 Kobe earthquake. 01:48:29.000 --> 01:48:35.000 So, Kobe earthquake aftershock activity not so strong. 01:48:35.000 --> 01:48:41.000 And G2, SW of Hokkaido earthquake and Akita earthquake they are 01:48:41.000 --> 01:48:52.000 not intraplate event, plate boundary event. But Noto Peninsula earthquake is intraplate event. 01:48:52.000 --> 01:49:04.000 So, Noto Peninsula earthquake event aftershock activity quite active. So what was the important feature of this earthquake is the large uplift. 01:49:04.000 --> 01:49:18.000 This is from InSAR data analysis and GNS analysis. So, from InSAR data quite a lot uplift reported and actually all meta uplift observed at this spot. 01:49:18.000 --> 01:49:28.000 And this spot was previously at the point, this point was previously a fishing port, but at the present no water 01:49:28.000 --> 01:49:33.000 because of the large uplift. 01:49:33.000 --> 01:49:41.000 And lab job process, some organization analyze the lab job process of this event. JMA and DPRI and NIED commonly estimated relatively small slip around the local area. 01:49:41.000 --> 01:50:03.000 Large slip area are reported by those east-part and southwest part [indiscernible]. Next, I already showed earthquake early warning. 01:50:03.000 --> 01:50:09.000 So [indiscernible] I like show this video. Okay, this video taken at [indiscernible] that means here. 01:50:09.000 --> 01:50:21.000 [video] Okay, here we go. 01:50:21.000 --> 01:50:40.000 [video] Sound and more strong shaking will come. 01:50:40.000 --> 01:50:45.000 [video] 01:50:45.000 --> 01:50:55.000 [video] 01:50:55.000 --> 01:51:10.000 Okay. So, magnitude 7.6 event occurred at 10, and 4 minutes before magnitude 5.5 event occurred. 01:51:10.000 --> 01:51:18.000 So because of the magnitude 5.5 event warning issued and 4 minutes later, 01:51:18.000 --> 01:51:31.000 magnitude 5.9 event occurred and that came later magnitude 7.6 event occurred. 01:51:31.000 --> 01:51:39.000 So this process was triggered by magnitude 5.9 event because of the magnitude 5.9 event, a warning issued. 01:51:39.000 --> 01:51:53.000 Not opinion GRJ, and second warning issued to these area and third warning issued to these area. Okay, so because of it 01:51:53.000 --> 01:52:05.000 the past warning is a [indiscernible]. Second one as strong as the shaking everywhere. 01:52:05.000 --> 01:52:14.000 For this event 2, 3, 26 forecast and 3 warning issued, past warning, second warning, and third warning. 01:52:14.000 --> 01:52:27.000 So, here shows a latitude, longitude, local depth, and magnitude. So from 90 assurance; under 20 assurance, 20 assurance means a second warning. 01:52:27.000 --> 01:52:37.000 So no change of hypocenter, location, no change of magnitude, but it was updated this is due to the problem mentioned. 01:52:37.000 --> 01:52:45.000 But other warning also, no change of the hypocenter data, but it was updated this is due to the problem mentioned. 01:52:45.000 --> 01:52:51.000 Okay, tsunami early warning issued, 01:52:51.000 --> 01:53:00.000 so, 3 minute later earthquake early warning updated and also tsunami warning 01:53:00.000 --> 01:53:14.000 issued. Tsunami warning advisory issued to the western coast of Japan. 01:53:14.000 --> 01:53:22.000 And all 22 tsunami early warning updated to major tsunami and this is a legend and expected high tsunami. So, [indiscernible] 01:53:22.000 --> 01:53:30.000 Okay, so this is some observation at 01:53:30.000 --> 01:53:34.000 Wajima. Kanazawa, Nanao, and Nigata. 01:53:34.000 --> 01:53:48.000 It's a very steep increase of tidal gauge. So, because of something wrong on the tidal gauge and no data transmission after that. 01:53:48.000 --> 01:53:55.000 So, very steep increase at Wajima Port. 01:53:55.000 --> 01:54:12.000 So, 2024 Noto peninsula earthquake this is the largest intra-plate event in Japan in less than 100 years and earthquake activity started a couple of years before and the biodata rupture of 150 km, 01:54:12.000 --> 01:54:20.000 and active aftershocks and large approved to our observed. And earthquake early warning, so more than several second before peak shaking 01:54:20.000 --> 01:54:29.000 it was triggered by magnitude 5.9 event, which [indiscernible] saw 7.6. 01:54:29.000 --> 01:54:38.000 And the tsunami are you on the post-warning issue, a whole, and it was updated to major tsunami at 01:54:38.000 --> 01:54:53.000 0:22 p.m. Good. That's all. Thank you very much. 01:54:53.000 --> 01:55:06.000 That was excellent. So. Last call, would anybody like to ask some questions about this really interesting earthquake that just record. 01:55:06.000 --> 01:55:17.000 Jolly, you wanna come up and ask your question? 01:55:17.000 --> 01:55:18.000 I think you're muted. They go. Yep. 01:55:18.000 --> 01:55:24.000 Sorry, can you hear me? Can you hear me okay? So, thank you very much for the presentation. 01:55:24.000 --> 01:55:33.000 You focused on the seismological aspects. Can you comment at all on any building damage or other kinds of infrastructure damage. 01:55:33.000 --> 01:55:37.000 And my particular interest, were there any fires that occurred afterwards? 01:55:37.000 --> 01:55:42.000 Miss Yuki, you wanna turn on your camera and unmute and I got it. 01:55:42.000 --> 01:55:43.000 Perfect. 01:55:43.000 --> 01:55:53.000 Alright, so, Hannah, hi, Rita. Yeah, so, I'm not sure, but about 100 met, or 100 meta or something, it's wired. 01:55:53.000 --> 01:56:00.000 At the time, because of the know what I don't remember, one is one, approved of the land. 01:56:00.000 --> 01:56:10.000 So water is not a bearable. Also it's water in Giba. It's not available because of larger uprift. 01:56:10.000 --> 01:56:20.000 So because of reasons, they cannot stop fire. So 100, I'm not sure about that, but it's a 100 metre, 100, I'm not sure, of what's a major, but it's a 100 data or 100 metre, fire yet. 01:56:20.000 --> 01:56:23.000 After the ask quick. 01:56:23.000 --> 01:56:26.000 Buildings damaged or infrastructure damage, water systems, for example. 01:56:26.000 --> 01:56:36.000 Yeah, quite a lot because of the not opinion, you know, it's a lot of pain in generation. 01:56:36.000 --> 01:56:39.000 It's a very old, not a pen in generation, it's a one where it's a very old order prices. 01:56:39.000 --> 01:56:45.000 So many all the building. So in Japan, building code is the departed in 98. 01:56:45.000 --> 01:56:56.000 So, many all the building people are 98. A yeah more than 10,000 or something the cast. 01:56:56.000 --> 01:56:58.000 100000, thank you very much. I got those. 01:56:58.000 --> 01:57:08.000 Yeah, sorry. I don't think you know about where the number. But anyway, many of the building were cut off to that. 01:57:08.000 --> 01:57:13.000 Thank you. Tom Heaton. You wanna take it away? 01:57:13.000 --> 01:57:24.000 Good. I was curious about the interpretation that this was an introplate versus interpolate versus interpolate versus interplay versus interplay versus interplay. 01:57:24.000 --> 01:57:30.000 Isn't this the boundary between, versus interplay? Isn't this the boundary between, North America and the Eurasian plate runs right through there. 01:57:30.000 --> 01:57:40.000 It's kind of like the the earthquake of 1964. Why is this not a interplay earthquake? 01:57:40.000 --> 01:57:48.000 How, how deep was it? 01:57:48.000 --> 01:57:51.000 Oh, I believe you're muted. 01:57:51.000 --> 01:58:10.000 A 4 car depth to the 16 kilometer and it's now a predator boundary, it's days think that from the with the coast of Hokkaido with the coast of Tohoku. 01:58:10.000 --> 01:58:15.000 Non-Citeer off. 01:58:15.000 --> 01:58:16.000 So, huh? 01:58:16.000 --> 01:58:21.000 But. Just the phone. Just to follow up on Tom's question, isn't there a platelet? 01:58:21.000 --> 01:58:28.000 Just offshore Noso Hunto, the Amor, A more plate. A platelet. 01:58:28.000 --> 01:58:32.000 So I think Tom, that's a good point. To characterize this as interpolate or interpolate. 01:58:32.000 --> 01:58:35.000 Is an interesting question. 01:58:35.000 --> 01:58:44.000 Yeah, it's, Yeah, Yep, Predator Bandary. 01:58:44.000 --> 01:58:50.000 Okay, so it's a, European threat. Yeah, more great. 01:58:50.000 --> 01:58:55.000 But there's a small platelet in between. There's an 8. Yes 01:58:55.000 --> 01:59:08.000 So at the present they don't know I don't know the report. There is a microwave around the earth. 01:59:08.000 --> 01:59:09.000 Okay. 01:59:09.000 --> 01:59:16.000 Oh, so anyway, I don't I don't remember the report of, microplate around there. 01:59:16.000 --> 01:59:19.000 Alice, would you like to ask your question? 01:59:19.000 --> 01:59:28.000 Sure, thank you. Thank you for this really fantastic summary. So I have a question about the effect of uncertainties in in rapid available source models for this event specifically and was very fascinating to see how the USGS model, for example, evolved. 01:59:28.000 --> 01:59:46.000 The first model was available had 2 snip patches which were inferred using telescope data and using genetic data, one of the slip patches was removed and most of sleep was constrained actually. 01:59:46.000 --> 02:00:04.000 And the part of the fault which is an unsure. So I'm wondering if this effect of source uncertainty has an effect on early warning or are you blending what's your what's your take on that because it's very puzzling to observe. 02:00:04.000 --> 02:00:05.000 Yes. 02:00:05.000 --> 02:00:08.000 Okay, do you mean the so structure process? Well, I'll ask you, do you want him? 02:00:08.000 --> 02:00:09.000 One general for your rapid response. 02:00:09.000 --> 02:00:23.000 Okay. So, so sorry, I'm not an expert of the source process, but anyway, so, several, segment, at least the 2 segment, activated. 02:00:23.000 --> 02:00:39.000 Yeah, why is it going to north, north east wines going to the southwest? So, some report, it's, time, with, about, second. 02:00:39.000 --> 02:00:46.000 Alright, there was a slow notation phase. Is this complicating? Your systems did you have it? 02:00:46.000 --> 02:00:47.000 Operations that are in operation is the same something that is concerning this high degree of complexity. 02:00:47.000 --> 02:00:56.000 That's a, S, and after 30 s, the second, the second, the second, the, UP, yeah, do you mean the ask a card warning or not? 02:00:56.000 --> 02:00:57.000 Yeah. 02:00:57.000 --> 02:01:10.000 Oscar, you want you? So, at the time, so, oh, yeah, as I talked to before, there's a, somebody, been, the, multiple point 9 event, okay, and then 30 seconds radar, which is 7.7 point 6 event o card. 02:01:10.000 --> 02:01:20.000 So, but, it's almost same. And also the to work to where, so, using a problem. 02:01:20.000 --> 02:01:28.000 And the warning, there's expanding using a product. 02:01:28.000 --> 02:01:29.000 Okay, thank you. 02:01:29.000 --> 02:01:34.000 So. I realize it's 5 PM. 02:01:34.000 --> 02:01:41.000 Of course, on the East Coast and it's even later on the West Coast that's even later on the East Coast and in parts farther East 02:01:41.000 --> 02:01:46.000 so people want to take off you absolutely go call it a night, feed your children, but if there are any last questions, I'm sure we could answer a couple of them. 02:01:46.000 --> 02:01:59.000 Are there any other questions? 02:01:59.000 --> 02:02:11.000 Okay, going once, going twice, going three times, then I'm gonna say thank you so much Hoshiba-san. This was a great summary. 02:02:11.000 --> 02:02:16.000 It was really nice to be able to learn everything that happened. Thank you to all of our speakers today. 02:02:16.000 --> 02:02:25.000 Thank you to all of our moderators and thank you to all of our attendees who asked great questions and had great chats. 02:02:25.000 --> 02:02:29.000 Thank you so much everyone. Have a wonderful day, morning, afternoon, evening, and we'll see you again tomorrow. 02:02:29.000 --> 02:02:39.000 Thank you very much. Thank you, Sarah. Thank you!