 Okay, I'm afraid we need to move on. Thanks, Mike, very much for stimulating talk. Okay, so next talk will be given by Phaedra Upton from GNS New Zealand, and we'll be hearing about the Kaikoura earthquake. Okay, thank you, Greg, and it's great to be back in Boulder again. But yeah, today I'm gonna talk to you about the Kaikoura earthquake. This occurred about 18 months ago on the 14th of November, 2006. It was a magnitude 7.8. It was felt through most of central New Zealand, more intensely by some than others. I have to admit, it woke me up. I thought, yeah, it's not a bad earthquake and I'll be back to sleep. My colleagues got up, raced to work. So yeah, when my friend called me at seven o'clock the next morning, I realised I raced to work too. And I ended up doing some GPS fieldwork after the next several days. But today I'm gonna talk about, well, I'll give you an introduction to the earthquake and then I'll talk about the landslides, the landsliding response and what we've learned in the 18 months since then. This landslide, I think, is the one that Michael just showed on the amazing satellite data. So we were really lucky, this earthquake happened at four minutes past midnight. So there were two deaths, both in houses, so damage by falling stuff in the houses. About 50 or 60 injuries. There was one train driver who was stuck in his train overnight between two slips, and it had been much fun. So I'll just give you a really quick introduction to the tectonic setting, why the earthquake happened. So New Zealand is situated on a plate boundary between the Australian and Pacific plates. The plate boundary runs off to the east of the North Island and along the Alpine Fault on the west and south island. And this earthquake happened in between those two, right in here, which is a mulberry fault system. There's a series of parallel strikes of faults that are taking up most of the plate motion. This is one of the faults along here, the Arwick Tree Fault, which early in its history had a lot of uplift and it's now largely strike slip. So the earthquake, in some ways it was unexpected in that we've been focused on the Alpine Fault for so long, the Hikarangi Subduction Zone, these are two big plate boundary faults. In Christchurch, Duffield and Christchurch occurred, which hit on faults that we weren't really aware of. And then BAM, another one on a fault that we weren't, this is the seismic hazard model. So it's obviously a region that we know was a high probability of a large earthquake. But as I said, it wasn't one of the two major faults. The Hikarangi seems to have disappeared. So large earthquake, there wasn't unexpected, but it was a very unusual event. And one of the things was it wasn't the faults that we would have expected to rupture in that region, particularly the Hope Fault, which is one of the mulberry faults. It's the most active of the mulberry faults. It didn't rupture at all. Instead, a series of faults to the south of it. It's our east of it ruptured. And then as they propagated north, the rupture did move on to faults that we knew were very active. So here's a map of what we've learnt in the last 18 months. So the earthquake, the epicenter was down here in North Canterbury. And in the red are all the faults that we know ruptured as part of that earthquake. So it started down here, propagated to the north. We've mapped a few offshore, then came up through here, up towards Wellington. So it was a 7.8. The shaking lasted two to three minutes, depending on when you were there. There were multiple effects on the landscape. About 180 kilometres of surface rupture, up to 12 metre offsets on some of the faults. That straight slip up to nine metres of vertical on the puppeteer, which was again what Michael showed in the satellite and I've got some photos of that. There was a large part of the coast was uplifted. And because it's so complicated, it was variable how much uplift. Locally, there were some tsunamis and tens of thousands of landslides and also many landslide dams because of course this area is really quite mountainous. And liquefaction was locally significant but nothing like what we'd experienced in Christchurch. So here's a couple of examples of what it looked like on the ground. This is where the puppeteer fault that Michael showed on the satellite image comes out to the coast. It actually comes out as two strands. That's Waipapa Bay. And this area here was a pop-up of about six metres. See the road and the railway. So that's just looking down the railway there. That's looking at the road and the railway from the air. This is one of the faults down South Canterbury near the epicentre. And this here is Bluff Station and Greg and others have known the landowners here and have been here. And Bluff Station is on the Kekarangu fault which had not quite, not here, but further north had 13 metres of offset. But here, this house, oops, what do I want? So this house was right on the fault and you can see how much offset on the driveway. And there's a lot of stories. I couldn't spend my whole time telling stories, but this is a good story. So the bum worker was in that cottage. And because the earthquake started to the south and so he was woken up by the shaking. He got out of the house and then 30, 40 seconds later, the fault ruptured right through his house and that's the offset. That's his house. The fault right through it, it moved it off its foundations by about eight metres. The wooden house and you see the windows aren't even broken. So, you know, wood is good. This is the puppeteer fault. So this is, as I said, that satellite image. This was the uplifted side. This is the downside. So it crosses the Clarence River coming through here. It comes along here. There's a farm track we went down that, you know, you stood at one end of the farm track and you looked down about six metres and the owners were saying, this was flat. And it was hard to believe except, you know, when you. So, yeah, there's approximately eight metres of strike slip here and about six metres of offset. It crosses the Clarence River in a couple of places. This ended up uplifting the downstream side. So we got a hauling down here. It created a new rapid on the Clarence River. That has since, I'm afraid that as far as I know, we haven't been tracking that in that point. We've been quite busy, but that has cut back. It's in soft rock and has cut back really quickly. So as I said, yeah, there was coastal, the coast uplifted and went down. So high amounts of uplift here, some subsidence here. See, really dramatic uplifts along the coast. The seaweed lines gave it. So there was, you know, in the week, a couple of weeks after it, there was just heaps of work happening to try and track all this, things like this. Seaweed and that were time dependent. We had to get in there and map them. There was local tsunami, but because one, it was low tide and two, the model of the coast went up rather than down, the potential tsunami was a lot less than it could have been. This is the permanent ground movement that we've measured in the last year since the earthquake. The scale bar is one metre. So this area here moved about six metres closer to Christchurch. And it's just really complicated, which is because of all those multiple faults. So moving on to landslides, the topic of this talk, it was immediately obvious that landslides were going to be a major issue. That first day after the earthquake, there were many recurrence, reconnaissance flights. The earthquake geologists managed to get on flights with the media as soon as it was light. And so, yeah, saw lots of landslides. But there was concern about the potential for the road access, the rail access, and landslide dams. And it was wet the couple of days after the earthquake, it really poured with rain. So that was, again, real concern for breaching of landslide dams. And so there were a couple of groups, one out of Durham, Tom Robinson, another out of the USGS. These are the ones I know about probably others working with colleagues in New Zealand that were trying to model the landslides in real time to see if we could, you know, before we could actually go and map everything to see if we could actually get a realistic idea of how many, of where the landslides were. This map here is from Tom's paper, but it's the result of months and months of work, so it's not. So it shows the shaking intensity. The epicenter was down here, the energy propagated to the north. So there's all map landslides and road blockages and landslide dams in the mountains. So there's two papers that have just come out. There's a special issue of BSSA on the Kaikoura earthquake that's just come out. So there's two papers that describe this, near real-time modeling of landslides. So what they did during the days after the earthquake and how successful it was. So I'll just talk about this Robinson et al, what their work. This is one of the many landslides. We actually saw this one as we were flying over to do GPS work. This is just an aside. There's the railway line coming into the landslide. There's the railway line having been pushed out by the landslider because it should be there. So it's actually being pushed out. So this is the Durham work. They had a model 21 hours after the earthquake. They based it on the USGS shake map. So they had an intensity from that early model. They modeled the likelihood of landslides and from that where you'd likely to get road blockages and where you were likely to have landslide dams. And then once the GNS shake map was developed, about 72 hours after the earthquake, they built another model. So with a more detailed shake map, yeah, shake map and again, the likelihood of landslides and the road and river blockages. We knew by this time, we knew on Monday morning that Kaikouda was completely blocked off. There was, so the main road comes south along the coast to Kaikouda and then go south along the coast and then out into the mountains. And there was, I think, 30 major landslides between north and south of Kaikouda. There's another inland road and that was also blocked by landsliding. The railway, as it runs along the road, was also completely blocked. So the residents of Kaikouda was, there was emergency helicopter for anyone that needed it and then within a few days, the Navy frigate had gone in and so people that wanted to leave could leave. But it wasn't, there was no road open for at least a month and then that was just an inland road that was open part of the day, there were convoys. So yeah, it was blocked off for a long time. So following their initial models and once we had a much better idea of the actual distribution of landslides, this is kind of what they got from their modelling. Well, they obviously highlighted that landsliding would be widespread, which we expected that the Kaikouda was likely to be cut off as it was. And these results were used by the first responders to try and figure out where they needed to go and look and where they might need to go and find people. Graphication showed that while the models captured a large percentage of the landslides that occurred, they overestimated and all the models overestimated the number of landslides and where they were gonna be. So yeah, so obviously we need some to modify the models and automate if we're gonna, if these are gonna be useful tools in a future event. So year after the earthquake, what can we say about what did control landsliding? And it turns out to be geology and relationship to faults. So there's three main rock types in this area. There's quaternary gravels, sands and silts, which are forming terrace deposits and alluvial bands. There's the red, the neogen sedimentary rocks that tend to be sandstones, limestones. This is the puppeteer fault between the, and there's the basement tool there. So most of the green is this basement greywack that was laid down at the mesozoic. So this puppeteer fault, which was one of the most, had the most dramatic offsets on that we've looked at, that was mapped as an inactive fault because there was no evidence that it was active and it was mapped as a boundary between these two units as we know now, quite an active fault. And there's a lot of work being done to now go back and look at the paleos seismicity of these faults. So the landslides tended to be in these neogen sedimentary rocks like this one here. This is the limestone and it's slowed down through here and then in the greywackies. And there was lots of slumps, lots of rock slides and there was some big rock, big landslides. A lot on the coast, as we've seen, there's the main road and the railway. So there was, yeah, as I said, there were about 30 or so of these big, big landslides. It took 10 months to reopen the railway. It was reopened to freight only and also they worked really hard to get that open first because that facilitated working on the road. The road was open with several big one-way sections but it was open like the 22nd of December. So it was open in time for Christmas traffic, 13 months after the earthquake. And it turned out that a lot of the proximity to faults was a really significant feature for landsliding. So this is just a plot of the number of landslides versus distance from faults and in meters. And so a lot of them were right at the fault zone. This is the seaward landslide. This is on the Puppeteer fault. There's the fault trace coming through and that's where the landslide has been triggered. So as I said, there are some very big landslides. This was the largest one, the Harpuku up on the mountains. That's a source area here. These both post-earthquake images. Again, fault running through the source area for the landslide. And this formed one of the biggest landslide dams and the early days was one of the most concerning to the civil defence, et cetera. In the end, it was relatively stable for quite a while. So this was in the Greywacky large source area. And as I said, it dammed the Harpuku River with a large landslide dam like in the back there. This is another large landslide. This is the leader sump. This is in the Neogen sediments. Again, big slump, big lake in the back. This, it completely dammed the river and there was no flow through here for about, not until May I think, and then it broke through once it got a bit wetter in the winter. And again, it's not shown on this map. I think it's shown on the next slide. Again, there's a fault, the leader fault. In this case, it's actually at the base of the landslide rather than the top. So there's probably a combination of the close proximity to shaking and also the damage zone, so it's a fault. So weakening the rock that meant that these two, that the fault proximity was a real risk factor for these large landslides. So after the earthquake, we've got lots of landslides, landslide dams. Lots of moose and material up on the hills. And since then, we've had four X cyclones come through. So in 2007, we had cyclone Cook and then cyclone Debbie. I think they were X by the time they got to us. So a lot of concern about what these landslides and debris flows were gonna do. And indeed we did have, so the rain from this event actually caused the Harpuku Dam to overtop and start to erode down. But it wasn't a dam burst, so it wasn't a dramatic. It didn't endanger anything downstream, but it has started the erosion of that feature. And lots of material from the coastline from these landslides was remobilized. So here's the rebuilt road with debris on top of it. That's a trace of the Papatier fault coming through there. And another place where the road was blocked. And again, this year, a few months ago, this was X cyclone Gita. And it was tracking west of New Zealand. And so they weren't really sure where it was gonna go. In fact, Wellington had a huge, there was big heavy rain warnings for Wellington, but it actually ended up kind of splitting in it when north of Wellington and then it basically tracked right over Kaikoura, which is unfortunate. So it was rainfall was highly low clothes on the coast. Kaikoura got 270 millen, 12 hours. And huge again, a large amount of debris ended up on the road. So the road was closed for over a week. And most of this, of course, occurred on slopes that had been damaged by the earthquake. Speed up. So this is pre-Gita, post-Gita scar on the landscape. Doesn't even look like that one, impacted the road post-earthquake. But it's a slightly different angle, but that there is that flat there. And you can see heaps of that material got mobilized by this event and down onto the road. And there's still a lot of material up there. So that kind of brings us to one of the ongoing hazard from this earthquake is the sediment cascade. This transport of all this material that has been loosened and will either deposit it at the base of a landslide or loosened on the hill slopes is gonna get transported to the sea over the next, or we don't know, years to decades. And these pictures are from Wen Shuang, so showing pre-earthquake, so pre-10 years ago, all the landsliding that occurred as a result of the earthquake and then how it's been transported downslope and into the river system there. And in our case, it's largely being transported to the coastal plain across our State Highway One. And we expect that these hazards will last for years or possibly decades and phases. So there's a huge amount of monitoring and modelling and work being done. We've got a large project to look at these landslides, to understand the landslides, understand the mechanics of why they occurred, but also what's gonna happen to monitor and understand what's gonna happen over the next years. So it involves a lot of lidar, structural promotion, field work, monitoring the rainfall, trying to look at the relationship between rainfall and movement of material, numerical modelling, and of course working with the communities and the stakeholders and the regional council with transit and everybody else. So we're gonna be using Eros, which is a bed load model to move material around that's been developed at University of Wren. There's a website there or if you're interested, you can ask me. So we can move the bed load, we can road the banks and look at the sediment transport downslope. And one of the reasons we chose it is because it's been calibrated against an early event from New Zealand, the Mount Adams landslide, the large landslide that occurred in the 80s. This wasn't earthquake, it just happened. It dammed the Pro River River and then there was an outburst flood with an expert rainfall. And so this sediment has been moving down the river system for the last 30 years and it's been monitored really closely. And so Tomas did his PhD on modelling this and calibrating Eros to this natural example here. And in our breakout yesterday, there was one of the things that was raised is something we need to do is to develop tools that the stakeholders can use. And that's definitely one of the planned outcomes from this project is that we can end up with some sort of tool that we can give to, we can either use with or give to depends on how it turns out, to butt to their councils and the people that are dealing with the infrastructure in the downstream area. So you can run ensemble models and work out the probability of, so this is for pro river, probability of flooding and probability of where you're gonna get the alluvial material going down. Okay, so rather than summarise, I'm going to point out why this is super important to us. It's important to, well, other places, I mean, Taiwan, Winshan, everybody has this, but we also have the alpine fault. And the latest paleo seismic study for the alpine fault suggests that there's a greater than 50% chance of a magnitude eight in the next 50 years. The last event on the alpine fault was pre-European history, but it's dated very accurately with tree rings to 1717. We had a 300 year anniversary conference last year, which turned into a Kaikoula conference, but so yeah, the last event was 301 years ago. We have a record going back 8,000 years in places, a combination of on fault studies and lake studies and wetlands. This is just going back to 500 years BC along the fault, and these are the previous events. So with mapping these lakes and these other marshes, we've got a really good idea of how often this fault ruptures, how often it ruptures a couple of, you know, whether it ruptures the whole 400 kilometres, how often it does that. So it's definitely time that we really thought about what we're gonna have to do. The lakes also give us the sedimentary response to the event. The alpine fault is right along the edge of the mountains. So these lakes, what they see is where they see a shaking deposit, then they see a post seismic layer, which is all the landslide material, which is high organic, so it's the trees and stuff, and then the a seismic period. This period for the alpine fault, so when that sediment response lasts 40 to 60 years on average, I think that's more than Kaikoura's gonna be. It's gonna be a big event, and about 40% of the sedimentation happens during that time. So that's why we are gonna use Kaikoura as we learn as much as we can from it. Thanks. Beautiful, thank you, Fedra. Questions for Fedra? You mentioned that the models had overpredicted the slides, and I was wondering if you could follow up with a little bit more detail on that. I figure a crustal fault of that magnitude with that much oblique slip in the wheat bicks of New Zealand would be a great landslide generator. So what's the nature of the model that you were referring to? So that was like the statistical models, and well, I'm not an expert on those, but I would have to, I refer you to Robinson and Alan, the other papers, but they were just purely based on the shaking magnitude and the slope, and so there was no geology in there. There was no idea that the proximity to the faulting made landslides more likely. So those initial models predicted over 100,000 landslides, whereas I think that's like 20 or 30,000. So, yeah, there was heaps of landsliding, but it just wasn't as extreme as those particular models predicted. Thanks for the talk, it was fun. You mentioned that they did a little bit of offshore mapping to look at displacement offshore as well. Did they have enough free merc week pedimetry to look at possible offshore landslides? They had a wee bit. So it was a real fluke that one of the research vessels was offshore, North Island at the time, and so they came down about a week after and did some mapping. They had enough that they knew where the faults were, and in the Kaikoura Canyon, they did have pre and post, and I think that's been published, and if you come and talk to me, I can find it for you. So they did see a large, so the edge of the canyon failed, and they could track that down, and they sampled the turbidite from that in that cruise. Terrific, let's thank Fedora one more time. Thank you. Thank you.