Kia ora, today we are in  Victoria University in Wellington  

New Zealand, and I am with Jamie Howarth, my  ex-colleague and good friend. Kia ora kotou,  

my name's Jamie Howarth, I'm a paleo seismologist  and a senior lecturer here at Victoria University  

and I'm excited to be here and talk  to you a little bit about the work  

we've been doing on the Alpine Fault and in  particular how that work informs the likelihood  

that we're going to see a large  earthquake on that fault in the future.  

Now my understanding of the Alpine Fault is it's a  globally significant faultline that generates very  

large earthquakes potentially, and so of course  it's really important for scientists to understand  

how often does that happen, when were the last  earthquakes, how big were they? You know and  

that that research has occupied many people for a  long time in a very difficult environment because  

the Alpine Fault crosses an alpine region in New  Zealand it's very inaccessible, there's lots of  

dense forests, wild remote areas in new zealand,  it's quite hard to get information.Aand that's  

what's so interesting about what Jamie's been  doing because he's been able to sort of in  

spite of those conditions find out some really  important information about the Alpine Fault.  

So Jamie I'm going to hand over to you and you're  going to take us through some of the ways you've  

done your research first of all, and then what  you've found out over time because of your methods  

and your approach and what that means for us.  Fantastic so from space the Alpine Fault is a very  

obvious and really straight feature that basically  cuts through the South Island. It's a plate  

boundary fault so it forms the boundary between  the Pacific Plate in the east and the Australian  

Plate in the west. And it's a particular type  of plate boundary - it's a transform plate  

boundary - and that basically describes the sense  of motion between these plates. The Pacific Plate  

is actually moving southwest relative to the  Australian Plate which is moving northeast. Now  

that plate motion is accommodated elastically  so that means that the fault is stuck  

and it's building up the stress associated  with that plate motion that's continuous.  

So it's building up then suddenly it's released  in a big big earthquake. Okay and we can see  

the Alpine Fault runs from offshore at Milford  Sound north along the base of the Southern Alps,  

through the central upper South Island where it  does a little bit of a dog leg before running  

offshore at Cloudy Bay near Blenheim. Okay so  there is a little bit of upward movement then  

on the fault to produce the mountains, but it's  not all upward. There is a component of vertical  

motion or compression and it's that compression  which has built the topography that we see.  

Okay let's start with what was previously known  before you did your work here about the fault and  

where that was studied. So a really famous record  of past earthquakes on a fault actually anywhere  

in the world, comes from the very southern onshore  part of the Alpine Fault from these sites called  

John O'Groats and Hokuri Creek. And if we  zoom in to John O'Groats I can explain why  

these are such good sites for recording  earthquakes. So we can see the stream here  

that is running across the Alpine Fault. And  what happens when we have a large earthquake  

is that this stream is offset and essentially  dammed, and as a result of that damning we have  

water and sediment spilling out over this wetland  and earthquakes are recorded then as interbeded  

sediment layers in the stratigraphy of this  wetland that alternate between sediments produced  

by the river and sediments produced by the  wetland itself. Okay so some of the kind of rocky  

material, gravel, sand that sort of thing? Yep,  so that would be an earthquake interbedded with  

very organic material formed by the little plants  that grow on the wetland. By using sediment cores  

that basically act as a time machine going back  through time, we're able to identify those layers,  

date them with radiocarbon dating and produce  a history of large earthquakes on the fault.  

And we've been able to do that at these  two sites John O'Groats and Hokuri Creek  

slightly to the north, over a time frame  that spans the last eight thousand years  

or 24 Alpine Fault earthquakes. So here  we have a plot of the timing of the last  

24 earthquakes on the Alpine Fault. So on the  x-axis here we have time in calendar years so  

the present day is all the way to the right of the  screen, and then we go back to around about eight  

and a half thousand years ago and then each one of  these distributions you see here is our estimate  

for the timing of an earthquake. And one of  the really neat things about this record is  

that the historic record spans this tiny little  area right up at the right hand end of the plot,  

so we're able to actually look at the recurrence  of these earthquakes over a really long time frame  

and that gives us a lot of confidence that we  actually understand how this fault behaves. There  

is around about 300 years between each of these  events and that timing between events is pretty  

constant as we go back over the record. So 300  years, and the last earthquake was? In 1717 around  

about 300 years ago. Okay yeah what does that mean  just from this information? What does that mean  

about the probability of an earthquake let's say  in the next 50 years? Well so because we've got  

a really long record of earthquakes from this  point at the southern end of the fault, we can  

use it to forecast the likelihood that we will  see an earthquake at some time in the future.  

So the probability from this record of  seeing an earthquake at this site on the  

Alpine Fault is around about 29% in 50 years.  So we can be reasonably confident that we can  

forecast when the next event is likely to occur  here, but what we really want to know is how much  

of the fault breaks during those earthquakes,  and whether or not the forecast from here is  

applicable along the length of the fault, Right,  because there could be a place way up north where  

there's a different record because the earthquakes  were not the same. Exactly! As it turns out  

lakes adjacent to the Alpine Fault provide  really excellent archives of past earthquakes.  

So the cool thing about these lakes is that  they were formed when glaciers retreated  

at the end of the last glacial maximum  sometime around about 17 000 years ago,  

and they've been archiving the processes that  occur in this environment ever since then,  

in the sediments that accumulate at the bottom.  Okay so the ice retreats, there's an empty  

basin - fills with water and then sediments slowly  pile up at the bottom of those lakes for the last  

17 000. Yeah yeah, and what's really interesting  when the Alpine Fault ruptures in a large  

earthquake, we generate strong ground motions,  they disturb the sediments or the the natural  

pattern of accumulation in a way that allows us  to recognize that a shaking event has occurred.  

And they do that continuously. So unlike that  fault-bounded swamp that we were looking at  

at John O'Groats and Hokuri Creek, the sediment  archive or record that we're after, is some 80  

meters under the surface of this lake and the  way we actually sample it is using a Mackereth  

Corer which is a system that is lowered to the  bottom of the lake and we use compressed air to  

push a tube into the sediment that samples many  thousands of years of accumulation, and then the  

whole thing lifts off. You can see in this video  the Mackereth Corer dramatically breaching the  

surface of the lake as it comes to the surface  with our sediment archive or sediment record  

attached to it. Once we have that core we take it  back into the lab, split it and then study it to  

identify these earthquake layers. The neat thing  about lakes for reconstructing the spatial extent  

of earthquakes they almost act like nature's  seismographs - they're continuously recording  

when there's an earthquake they change the way  the sediments are being laid down, and we can  

identify that and then they continue recording and  one of the really powerful things about that is  

it means that they're able to record earthquakes  that occur very closely spaced in time.  

And that's important because it means we can  distinguish events that rupture one part of the  

fault from an earthquake that occurs in another  part of the fault, but really closely spaced  

in time. So that's useful for starting to piece  together this picture of rupture extent along the  

fault. Great! so just a question: you know, let's  say you found a layer in the bottom of the lake,  

how do you date it? How do you  know? So we use radio carbon dating  

of leaf material from the vegetation that  grows around the lake, and we can determine  

the timing of an earthquake within a decade or  so using these lake sediments. Can you tell me a  

little bit more about the different lakes? They're  distributed along the entire length of the fault  

and we've investigated four of them here. So  here's that site that we talked about right  

at the southern end of the fault and we've got  really nice earthquake records from Lake Ellery,  

Lake Puringa, Mapourika - you've  seen, and Lake Kaniere up by Hokitika.  

And this allows us to put a picture together  of earthquakes over the last 4 000 years  

along this 350 kilometer stretch of the Alpine  Fault. But one thing I just want to bring to your  

attention is that while the Alpine Fault looks  like this very straight linear structure, there  

are actually really important, albeit subtle,  changes in the fault that occur along its length.  

And those changes are predominantly related to  the strike of the fault - or the direction of it -  

and the dip of the fault - so the angle at which  it's diving away into earth's crust. For example  

on the South Westland section here, which is  a really important section with respect to  

some of the earthquake records we'll talk about,  the fault has a sub-vertical dip,so it's almost  

straight up and down. And when we transition  into the Central section which is also a really  

important section which is the yellow section,  here we have a much shallower dip and it's dipping  

away to the east underneath the Southern Alps.  And indeed that change in the strike of the fault  

and the dip is what distinguishes the South  Westland from the Central section and that will be  

important for the earthquake behavior that we're  about to talk about. Here we have a space and time  

diagram. So it's complicated, but I'll take you  through it. And essentially what it's showing us  

is: unlike the last diagram we just saw the timing  of earthquakes going back through the last 8000  

years, this is showing us the timing but also how  much of the fault broke in each earthquake. At  

the bottom of the screen here we have the Alpine  Fault laying on its side such that north is to the  

left and south is to the right, and then going  vertically on the screen here we've got time.  

So here's the year 2000 and we go back about  4000 years in this record here. Each one of these  

red marks here is an Alpine Fault earthquake  that has occurred at this particular site  

and what we can do is using statistics we  can correlate between sites, allowing us to  

map the spatial extent of rupture along this  fault. The important take-home message to  

see from this is that where we have these red  horizontal lines we have these large magnitude  

eight earthquakes that break the majority of the  south western - central section of the fault,  

in excess of 320 kilometers. But what's really  interesting is that there are other periods in  

the past such as this one here, demarcated by  these yellow earthquakes, where earthquakes  

rupturing the southwestern section or the  central section actually stop at this location,  

or the boundary between these two segments.  And they have a much smaller magnitude  

in the order of mid-sevens as opposed to in excess  of eight. So it's still a large earthquake but  

the ground motions associated with a magnitude  eight that breaks in excess of 300 kilometers of  

the fault is going to be significantly greater and  the spatial extent of those significantly larger  

than magnitude 7. So distinguishing between the  two is a really important thing to do. So based  

on this paleoseismic record alone we are able  to say that the central part of the Alpine Fault  

has a conditional probability- so a  forecast - of producing an earthquake in  

the next 50 years of about 75 percent. So that's  dramatically different from 29percent! - That is  

dramatically different from 29 percent. Okay so  i'll just get this right: you've discovered that  

the likelihood of an earthquake on the Alpine  Fault central section has been, if you like,  

increased with this new information to 75%  in the next 50 years. So it's pretty hig!  

Very high! On a global level, what does that  look like in terms of: is that big? high?  

I would say that that's very high! There are some  faults that have higher conditional probabilities,  

but not many. Okay yeah what else can you tell  me? Well what we can actually do is tell you  

what the magnitude or the amount of the fault  that's going to break in the next earthquake, and  

we do that by combining our empirical observations  of how the fault has behaved in the past  

with a computer model that's based on  the physics of how earthquakes behave.  

And by bringing those two sources of information  together we can actually forecast not just the  

likelihood that we'll see an earthquake, but what  its magnitude will be and what part of the fault  

will actually rupture. Okay and what we see when  we do that is there's about an 82 percent chance  

that the next earthquake is also going to be a  magnitude 8 through-going rupture that breaks  

both the central and the southwestern section.  Okay um it feels very sobering what you're saying,  

because what I'm hearing is that our understanding  of the Alpine Fault now, is telling us that the  

chances - all we can talk about is chances or  statistics or probabilities - the probabilities  

or chances of a future earthquake on the Alpine  Fault are higher that it will happen in the next  

50 years than we thought, and that the fact that  it's most likely going to be a magnitude 8 plus.  

I think you've summarized it perfectly. The one  thing that I will add is that I think knowledge  

is power in this instance Julian. At the end  of the day the fault for all intensive purposes  

is no more hazardous today than what it was  yesterday before we knew about this information.

But what we're able to do knowing that  there's a really high likelihood of  

a magnitude 8 earthquake on the Alpine  Fault is actually start to plan for it.  

And if we plan for it we can build resilience  into our communities and into our infrastructure.  

Does it change anything in terms of - I'm a person  living in a house somewhere on the South Island  

potentially quite near to the Alpine Fault, do I  do anything different? What do I do about this?  

I think as an individual it doesn't really change  that much at all. You know we've had a spate of  

large earthquakes in New Zealand over the last  decade starting with the Christchurch earthquake  

sequence and then with Kaikoūra. We all know  what we need to do as individuals with respect  

to having our earthquake response plan for the  household in place, making sure we have supplies  

in the house in the event that we are cut off from  our supply lines, and really I think that all this  

new information does is sharpen the point a little  bit. It makes us just a little bit more aware,  

a little bit more conscious that we need to have  those plans in place. And if we haven't done it  

maybe it provides us with the encouragement  we need to go out tomorrow or on the weekend  

and make sure that those supplies are stocked  up, those water bottles have been refreshed,  

and that you sit down and have a korero with  your family about what your response plan is  

going to be in the event that you experience a  large earthquake on the Alpine Fault or indeed  

on any one of the other 500 or more  active faults that we have in New Zealand.  

So Jamie I'd like to really thank you. This has  been an eye-opening discussion with you today.  

I think your work has been phenomenal - I've seen  you in action - we've been in the field together,  

I've seen you working on getting those cores.  It sounds easy it's incredibly technical,  

it's really problem solving at its best in  the most incredibly wild environments with  

all the weather and all of the wilderness  around you. So fantastic! Thanks so much  

for this opportunity to hear about your research  and so I hope people at home have enjoyed -or  

if that's the right word - have felt enlightened,  Something I'd really like to say too about it is:  

this is what science is. Science is value for  society. We now have some insight which guides  

us and gives us really good information.  So that's great thanks so much. Yeah thanks  

for coming along Julian. As always it's a  pleasure discussing science with you, and yeah  

I look forward to taking our  message out to the New Zealanders.