WEBVTT

00:00:03.000 --> 00:00:09.000
Good morning, everyone. Welcome back to the third day of the Northern California Earthquake Hazards Workshop.

00:00:09.000 --> 00:00:19.000
We have one more day of awesomeness for you. Can you stand it!! So, this is both the 20th anniversary of the 2004 magnitude 6.0 Parkfield earthquake.

00:00:19.000 --> 00:00:26.000
The 10th anniversary of the 2014 magnitude 6.0 South Napa earthquake. Were both of those earthquakes very impactful?

00:00:26.000 --> 00:00:30.000
Yes, they were. Do both of them deserve their own special session to look back at them?

00:00:30.000 --> 00:00:41.000
Yes, they do. Do we have time for that? No, we do not. And so when we look at these earthquakes, we thought the lesson from them is really the extraordinary hazards that after-slip posed for these earthquakes,

00:00:41.000 --> 00:00:53.000
and so what we'd like to do is instead look all around the world at all the different kinds of after-slip behavior we have seen and what that might mean for hazards that we might see in northern California.

00:00:53.000 --> 00:01:04.000
And to take us on this around the world trip, we have the amazing moderating team of Junle Jiang and Kathryn Materna.

00:01:04.000 --> 00:01:34.000
Hi everybody! Hello. To start off our session, we have our first speaker, Jan Primus, from Géoazur Laboratory in Nice, France, who will be talking about friction-driven seismic and postseismic slip of the 2014 South Napa earthquake.

00:01:35.000 --> 00:01:38.000
[technical problems]

00:01:38.000 --> 00:01:43.000
[technical problems]

00:01:43.000 --> 00:01:47.000
[technical problems]

00:01:47.000 --> 00:01:48.000
Thanks! Perfect.

00:01:48.000 --> 00:01:54.00
Hello, my name is Jan Premus and I will present my work on giant dynamic modeling of the 2014 South 

00:01:54.000 --> 00:02:03.000
Napa earthquake and it's afterslip.  I have done this research with my co-authors Frantisek Gallovic and Jean-Paul Ampuero.

00:02:03.000 --> 00:02:11.000
Napa earthquake was a magnitude 6.0 strike-slip earthquake, you can see the position of the hypocenter on the map on the right.

00:02:11.000 --> 00:02:19.000
It is denoted as a star. At the nucleation, the earthquake rupture proceeded towards the north.

00:02:19.000 --> 00:02:24.000
Earthquake duration was about 10 seconds.

00:02:24.000 --> 00:02:36.000
In the months after the earthquake it was followed by unusually light afterslip that was spreading in the opposite southward direction.

00:02:36.000 --> 00:02:47.000
We want to study the earthquake physics through a dynamic invention method. On one side we have the real event the Napa earthquake and data that it produced.

00:02:47.000 --> 00:02:58.000
We model it by a dynamic model parameterized by the choice of the friction law and an unfold distribution of stress and frictional parameters.

00:02:58.000 --> 00:03:07.000
From there we proceed with a rupture simulation that produces synthetic data that we then compare with the real data set.

00:03:07.000 --> 00:03:14.000
Next, we sample the model parameters by a Monte Carlo method.

00:03:14.000 --> 00:03:17.000
The dynamic inversion method in our case is formulated using a bayesian formulation as an update of 

00:03:17.000 --> 00:03:36.000
prior distribution of dynamic parameters. The prior distribution of model parameters are set as uniform on very wide intervals.

00:03:36.000 --> 00:03:52.000
We are seeking the posterior distribution of model parameters constrained by the data measured during the earthquake and we sample these using a Markov Chain Monte Carlo

00:03:52.000 --> 00:04:00.000
method. Completely the parallel tempering algorithm.

00:04:00.000 --> 00:04:13.000
For that we use seismic data from technical station. Some GPS stations and some additional service slip data.

00:04:13.000 --> 00:04:20.000
We expect the data to have Gaussian distribution 

00:04:20.000 --> 00:04:31.000
and we assign misfit to our synthetic data as an L2 norm of the difference between the synthetics and the real data.  

00:04:31.000 --> 00:04:45.000
We accept more permeably models that have a low misfit between the real and synthetic data.

00:04:45.000 --> 00:04:56.000
The inversion was done as a parallel run on free and Nvidia GPUs where it took around 35 seconds to run 1 forward simulation.

00:04:56.000 --> 00:05:13.000
During the dynamic inversion we simulated 500,000 models from which 7,500 were accepted as well fitting models.

00:05:13.000 --> 00:05:20.000
For the rupture propagation modeling, we use our fully dynamic finite difference called FD3D.

00:05:20.000 --> 00:05:32.000
While the postseismic phase was simulated quasidynamic. The final state of the coseismic model is used as an initial state of the postseismic model 

00:05:32.000 --> 00:05:37.000
while they share the frictional parameters.

00:05:37.000 --> 00:05:44.000
We choose our rate and state friction law with fast-velocity-weakening as the friction law in our modeling.

00:05:44.000 --> 00:06:09.000
The friction is dependent on rate as stored and the state variable. Well, the parameters that we invite are BAL and the beginning, B-A gives the frictional reality with friction being either weakening or strengthening based on the sign of the difference.

00:06:09.000 --> 00:06:18.000
Becoming those cities that's a slippery limit above each friction of style stripping quickly.

00:06:18.000 --> 00:06:28.000
Looking at the results, first I show the fit of the date. We used strong motions seismograms from 10 stations close to the fault.

00:06:28.000 --> 00:06:37.000
You can see their position on the map with respect to the fault plane. Graphs on the right show data is a black line.

00:06:37.000 --> 00:06:46.000
Best model seismograms [indiscernible] best estimates of the modeling uncertainty in blue.

00:06:46.000 --> 00:07:06.000
Number on the far right shows the maximum amplitude of the seismogram. Large portion of the unexplained data here is due to the model simplification as we used only a one dimensional velocity model to model the synthetic waveforms.

00:07:06.000 --> 00:07:15.000
We also use data from 7 close GPS stations. Map shows that position and coseismic displacement.

00:07:15.000 --> 00:07:26.000
Graph shows that postseismic displacement in 30 days after the earthquake.  Again, like here is the real data.

00:07:26.000 --> 00:07:39.000
Well, red is the best model and blue the channel density estimate. You can see that the postseismic signal from afterslip is not very large in comparison with the uncertainty of the data.

00:07:39.000 --> 00:07:45.000
As the earthquake has only a magnitude 6.

00:07:45.000 --> 00:07:56.000
Better source of the postseismic information where the alignment fields installed on top of the surface rupture there is less data points.

00:07:56.000 --> 00:08:02.000
around 5 data points in the month after the earthquake.

00:08:02.000 --> 00:08:09.000
But the postseismic signal is larger, 5 to 10 cm 

00:08:09.000 --> 00:08:17.000
and you can also see that it is very well fitted by our models.

00:08:17.000 --> 00:08:28.000
Here I show the on-fault distribution of coseismic slip, the south would be towards the right in this picture.

00:08:28.000 --> 00:08:42.000
The maximum of the coseismic slip is around 5 to 6 km depth, it achieves 3 to 3.5 m.

00:08:42.000 --> 00:08:58.000
The red contour shows the main contour of the coseismic slip with its uncertainty. The blue lines show the contours of the rupture front 

00:08:58.000 --> 00:09:04.000
every second of the rupture propagation.

00:09:04.000 --> 00:09:17.000
The grey dots show the position of aftershocks that are located mostly below the coseismic rupture.

00:09:17.000 --> 00:09:25.000
Earthquake is followed by postseismic slip that we show on the lower picture now.

00:09:25.000 --> 00:09:33.000
We have a large edge of shallow afterslip that's press towards out that can be shown 

00:09:33.000 --> 00:09:39.000
on the daily contours of afterslip shown as the blue lines.

00:09:39.000 --> 00:09:52.000
Additionally, our models produced patches of deep afterslip that are overlaid with the aftershocks.

00:09:52.000 --> 00:10:07.000
Vital both, the coseismic and postseismic slip together we see that afterslip occurs on the border of coseismic slip and then spreads out.

00:10:07.000 --> 00:10:23.000
The dynamic produced distributions of all model parameters. Here I will mostly focus on the B-A that decides the strengthening beginning fictional rheology.

00:10:23.000 --> 00:10:34.000
Here on the right, I show the distribution of B-A. Blue parts, are strengthening and red parts are weakening.

00:10:34.000 --> 00:10:43.000
The contours again show the contours of the coseismic slip and the postseismic slip.

00:10:43.000 --> 00:10:55.000
The near surface layer that produced the shallow afterslip is predominantly strengthening while the main coseismic area around 6 km depth

00:10:55.000 --> 00:10:59.000
Is becoming as people expected.

00:10:59.000 --> 00:11:09.000
Below the coseismic area is another strengthening patch that produced the deep afterslip.

00:11:09.000 --> 00:11:17.000
If interested in the deeper afterslip and we think that's the case for this deeper afterslip is

00:11:17.000 --> 00:11:29.000
supported by the occurrence of around 200 off-fault aftershocks around this area that I show here as a green rectangle.

00:11:29.000 --> 00:11:38.000
Now it's a theoretical relationship between elevated stress, right? Sigma mode and aftershock grade and they are supposed to be proportional.

00:11:38.000 --> 00:11:46.000
Here I show the stress rate in the deep afterslip badge from other simulation as a blue line.

00:11:46.000 --> 00:11:53.000
The error bars show again the variability between the different models.

00:11:53.000 --> 00:12:00.000
I show the relationship with the real aftershock rate shown as black dots.

00:12:00.000 --> 00:12:24.000
This gives us an independent validation of our results. We could also maybe in the future use this afterslip rate as an additional data point for future dynamic inversions constraining the stress rate in different areas of the fault.

00:12:24.000 --> 00:12:37.000
The shallow part of the fault is also interesting. It is essentially divided into two halves; northern, left; half fractures coseismically and southern half fractures of seismic.

00:12:37.000 --> 00:12:46.000
From local rheology, the southern half is embedded in Quaternary sediments while northern in harder Cretaceous rocks.

00:12:46.000 --> 00:13:01.000
This is looking at the horizontal slice of P-A. On the left plot we see values of -0.01 in the southern aseismic part
 
00:13:01.000 --> 00:13:11.000
And by a small value in the northern seismic part, while the transition between them approaches 0.

00:13:11.000 --> 00:13:21.000
In conclusion, we showed a joint dynamic model of seismic and aseismic slip of the 2014 Napa earthquake.

00:13:21.000 --> 00:13:31.000
The constraint is dynamic parameters from seismic and GPS data. You can see many additional details in our paper.

00:13:31.000 --> 00:13:44.000
The dynamic inversion is a very difficult nonlinear problem. In the case of the Napa earthquake it's solution was made possible by two components.

00:13:44.000 --> 00:13:54.000
First, we had available the live update dataset of both seismic and GPS data.

00:13:54.000 --> 00:14:00.000
To constrain the details of the earthquake and postseismic slip.

00:14:00.000 --> 00:14:06.000
Second component was a very fast forward solver.

00:14:06.000 --> 00:14:13.000
GPU acceleration allowed us to make one structure simulation in less than 1 minute.

00:14:13.000 --> 00:14:23.000
And those we got simulate our 500,000 bottles needed for the dynamic invasion.

00:14:23.000 --> 00:14:41.000
That's it from me. Thank you. Cheers.

00:14:41.000 --> 00:14:52.000
Alright, cool. Thank you. And for anybody who has questions, please put them in the chat and if Jian doesn't get to them will be monitoring and we can bring them back up during the discussion.

00:14:52.000 --> 00:14:57.000
Thanks for a really nice talk.

00:14:57.000 --> 00:15:00.000
I'll just go. Hey, go for it.

00:15:00.000 --> 00:15:09.000
Shall we move on to the next. Okay, so our next speaker is Nicola D'Agostino.

00:15:09.000 --> 00:15:17.000
He's going to talk about the afterslip for the 2009 L'Aquila earthquake.

00:15:17.000 --> 00:15:27.000
Good morning, my name is Nicolas D'Agostino and I am a researcher at the National Institute of Geophysics in Rome and today I will talk about the afterslip 

00:15:27.000 --> 00:15:33.000
after the 2009 Mw 6.1 L'Aquila earthquake.

00:15:33.000 --> 00:15:46.000
Depending on the availability of GPS and in some measurements coseismic and postseismic slip inversions are frequently biased by poor time resolution.

00:15:46.000 --> 00:16:04.000
Entire information of the surface rupture is also poorly determined by temporal point of view, and the availability of generic instrumentation in place in the near-field is thus critical for resolving 

00:16:04.000 --> 00:16:21.000
coseismic and postseismic deformations. Afterslip on a normal fault also has been rarely reported, implying that the relationship between coseismic and postseismic slip in extension setting is not very well known.

00:16:21.000 --> 00:16:32.000
In the next slides, I will show you that the 2009 L'Aquila earthquake provides a unique example of afterslip on a normal fault.

00:16:32.000 --> 00:16:40.000
Where the existing data allows the separation of coseismic early-afterslip and late-afterslip.

00:16:40.000 --> 00:16:49.000
In this talk, I will start with the description of the regional contest of the L'Aquila earthquake.

00:16:49.000 --> 00:17:00.000
I will then describe the coseismic and late-afterslip distribution inferred from GPS and InSAR inversion.

00:17:00.000 --> 00:17:09.000
Will then describe the early-afterslip inferred from seismic, laser strain meter and GPS data.

00:17:09.000 --> 00:17:23.000
And I will then conclude by comparing geodetic afterslip with temporal evolution of surface ruptures and other seismological findings from the Paganica fault.

00:17:23.000 --> 00:17:34.000
The L'Aquila earthquake occurred in the Apennines, which are dominated by crustal extension on the right; on the left we see a GPS velocity field 

00:17:34.000 --> 00:17:43.000
with increasing vectors from the terraniun side to Adriatic side with an overall extension rate of 3 mm per year.

00:17:43.000 --> 00:17:54.000
From this GPS velocity field it is possible to calculate the strain rate and we can see that the L'Aquila is located exactly on the actively straining belt.

00:17:54.000 --> 00:17:56.000
After the 2009 L'Aquila earthquake the area has been affected by two more

00:17:56.000 --> 00:18:15.000
earthquakes of magnitudes larger than 6.0, the Amatrice and the Norcia earthquakes. These two events showed evidence of postseismic deformation, but they lack geodetic data in the near-field 

00:18:15.000 --> 00:18:34.000
readily available after the events. So there afterslip is not very well resolved and also these earthquakes occurred on rugged terrain, so surface structures were more difficulty accessible.

00:18:34.000 --> 00:18:48.000
For this reason, the diversity and availability of near-field measurements for the L'Aquila earthquake make this event one of the best normal faulting earthquake for both coseismic and postseismic deformation.

00:18:48.000 --> 00:18:53.000
The L'Aquila earthquake was preceded by 6 months-long swarms culminating with the magnitude 4 on March 30th,

00:18:53.000 --> 00:19:12.000
and the mainshock was given a magnitude 5.9 with a moment magnitude 6.1 and occurred on a normal fault southwest-dipping that was soon identified to coincide with the Paganica fault here.

00:19:12.000 --> 00:19:38.000
The strong motion and highly rate GPS data provided resolution on the temporal evolution of the rap coseismic rupture showing an initial upward-pulse right below the city of L'Aquila, which caused extensive damage in the city,

00:19:38.000 --> 00:19:50.000
and this town was already known for its seismic hazard and the previous earthquake were in 1461 and the 1703 with a double earthquake 

00:19:50.000 --> 00:20:16.000
in 1701. This is the geometry of the Paganica fault showing the hypocenter, the slip distribution, and the emergence of the fault near the Paganica village.  In this area, the surface ruptures were soon identified after the earthquake, 

00:20:16.000 --> 00:20:30.000
and they were located just at the outskirts of the village, and so they were repeatedly observed and it was clear that the offset and the deformation across these ruptures was increasing 

00:20:30.000 --> 00:20:40.000
in the days following the event, and some of the rupture affected the buildings, so different agencies started taking measurements.

00:20:40.000 --> 00:20:44.000
Documenting the evolution of these ruptures.

00:20:44.000 --> 00:20:53.000
The dataset that can be used for postseismic deformation include continuous GPS station 

00:20:53.000 --> 00:21:17.000
some of them were already existing and some were installed right after the earthquake, and there is also a very dense temporary dense acquisition from InSAR from the COSMO-Skymed constellation with 33 acquisitions, 1 pre-seismic which lasted for about 3-6 months.

00:21:17.000 --> 00:21:35.000
With the first postseismic acquisition taken 6 days after the earthquake. Early afterslip in the first 6 days has also been measured by laser strainmeter, 20 km from the fault, and from field measurements near the village of Pagan.

00:21:35.000 --> 00:21:54.000
This is a map of the coseismic and postseismic displacement captured by the coseismic COSMO-Skymed sudden light, which shows a compact patch of deformation for the coseismic with maximum 

00:21:54.000 --> 00:22:07.000
line of side displacement of about 15 km and a more articulate pattern for the postseismic slip with 3 patches of slip surrounding the main 

00:22:07.000 --> 00:22:31.000
coseismic slip, and the difference from the coseismic and postseismic is also visible on this two section which shows a progressive motion of the L'Aquila highest deformation towards the faults, which suggests a shallowing of slip 

00:22:31.000 --> 00:22:41.000
along the Paganica fault. This image here shows the continuing evolution of the afterslip 

00:22:41.000 --> 00:22:55.000
on the Paganica fault. This is the pattern of LOS displacement at different temporal frames which shows the characteristic asigmatic pattern and continuous gradient of LOS.

00:22:55.000 --> 00:23:04.000
This measurements suggest that the postseismic deformation can be explained by slip

00:23:04.000 --> 00:23:12.000
at shallow depth along the Paganica fault without full propagation at the surface. This is the results from the coseismic and postseismic 

00:23:12.000 --> 00:23:23.000
conversion. The same fault geometry was used for both coseismic 

00:23:23.000 --> 00:23:37.000
and postseismic is the distribution is rather compact with the main patch, which basically stopped a few kilometers below the surface.

00:23:37.000 --> 00:23:51.000
And the afterslip is basically surround this high coseismic patches with three main patches which were characterized by the characteristic temporal decay.

00:23:51.000 --> 00:23:57.000
And they were characterized by a stationary evolution so they didn't, migrate 

00:23:57.000 --> 00:24:07.000
in time. The early afterslip was captured by the lesser stream meter located in the Gran Sasso tunnel 20 km

00:24:07.000 --> 00:24:19.000
northeast of L'Aquila. These strain meters were blind to coseismic deformation due to low-pass filtering in the electronics, but they're very sensitive to postseismic deformation.

00:24:19.000 --> 00:24:31.000
Strain has been measurements along two alignments parallel perpendicular to defaults and two phases of the formation has been 

00:24:31.000 --> 00:24:49.000
recorded. A first phase non monotonic pattern with initial shortening followed by elongation along the alignments, BA followed by a monotonic pattern for both BA and BC alignments.

00:24:49.000 --> 00:24:58.000
This deformation has been modeled and interpret in terms of the first initial phase of slow diffusive 

00:24:58.000 --> 00:25:09.000
slip propagation from the area of highest coseismic slip toward the surface and toward the regions where surface ruptures were observed.

00:25:09.000 --> 00:25:17.000
And after this initial phase the afterslip was stationary in this area where it was

00:25:17.000 --> 00:25:37.000
corresponding which paths of inferred from InSAR from GPS. and The early phase of afterslip as well also been studied integrating seismic and postseismic data for the first, day and half,

00:25:37.000 --> 00:25:52.000
and in this study, we can see the coseismic distribution, shown with pink contours, surrounded by afterslip representative of the first half-day and half of the formation.

00:25:52.000 --> 00:26:22.000
And you can see that this study confirmed the inference from the LOS stream meter of our propagating slip from the region of Icoseismic slip to the surface right below the regions of surface ruptures.

00:26:22.000 --> 00:26:26.000
This study also refines gravity of contribution of afterslip and coseismic slip.

00:26:26.000 --> 00:26:41.000
The surface fractures observing Paganica were measured like this one; this was a crack where, track meters were installed near the surface ruptures and it was not clear whether this surface evolved 

00:26:41.000 --> 00:26:55.000
for local factors, local ability or where somehow related to the deeper afterslip. This evolution here, which shows a characteristic exponential decay, shows the same pattern of the formation,

00:26:55.000 --> 00:27:09.000
reconstructed from the inversion of GPS data suggesting that these cracks 

00:27:09.000 --> 00:27:24.000
evolved under the effect of afterslip on the underlying Paganica fault. Soon after the event different groups started  

00:27:24.000 --> 00:27:48.000
trenching across the Paganica fault to see the previous events along this fault, and it's interesting to compare the results of the 2009 afterslip studies with the findings from the trenches and in particular these studies reconstructed a long-term slip rate of about 0.2-0.5 mm/year.

00:27:48.000 --> 00:27:56.000
But, what emerged from these studies is an heterogeneous behavior with the small ruptures 

00:27:56.000 --> 00:28:09.000
associated with events like the one in 2009 or the 1461 event with small ruptures and recurrence time of about 500 km.

00:28:09.000 --> 00:28:21.000
At the same time, there was evidence for larger offset, larger than 0.5 m, that they are probably a characteristic of larger [indiscernible] segment events 

00:28:21.000 --> 00:28:32.000
with much longer, recurrence times. So, taking into account the result of 2009 afterslip study,

00:28:32.000 --> 00:28:56.000
it's clear that the same portion of the fault can undergo different behavior. So we can have the seismic slip or higher energy events rupturing to the surface and a seismic slip during the postseismic phase for lower energy event, this type of behavior has been already described 

00:28:56.000 --> 00:29:06.000
as aware in different places and is known as "conditionally stable" in the framework of R&S frictional 

00:29:06.000 --> 00:29:16.000
theory. So going to the conclusion. Our study and all the studies allowed to reconstruct two afterslip 

00:29:16.000 --> 00:29:25.000
phases. Initial phase of afterslip with propagation of slip from the region of high coseismic slip toward the surface.

00:29:25.000 --> 00:29:36.000
This initial phase was followed by a stationary afterslip, which encircled at the edges of the area of high coseismic slip.

00:29:36.000 --> 00:29:46.000
The timing of surface ruptures is not tightly constrained, but the proximity to the shallow afterslip is consistent with their postseismic formation.

00:29:46.000 --> 00:30:04.000
Reconstructing the heterogeneous behavior of the Paganica fault, which shows aseismic afterslip, which is probably characteristic for low energy ruptures like the 2009.

00:30:04.000 --> 00:30:16.000
And it's probably aseismic in his shallower part higher energy ruptures like the 1703-like event.

00:30:16.000 --> 00:30:28.000
Thank you very much and bye, bye.

00:30:28.000 --> 00:30:36.000
Alright, well that was a super interesting talk. I'm keeping up with a lot of questions and Nicola is too in the chat.

00:30:36.000 --> 00:30:56.000
So our next speaker is Ziyadin Zakir here from Istanbul Technical University and he will speak about "surface creep along the East Anatolian fault and the magnitude 6.8, February, 24, 2020 Elazig earthquake."

















00:30:56.000 --> 00:31:07.000
Hello everybody in this talk I'm going to present with my colleagues from {indiscernible] Observatory Istanbul Technical University 

00:31:07.000 --> 00:31:19.000
on the creeping section of the East Anatolian fault and the 2020 Elazig earthquake that took place on the central section of the East Anatolian fault.

00:31:19.000 --> 00:31:28.000
It's an important earthquake because this is the first large earthquake magnitude 6.8 to take place on a 

00:31:28.000 --> 00:31:46.000
well monitored strike-slip fault that shows surface creep. Therefore it provides important observations and insights into the behavior of creeping faults during the earthquake cycle that may be useful for those who do 

00:31:46.000 --> 00:31:57.000
mechanical and dynamic earthquake modeling and studies.

00:31:57.000 --> 00:32:12.000
As you all may know the east left-lateral Anatolian fault conjugates in the North Anatolian fault help totally escape from the collisions on

00:32:12.000 --> 00:32:31.000
the Arabian plate and the Eurasian plate in eastern Turkey where they meet at the triple junction called the Karliova triple junction. The 2020 Elazig earthquake took place in the central section of the Anatolian fault

00:32:31.000 --> 00:32:50.000
and 3 years later the [indiscernible] earthquake happened on the western continuation of the Elazig. Let me zoom in here in this area and show you the GPS velocity fields 

00:32:50.000 --> 00:33:03.000
with respect to [indiscernible] Anatolia that shows very nicely and clearly the strain accommodation 

00:33:03.000 --> 00:33:15.000
along the North Anatolian fault here and the [indiscernible] fault here. Here are the subsurface of the 2020 and 2023 

00:33:15.000 --> 00:33:25.000
earthquakes, first earthquake here that started here and then propagated onto the main fault and then rupturing bilaterally 

00:33:25.000 --> 00:33:39.000
and breaking about 350 km long section of [indiscernible] about 9 hours later, the second shock magnitude 8.6, 7.6.

00:33:39.000 --> 00:34:04.000
happened on a subsidiary fault display for [indiscernible]. Magnitude of the first shock was M7.8.  Let me zoom in this area and show you InSAR velocity obtained from SentinelT21 data this is descending [indiscernible]

00:34:04.000 --> 00:34:12.000
track where blue areas indicating motion away from satellite that is north to west,

00:34:12.000 --> 00:34:21.000
and red areas are stable on most of the satellite indicating left-lateral plate motion [indiscernible].

00:34:21.000 --> 00:34:34.000
Extended data, process data further north than the North Anatolian fault also comes into the picture showing nicely

00:34:34.000 --> 00:34:53.000
nicely all the three plates are in motion. If we take a profile across these two earthquakes that happened last year we get this typical interseismic accumulation shape of [indiscernible]

00:34:53.000 --> 00:35:13.000
indicating that the faults are here locked.  When we move to the east and section of the Anatolian fault and checking and analyzing the descending and ascending [indiscernible]

00:35:13.000 --> 00:35:34.000
and SentinelT21 data we see a very sharp contrast in the velocity field across the fault here along the central section [indiscernible] of the North Anatolian fault, which indicates of course, surface group 

00:35:34.000 --> 00:35:43.000
contribute to the other sections along the North Anatolian western sections. You see, this is better seen in profiles also 

00:35:43.000 --> 00:36:07.000
let's look at the profiles from ascending and descending data sets. Right, the default you see [indiscernible] velocity field  [indiscernible] Let me measure the offset, which is the creep rate.

00:36:07.000 --> 00:36:20.000
All along the fault can be obtain a variation of correct rate along the fault that varies from 0 this means from the [indiscernible]

00:36:20.000 --> 00:36:31.000
to, up to 10 mm per year in more, which is roughly the slip rate of the East Anatolian fault. 

00:36:31.000 --> 00:36:39.000
All different data sets in different tracks and GPS are all in [indiscernible].

00:36:39.000 --> 00:36:52.000
When the model [indiscernible] of these profiles we also obtain creeping depths.

00:36:52.000 --> 00:37:14.000
However, these creeping depths estimates are not very reliable and the background we have the seismicity along the fault and the green circles are repeaters just around the hypocenter of the 2020 

00:37:14.000 --> 00:37:17.000
[indiscernible] earthquake.

00:37:17.000 --> 00:37:39.000
Look at the geology along the section of the fault we see [indiscernible] rocks and metamorphic rocks that all have [indiscernible] promote the creep observed in this area.

00:37:39.000 --> 00:37:47.000
I mean, back to the velocity field this time, you have Sentinel on track 

00:37:47.000 --> 00:37:59.000
3 and the GPS velocity respect to the [indiscernible] Arabia. If I show you another profile 

00:37:59.000 --> 00:38:22.000
along the rapture here of January, 2420, 2020. Again, you can see nicely the offset fault indicating surface creep, all the different data sets and the compost for all [indiscernible] etc., all agree with each other.

00:38:22.000 --> 00:38:34.000
This part of the fault is [indiscernible]. However, when the earthquake happened our colleagues went back to the field and observed 

00:38:34.000 --> 00:38:49.000
not clear surface rupture except some echelon fractures, etc. The absence of such surface loss indicated by postseismic interferograms like these ones, Sentinel 

00:38:49.000 --> 00:38:57.000
and also by GPS, here you see G4 just next to the fault has much 

00:38:57.000 --> 00:39:14.000
less coseismic displacement compared to those away from the fault G2 and G3 indicating that indeed the surface structure [indiscernible] did not rupture to the surface.

00:39:14.000 --> 00:39:21.000
The fault at the surface following the earthquake, as you can see in this

00:39:21.000 --> 00:39:33.000
postseismic [indiscernible] 1.5 years and also we have a creep method here by Roger Bilham also indicated the afterslip. So let me show you a profile along this line here,

00:39:33.000 --> 00:39:46.000
a prime. This is the same profile that I've shown you, and here the zoom, new field area plus

00:39:46.000 --> 00:40:02.000
-5 km and the earthquake happened you see, now this surface because just like interseismic slip accumulation

00:40:02.000 --> 00:40:19.000
we should see that there is a rupture [indiscernible] at the surface graphic-like this but we don't... and that means slip happened at depth but did not reach the surface.

00:40:19.000 --> 00:40:35.000
But analyzing the postseismic slip data indicates that yes, the fault slip reached the surface later on after the earthquake [indiscernible].

00:40:35.000 --> 00:40:46.000
When we model the postseismic slip distribution we get the high [indiscernible] fault geometrics of depth just above the epicenter

00:40:46.000 --> 00:40:58.000
which rapidly diminishes to 0 near 0 to the surface. Just because the surface had been creeping already so the rupture basically was injured or stopped 

00:40:58.000 --> 00:41:28.000
by this surface creep. After the earthquake postseismic slip, afterslip took place mostly at shallow depths around [indiscernible].  Creep also appears to be deep in this section [indiscernible].

00:41:30.000 --> 00:41:34.000
I'm looking at the 

00:41:34.000 --> 00:41:51.000
gap between the 2020 and 2023 earthquakes which is above the [indiscernible] we see again the sharp contrast across the builds of

00:41:51.000 --> 00:42:03.000  
6-months postseismic data updated from Sentinel data we see afterslip.

00:42:03.000 --> 00:42:31.000
So the area also the same, we have metamorphic rocks, mainly gnays and schist. So in conclusion, the January 24th magnitude 6.8 2020 Elazig earthquake has shown that the earthquake ruptures can be stopped or injured.

00:42:31.000 --> 00:42:45.000
Large earthquakes ruptures can be stopped by shallow earthquakes. In a similar way, rupture of the magnitude 7.8.

00:42:45.000 --> 00:42:51.000
Kahramanmaras earthquakes that happened last year.  Also, I'll stop my [indiscernible], and this surface rupture only started to unfold is mostly shallow.

00:42:51.000 --> 00:43:12.000
But locally it reaches down to seismogenic depths and the presence of your lights, metamorphic rocks and mafic rocks is suggested.

00:43:12.000 --> 00:43:27.000
Creep is likely promoted by weak minerals within these rocks. Thank you.























00:43:27.000 --> 00:43:36.000
Alright, thank you for very interesting talk. So we want to move on to the next speaker. We have Celeste Hofstetter, a speaker from UC Riverside.

00:43:36.000 --> 00:43:45.000
The title is "Did the creep stop the 2023 magnitude 7.8 Puturge earthquake rupture?"

00:43:45.000 --> 00:43:50.000
Hi, first I would like to thank the USGS and my co-authors for the opportunity to speak today.

00:43:50.000 --> 00:44:01.000
Thank you. And today we'll be speaking on the 2023 Turkey earthquakes and specifically what may have caused the [indiscernible] rupture to terminate to the northeast.

00:44:01.000 --> 00:44:07.000
So a quick look at what I will be focusing on today. I will be talking about the background of Turkey's plate mission.

00:44:07.000 --> 00:44:20.000
The 2023 Turkey earthquake sequence. What is fault creep? A closer look at the Puturge Segment of the East Anatolian fault and applying lessons learned about afterslip to northern California.

00:44:20.000 --> 00:44:36.000
So first we'll take a step back and look at the large scale plate movement in this area. Here's a map that shows African and Arabian plates are moving north and northwest to respectively towards the Anatolian plate which moves in a southwest counterclockwise motion and an example of escape tectonics, 

00:44:36.000 --> 00:44:42.000
and the North Anatolian fault is shown a yellow and the eastern Anatolian fault shown in pink.

00:44:42.000 --> 00:44:52.000
Both compensate for this plate mission dynamics with the nxorth Anatolian fault being a right-lateral strike-slip fault and the East Anatolian fault being a left-lateral strike-slip fault.

00:44:52.000 --> 00:45:00.000
So next I will focus on the 700 km East Anatolian fault, which has a slip rate of about 10mm/year 

00:45:00.000 --> 00:45:06.000
and we'll focus on the section in the red box.

00:45:06.000 --> 00:45:13.000
So, February 6, 2023, the Turkey earthquake sequence occurred and it includes two earthquakes that happened.

00:45:13.000 --> 00:45:20.000
The first was a magnitude 7.8 Pazarcik earthquake that struck on the Narli fault and propagated to the East Anatolian fault.

00:45:20.000 --> 00:45:25.000
The second occurred about 9 hours later, ir's a magnitude 7.5 Elibistan earthquake 

00:45:25.000 --> 00:45:32.000
it struck on the Cardak fault, and so here we are seeing it and this is the dense offset maps for built these figures,

00:45:32.000 --> 00:45:40.000
nd, so for this purpose, the satellite is descending flying south and it is looking west.

00:45:40.000 --> 00:45:56.000
And red is moving away from the satellite. Blue is moving towards the satellite, and we can see that the range is taking measurements that are much more clear because it is perpendicular to the flight path, whereas with a, whereas with azimuth it is in the direction of flight path.

00:45:56.000 --> 00:46:02.000
And so for this talk I will focus on the main magnitude 7.8 Pazarcik earthquake.

00:46:02.000 --> 00:46:09.000
And I'm going to be focusing specifically on the yellow box, which is where the earthquake terminates.

00:46:09.000 --> 00:46:16.000
And also includes the Puturge segment of the East Anatolian fault, which previous research and colleagues have

00:46:16.000 --> 00:46:25.000
found to be creeping. That I understand the seismic systems,  I am studying a seismic or fault 

00:46:25.000 --> 00:46:30.000
creep. Fault creep is a measurable surface displacement along the fault when there is no earthquake occurring 

00:46:30.000 --> 00:46:39.000
and creep segments display different frictional properties and lock segments such as the ability to slip during afterslip and inner seismically, but not during an earthquake.

00:46:39.000 --> 00:46:47.000
So for this talk we will assume shallow afterslip is the same as creep and so here are some examples of fault creep.

00:46:47.000 --> 00:46:52.000
on the North Anatolian fault. A is a railroad station at Ismetpasa,

00:46:52.000 --> 00:47:03.000
And you can see the bricks are slightly offset to the right in the red box. And B, you can see a brick wall with diagonal cracks climbing to the left at Himalaya Village near Ismetpasa suggesting the crack is from shear force.

00:47:03.000 --> 00:47:20.000
So how do we measure the surface displacement? So I am using InSAR, Interferometric Synthetic Aperture Radar data from Satellite Sentinel-1, which was launched in 2014 and currently orbits in 12-day cycles, 

00:47:20.000 --> 00:47:27.000
and from the data we are able to generate a product called an Interferogram, which allows for high resolution measurements of ground deformation.

00:47:27.000 --> 00:47:37.000
And we can also measure creep with campaign and GNSS data, which colleagues in Turkey have collected and to which we will compare our results.

00:47:37.000 --> 00:47:44.000
Here is a cartoon diagram that shows how InSAR works. The satellite emits a radar pulse and records the return signal phase.

00:47:44.000 --> 00:47:51.000
And on the second pass the satellite will record a second measurement. The two images are then different to generate than or interferogram.

00:47:51.000 --> 00:47:54.000
Any changes in phase from surface displacement such as the red line shown in the yellow circle can then be measured 

00:47:54.000 --> 00:48:08.000
and a few key questions we are investigating with this technology are "what is a creep behavior across three time periods on the Puturge segment of the eastern Anatolian fault?

00:48:08.000 --> 00:48:15.000
How does the fault creep behavior vary spatially and temporarily and to creep impact with the Pazarcik rupture terminated.

00:48:15.000 --> 00:48:23.000
So first we will look at is there fault creep in this area. Short answer, yes, there is creep in this area.

00:48:23.000 --> 00:48:31.000
Colleagues in Turkey have been doing exciting work collecting data on this section of the East Anatolian fault around Hazar Lake since 2014.

00:48:31.000 --> 00:48:43.000
And here you can see their campaign and permanent stations as the inverted yellow triangle and a creep profile here traced by the green line 

00:48:43.000 --> 00:48:52.000
displays an arc tangent pattern. While a lot to profile that moves in an earthquake event would display a step pattern,

00:48:52.000 --> 00:48:58.000
which is more traced by this red line. Where it goes across, it goes up and then it goes across again.

00:48:58.000 --> 00:49:06.000
So they're parfiles show that they have found creep with rates varying between 5.5mm to 6.8mm per year,

00:49:06.000 --> 00:49:12.000
and so now the next step was to process our InSAR data.

00:49:12.000 --> 00:49:22.000
So to generate all of our velocity and cumulative displacement maps, we used over 1,700 area standard product interferograms and processed them using Mintpy software.

00:49:22.000 --> 00:49:34.000
Atmosphere corrections were applied using ERA-5 data, and the velocity map contains data from 2014 to 2020 and includes the GNSS campaign data from our colleagues in Turkey.

00:49:34.000 --> 00:49:41.000
The stations are again the inverted triangles, and their velocity was converted to our line of sight velocity and shows a good data fit.

00:49:41.000 --> 00:49:50.000
In this figure, the satellite is traveling northward and looking east. With the red showing displacement towards a satellite and blue showing displacement away from the satellite.

00:49:50.000 --> 00:49:58.000
The black line traces the East Anatolian fault and the blue line to the southwest traces the 2023 Pazarcik rupture.

00:49:58.000 --> 00:50:08.000
Creep can be seen in the areas along the fault where there is a sharp color change from blue to red at the fault line with a darker color suggesting faster line of sight velocity.

00:50:08.000 --> 00:50:16.000
So we see different fault group behavior in these three fault sections outlined by the boxes. There is very little to no creep to the southwest in the yellow box.

00:50:16.000 --> 00:50:28.000
In the blue box slightly to the northeast, there is a larger area of slower creep, shown by the paler color, and the pink box closer to Hazar Lake shows darker colors suggesting faster creep.

00:50:28.000 --> 00:50:34.000
The slow creep area to the southwest starts within a approximately 10 km over the Pazarcik rupture terminated.

00:50:34.000 --> 00:50:51.000
So the next step is to generate profiles and measure the creep rates. So here the creep rates vary from 0 mm/year to the southwest to 0.05 mm to 1.5 mm/year in the center area and up to 4 mm to 6 mm/year to the northeast.

00:50:51.000 --> 00:51:01.000
These creep rates correlate well with our colleagues at Turkey and give us confidence in the data. So this works the end of our first period due to an earthquake.

00:51:01.000 --> 00:51:13.000
And so in January, 24th, 2020, the magnitude 6.8 Sivrice earthquake struck along the partook segment of the East Anatolian fault, southwest of Hazar Lake and northeast of the Pazarcik rupture.

00:51:13.000 --> 00:51:21.000
So how did this earthquake change the fault creek behavior? So in this cumulative displacement map from 2020 to 2023 after this Sivrice earthquake 

00:51:21.000 --> 00:51:32.000
the amount of displacement has increased. The yellow box traces the extent of the earthquake rupture with the ends having higher displacement than the rupture area.

00:51:32.000 --> 00:51:39.000
And surface displacement now extends to the southwest within about 5 km of the Pazarcik rupture.

00:51:39.000 --> 00:51:53.000
So what pattern of afterslip is in the time series. So to see this we difference the time series by looking at 4 km long profiles along the East Anatolian fault in 5 km increments and took the time series at the end of each.

00:51:53.000 --> 00:51:58.000
So here's an example where the red graph shows the time series displacement for the north end of the profile.

00:51:58.000 --> 00:52:07.000
The black graph shows the time series displacement for the south end of the profile, and we difference them to minimize atmospheric and seasonal noise.

00:52:07.000 --> 00:52:16.000
And you can see in the blue graph a time dependent afterslip pattern traced by the orange line. So what do we see when we look at the highest displacement area to the southwest?

00:52:16.000 --> 00:52:22.000
So we see an interesting step in the data from a sharp increase in the cumulative displacement that looks like a creep event.

00:52:22.000 --> 00:52:26.000
It occurs between July 31st and August 18th, 2020 displaces about 8 cm over an 18-day period and is seen over 15 km along the fault.

00:52:26.000 --> 00:52:41.000
There also appears to be minimal afterslip before and after the event. And so next we isolate the creep event in the cumulative displacement data.

00:52:41.000 --> 00:52:54.000
We can see the event here in the insert data. We average the time series at the beginning and end of the event and difference to two to isolate the event displacement and the displacement is similar to a small earthquake.

00:52:54.000 --> 00:53:02.000
So to better understand this event, we generated interferograms from ascending and descending data and applied in elastic dislocation model.

00:53:02.000 --> 00:53:10.000
The left column shows the two interferograms, which are the measurements for surface displacement from July 20th to August 19, 2020.

00:53:10.000 --> 00:53:17.000
The middle column shows the model for both tracks. These models are the best fit for the mean fringe features seen in the data.

00:53:17.000 --> 00:53:25.000
And the right column shows the residual displacement, subtracting the model from the data and shows what part of the data the model does not account for.

00:53:25.000 --> 00:53:34.000
So from this elastic dislocation model, we can infer the creep event was a magnitude 5.77, had 34 cm of slip.

00:53:34.000 --> 00:53:40.000
At a depth of 0.7 to 7.1 km and a length of 7.7 km.

00:53:40.000 --> 00:53:47.000
So for the next step we difference the time series for the length of the fault segment and examine the profiles.

00:53:47.000 --> 00:53:59.000
So here's a cumulative displacement map after this Sivrice earthquake and it shows an aftership displacement pattern that decays with time occurred to the displacement pattern that decays with time occurred to the northeast and gradually decreases as we move to the northeast and gradually decreases as we move to the southwest.

00:53:59.000 --> 00:54:08.000
The creep event starts near the southwest end of the 2020 rupture extent and there continues to be minimal creep within 5 km of the Pazarcik rupture.

00:54:08.000 --> 00:54:16.000
So the profile showed us placement increased after the Sivrice earthquake and the high displacement to the southwest included a large creep event.

00:54:16.000 --> 00:54:32.000
So how did the fault creep behavior change after the Pazarcik rupture? So here we can see a cumulative displacement map for 2023 for the greatest amount of slip was on the southwest segment of the fault which gradually decreases to minimal or no discernible creep by his our lake to the northeast.

00:54:32.000 --> 00:54:44.000
And when the cumulative displacement profiles are plotted after differences in the time series. We can see that there has been no large creep event and the profile show time dependent decaying after select patterns.

00:54:44.000 --> 00:54:50.000
But the largest cumulative displacement to the southwest and the smallest to the northeast, 

00:54:50.000 --> 00:54:54.000
and so overall, the creep behavior changed a long fault in the Puturge segment of the east Anatolian fault across three periods from 2014 to 2020.

00:54:54.000 --> 00:55:05.000
There was moderate creep to the northeast in the pink box. 

00:55:05.000 --> 00:55:10.000
Low creep in the blue box in the middle and no detectable creek to the southwest and the yellow box.

00:55:10.000 --> 00:55:15.000
And for 2020 to 2023 the creep increased across this entire section of the fault 

00:55:15.000 --> 00:55:34.000
to rapid creep to the northeast in the pink box, rapid creep with a large creep event in the middle box and low creek to the southwest in the yellow box. And during 2023 after the Pazarcik earthquake, the creep behavior changed again with the patterns facially reversing where the pink box to the northeast now shows no detectable creep, the

00:55:34.000 --> 00:55:35.000
blue box in the middle shows low creep,

00:55:35.000 --> 00:55:41.000
and the yellow box to the middle shows low creep and the yellow box to the southwest now shows the most traffic creeper after su.

00:55:41.000 --> 00:55:52.000
So this suggests that when the rupture reached the Puturge segment, it reached an area that had showed no detectable creep that may have had creep properties, which abutted a lock segment preventing it from creeping.

00:55:52.000 --> 00:56:01.000
So yes, creep could have stopped the Pazarcik rupture. So how do we apply the pattern observed in this changes to creep behavior to northern California?

00:56:01.000 --> 00:56:07.000
So here's a map of California with the faults represented by the grey lines and the creep segments represented by the red lines.

00:56:07.000 --> 00:56:27.000
A creep is observed in many faults in this area, including on the Hayward, Calaveras, and Rodgers Creek faults and we will take a closer look at the 2014 South Napa earthquake which experienced rapid post-seismic creep similar to the Puturge segment after the 2023 Pazarcik earthquake.

00:56:27.000 --> 00:56:37.000
Here's a diagram from the magnitude 6.0 South Napa earthquake which shows the accumulated coseismic slip and afterslip along fault for 15 km.

00:56:37.000 --> 00:56:50.000
The coseismic slip had the largest displacement of 45 km at 10 km from the epicenter and over the next 60 days the greatest afterslip of 35 cm occurred about 6 km from the epicenter.

00:56:50.000 --> 00:56:59.000
So before this earthquake, creep was not really detected. However, an area with a high coseismic of that's an area with a high afterslip.

00:56:59.000 --> 00:57:04.000
So the lock segment ruptured coseismically and the section with creep properties had the highest afterslip.

00:57:04.000 --> 00:57:11.000
Similar to the pattern observed with the Pazarcik rupture and the adjacent high afterslip on the creeping fault segment.

00:57:11.000 --> 00:57:23.000
This suggests that high afterslip can be experienced next to lock segments, missed centimeters for months, and continue to break pipes, water lines, and other infrastructure that crosses the fault.

00:57:23.000 --> 00:57:32.000
So in conclusion from 2014 to 2020 the Puturge segment of the Eastern Anatolian fault shows creep it rates up to 6 mm/year.

00:57:32.000 --> 00:57:39.000
During 2020 to 2000, and 2023, the northeast Puturge segment shows up to 12 cm of afterslip.

00:57:39.000 --> 00:57:45.000
With essential part slipping 8 cm in an 18-day creep event in August, 2020.

00:57:45.000 --> 00:57:52.000
And from February to November of 2023 there was smaller cumulative displacement of less than 6 cm.

00:57:52.000 --> 00:58:01.000
And suggesting perhaps that the southwest section of the Puturge segment has creep properties that allow it to slip with afterslip but not coseismic-slip.

00:58:01.000 --> 00:58:09.000
Our results are consistent with creep along the whole Puturge segment and is plausible that it arrested the Pazarcik earthquake rupture.

00:58:09.000 --> 00:58:24.000
In northern California could face similar changes to fault behavior after a large-scale earthquake, including the highest afterslip in areas of lowest coseismic slip and centimeters of afterslip offset for a year or longer.

00:58:24.000 --> 00:58:30.000
We like to give a special thanks to NSF and Tubatak for supporting our research.

00:58:30.000 --> 00:58:40.000
And thank you for listening.

00:58:40.000 --> 00:58:50.000
Alright, great talk. A lot of interesting discussion. So our last talk for this session is given by Rob Churchill.

00:58:50.000 --> 00:58:59.000
And it's entitled, "Aseismic afterslip an important but not universal driver of aftershock sequences."

00:58:59.000 --> 00:59:08.000
Hi everyone. My name is Robert Churchill and today I'm going to be talking about aseismic afterslip which is a post- seismic relaxation mechanism, 

00:59:08.000 --> 00:59:19.000
and something that is often proposed to drive aftershock sequences. The work I'm presenting today was undertaken mostly during my PhD in a short postdoc at the University of Bristol, 

00:59:19.000 --> 00:59:27.000
where I worked alongside co-authors and supervisors, Max Werner, Juliet Biggs and Ake Fagereng.

00:59:27.000 --> 00:59:33.000
To start things off, it's probably worth going into what an aftershock is and why we should care.

00:59:33.000 --> 00:59:43.000
Obviously aftershocks are a deadly and costly global hazard and if we want to be able to model them and potentially even forecast them one day it's important we understand them better.

00:59:43.000 --> 00:59:51.000
Purely from a statistics point of view, we can define an aftershock as an earthquake that follows a larger earthquake closeby in space and time.

00:59:51.000 --> 00:59:55.000
And there's a number of statistical laws that help describe the average or typical behavior of aftershock sequences such as how they decay through time and how many you should expect 

00:59:55.000 --> 01:00:14.000
for a mainshock of a given size, and essentially a few of these laws, bunch together can create frameworks and models like ETAS that describe the generic behavior of aftershock sequences really quite well.

01:00:14.000 --> 01:00:27.000
The issue is when we see sequences which are outlying in terms of their behaviors, commonly this is complexity and spatial temporal distributions, but maybe that they were productive or unproductive.

01:00:27.000 --> 01:00:33.000
It may be that they had swarm-like behavior, lots of large aftershocks, that sort of thing.

01:00:33.000 --> 01:00:41.000
If we really want to understand those case studies, we probably need to have a deeper dive into the physics and understand the physics a bit better.

01:00:41.000 --> 01:00:57.000
So in terms of the physics of aftershock triggering, you'd probably define an aftershock as any earthquake that occurs because part of some fault has been pushed to failure by a nearby earthquake and the stress change that it has imparted.

01:00:57.000 --> 01:01:09.000
And there's a few different mechanisms that are often talked about in the literature, and I'll go through these quickly and I've tried to find notable Californian examples where this mechanism has been suggested to trigger aftershocks.

01:01:09.000 --> 01:01:16.000
So the principal mechanism or the most common, the most established is coseismic Coulomb static stress change.

01:01:16.000 --> 01:01:23.000
This is essentially just the change of stress in the crust due to the mainshock rupture, 

01:01:23.000 --> 01:01:41.000
and I've included the classic Landers '92 figure from King et al. There's also dynamic triggering, which is spoken about lots and this is essentially just the passage of seismic waves that promotes failure on faults and this is obviously transient triggering,

01:01:41.000 --> 01:01:48.000
but again at Lander's their is evidence that this was in part triggering some aftershocks.

01:01:48.000 --> 01:02:00.000
The third mechanism that people invoke is "secondary triggering." So this is aftershock triggering due to previous generations of aftershocks and the sort of evolving stress field.

01:02:00.000 --> 01:02:10.000
And then on the right of the table we have three post-seismic mechanisms. So these are things that happen, you know, in the days, weeks, months after the mainshock 

01:02:10.000 --> 01:02:17.000
and there's a strong time dependence to these. So pore-fluid processes, things like pore-fluid migration and that sort of thing.

01:02:17.000 --> 01:02:24.000
There's evidence that at Northridge in '94 this may have triggered aftershocks,

01:02:24.000 --> 01:02:27.000
aseismic afterslip, obviously this is what I'm going to be talking about today and I'm going to go into a lot of depth.

01:02:27.000 --> 01:02:39.000
But it's worth mentioning that at Parkfield in 2004 there was lots of evidence of not only a lot of afterslip but that afterslip potentially triggered aftershocks.

01:02:39.000 --> 01:02:45.000
And finally, viscous and viscoelastic processes, these tend to be over longer timescales and happen deeper,

01:02:45.000 --> 01:02:53.000
deeper enough and the Hector Mine earthquake, which was supposed aftershock 7 years after the Landers event 

01:02:53.000 --> 01:03:00.000
Is believed by some to have been triggered by Viscoelastic relaxation. So what is afterslip?

01:03:00.000 --> 01:03:06.000
Aseismic afterslip is gentle fault-sliding that redistribute stresses left by an earthquake.

01:03:06.000 --> 01:03:11.000
And it happens in the weeks, months, and sometimes years following the earthquake, and it decays with time.

01:03:11.000 --> 01:03:23.000
It's really important to note that afterslip is aseismic and it does not include the slip associated with aftershocks, although very occasionally you might see people refer to sort of a generic term,

01:03:23.000 --> 01:03:30.000
afterslip to mean all slip that happens after an earthquake. We specifically are talking about aseismic afterslip.

01:03:30.000 --> 01:03:37.000
And the most common interpretation of this is brittle creep that occurs on velocity strengthening parts of the fault.

01:03:37.000 --> 01:03:42.000
So if you look at this diagram on the right, I've shown the afterslip as the blue, 

01:03:42.000 --> 01:03:47.000
and this happens below the seismogenic fault zone; above the seismogenic zone,

01:03:47.000 --> 01:03:53.000
and sometimes it's patches within the seismogenic zone. Like coseismic-slip,

01:03:53.000 --> 01:03:57.000
afterslip can be modeled or slip on a plane, it has a moment, it has a distribution.

01:03:57.000 --> 01:04:03.000
But it must derive from geodetic data because obviously it's aseismic and 

01:04:03.000 --> 01:04:06.000
it also evolves with time.

01:04:06.000 --> 01:04:17.000
In a study published in 2022, me, and co-authors wanted to address questions such as "how does afterslip moments scale with the size of the mainshock? 

01:04:17.000 --> 01:04:25.000
What controls variability around this and you know, just general typical behaviors of afterslip, we believe this wasn't particularly well constrained,

01:04:25.000 --> 01:04:36.000
and we thought this might be interesting to people that care about aftershock triggering, who cared about earthquake cycle behaviors and who also cared about fault zone structure and properties.

01:04:36.000 --> 01:04:47.000
And on this diagram on the right you can see the earthquakes that we looked at. We compiled about 150 studies and I think it was about 50 earthquakes so there was multiple studies for some earthquakes,

01:04:47.000 --> 01:04:54.000
and in the panel below, we've shown a breakdown by the data that was used primarily in these studies.

01:04:54.000 --> 01:05:02.000
And what we found was principally, afterslip moments scales with coseismic moment, this is what the panel on the left shows.

01:05:02.000 --> 01:05:07.000
It's nearly a 1-1 scaling, and that's very useful to know,

01:05:07.000 --> 01:05:19.000
but if we redefine how we measure afterslip, so instead of afterslip moment, if we think about relative afterslip moment, so afterslip moment divided by coseismic moment, we end up with this metric on the right, 

01:05:19.000 --> 01:05:30.000
Emeril and we found that this is typically sort of 10% to 30% of coseismic moment, but there are outliers such as the 2004 Parkfield earthquake as I said earlier had loads of afterslip.

01:05:30.000 --> 01:05:41.000
And it's not abundantly clear why and we look for specific metrics that maybe correlate with with having more or less relative afterslip moment.

01:05:41.000 --> 01:05:50.000
And whilst we found the Emeril weekly correlates with things like rupture aspect ratio and fault slip-rate, It didn't correlate strongly with anything.

01:05:50.000 --> 01:05:58.000
So we found that quite interesting and obviously lots more work is needed there. In this study, we also found that a size make after slip occurs throughout the fault zone.

01:05:58.000 --> 01:06:08.000
So as shown on this diagram on the right where after slip are the blue bands, cos size being slipped up to extent the red bands, you can see that after slip occurs.

01:06:08.000 --> 01:06:18.000
Below, above and within, seismogenic rupture debts. So that classic sort of stratified model of riology in the fault zone.

01:06:18.000 --> 01:06:28.000
This sort of flies in the face of that a little bit. We also noted that modeling methodologies very considerably, there is a lot of uncertainty in these distributions, these slip distributions, these moment estimates.

01:06:28.000 --> 01:06:36.000
So that's something we also spoke about quite a lot.

01:06:36.000 --> 01:06:45.000
So moving on to afterslip as a driver of aftershocks, there's a lot of compelling case study evidence out there that afterslip indeed does drive aftershocks.

01:06:45.000 --> 01:06:53.000
One argument you see made often is that these two things decay very similarly. Both exhibit and a Omori style decay.

01:06:53.000 --> 01:06:59.000
And you also see lots of evidence that after-slipping aftershocks, co-migrate, so they move together 

01:06:59.000 --> 01:07:09.000
similar to that, there's also lots of evidence. Spatial correlations so that perhaps afterslip and aftershocks correlate around the edges of coseismic slip for example.

01:07:09.000 --> 01:07:16.000
And then there's also a few numerical and mechanical models that people have proposed to explain triggering of aftershocks by afterslip 

01:07:16.000 --> 01:07:23.000
and these things are all consistent with one another and they sort of present a quite compelling case at afterslip.

01:07:23.000 --> 01:07:29.000
At least has the strong potential to trigger aftershocks and when you take into account the sort of case study evidence in case study observations.

01:07:29.000 --> 01:07:49.000
You know it's quite easy to believe that afterslip does drive aftershocks so in another study in 2022, we set out to sort of show whether this was apparent globally, whether there were statistical markers that were indicative of triggering of aftershocks by afterslip.

01:07:49.000 --> 01:08:01.000
So what we did is we took that data set of compiled afterslip moments. From our previous paper and we compared that to the corresponding aftershock sequences and characteristics of those.

01:08:01.000 --> 01:08:10.000
So we know that aftershock number is a first order function of mainshock size. This is the Utsu-Seiki productivity scaling or this is what's shown in this Panel A.

01:08:10.000 --> 01:08:19.000
It was very apparent in our data. No surprises there. What we found is that neither absolute nor relative afterslip moment 

01:08:19.000 --> 01:08:34.000
correlated with the number of aftershocks. In absolute terms, that shown in Panel B. But in Panel C we show aftershock numbers relative to what's expected given the Utsu-Seiki productivity.

01:08:34.000 --> 01:08:44.000
So here we're testing whether relative afterslip moment correlates with relative aftershock number and we didn't see a relationship here either.

01:08:44.000 --> 01:08:53.000
In fact, we didn't see relationships between the afterslip and cumulative aftershock moment, seismic rate change, B-value, decay parameters or

01:08:53.000 --> 01:08:59.000
background seismicity rate. So this was really interesting and this presented us a bit of a conundrum.

01:08:59.000 --> 01:09:05.000
Essentially, the problem is afterslip clearly matters as an aftershock triggering mechanism in some case studies.

01:09:05.000 --> 01:09:16.000
There's really good evidence to suggest that it does, but we were unable to support this with a global productivity-based study.

01:09:16.000 --> 01:09:25.000
So this either suggests afterslip matters in some cases settings for triggering aftershocks but not others,

01:09:25.000 --> 01:09:35.000
and or it suggests that afterslip is a driver of the spatial temporal distribution of aftershocks that potentially not a strong driver of overall productivity.

01:09:35.000 --> 01:09:44.000
So this is the final thing we sort of wanted to look at, and in a paper let's hopefully comes up by the time this workshop's running.

01:09:44.000 --> 01:09:55.000
We essentially did that. We looked at the evolving spatial temporal relationships between distributions of coseismic slip, afterslip and on-fault aftershock density.

01:09:55.000 --> 01:10:05.000
For a number of continental earthquakes which for which we had good coseismic slip data afterslip data and aftershock data.

01:10:05.000 --> 01:10:10.000
And as you can see on the right here this is for Parkfield, I believe the data was showing here, 

01:10:10.000 --> 01:10:17.000
we projected onto a best fitting plane and re-gridded coseismic-slip, calculated coseismic gradient.

01:10:17.000 --> 01:10:30.000
Afterslip, which obviously has lots of time steps, calculated those gradients and on-fault aftershocks and calculated the density of those and obviously that has lots of time steps and evolves.

01:10:30.000 --> 01:10:38.000
And essentially we just compared everything and we look for correlations through time. We had a number of hypotheses that we were testing, so we expected coseismic-slip

01:10:38.000 --> 01:10:43.000
to negatively correlate with aftershock density because these are patches that have already slipped.

01:10:43.000 --> 01:10:50.000
We expected coseismic slip to negatively correlate with afterslip because these should be on different host realities.

01:10:50.000 --> 01:11:00.000
We expected afterslip to positively correlate with aftershock density because there's already compelling evidence that exists of co-migrations, 

01:11:00.000 --> 01:11:13.000
and we also expected high slip gradients so either coseismic-slip, afterslip, but just where the gradient of those is high to correlate positively with aftershock density because we thought those would suggest high stress change regions.

01:11:13.000 --> 01:11:20.000
However, we tested everything and we found that spatial relationships differed between different earthquakes.

01:11:20.000 --> 01:11:38.000
And this really was indicative to us that case specificity rules out so essentially every setting is very different and there's no hard and fast rules and this is probably because of those six proposed aftershock triggering mechanisms.

01:11:38.000 --> 01:11:50.000
Some matter more in some case studies than others and there's no simple way to go about predicting, in this case, spatial temporal distribution, but potentially also productivity.

01:11:50.000 --> 01:12:02.000
We also suggest that bolts likely exhibit fine scale frictional and geometric heterogeneity that we can't resolve and that might be part of the reason why we see things we expected to be 

01:12:02.000 --> 01:12:09.000
inversely correlated on top of one another. So to conclude, should we care about afterslip?

01:12:09.000 --> 01:12:14.000
Afterslip is aseismic fault sliding the last days to years after an earthquake.

01:12:14.000 --> 01:12:29.000
Its characteristics can vary hugely between different earthquakes. Afterslip itself is one of six typically sighted aftershock triggering mechanisms and there's great evidence to support the notion that it does play an important role in some settings.

01:12:29.000 --> 01:12:44.000
But we could not find evidence that it plays a ubiquitous universal principal role. Lots more study is needed to drill down into why afterslip varies from earthquake to earthquake and why it matters more for aftershock triggering some settings than others.

01:12:44.000 --> 01:12:50.000
And I think we could probably say the same about all six proposed aftershock triggering mechanisms,

01:12:50.000 --> 01:13:00.000
they clearly all matter in some places, but what we need is statistical markers that help us understand in a given setting.

01:13:00.000 --> 01:13:10.000
Which one should matter more or where one might matter less? So I'm hoping that's the sort of bigger picture that came out at the end of my PhD.

01:13:10.000 --> 01:13:19.000
Thank you very much.















01:13:19.000 --> 01:13:30.000
Alright, great talk. And great, great job to all of our presenters. So, I'll start us off for the question from me.

01:13:30.000 --> 01:13:50.000
And then afterwards, everybody raise your hand or write in the chat if you have questions. One of the things I noticed kind of synthesizing the different talks that we saw today is that I think Jan noticed afterslip that migrates with time.

01:13:50.000 --> 01:13:57.000
And Nicola, after the first 1.5 days, specifically did not notice afterslip that migrates with time.

01:13:57.000 --> 01:14:05.000
Zia, noticed afterslip that propagates up towards the surface over time. And I'm wondering for Rob, did you 

01:14:05.000 --> 01:14:15.000
try to categorize or did you look at it? Do you think we should look at the way afterslip might evolve in time?

01:14:15.000 --> 01:14:19.000
As it migrates around.












01:14:19.000 --> 01:14:37.000
When we were trying to characterize afterslip, I think in that first paper a couple of years ago, We found it quite difficult to come up with sort of single scalar metrics that we could use to categorize sorts of things but some sort of measure of It's migratory, characteristic will probably be really useful.

01:14:37.000 --> 01:14:46.000
It did seem very clear that some We're much more migratory than others. So, you know, as we were shown today, black collar didn't migrate very much at all.

01:14:46.000 --> 01:14:54.000
It just sort of intensified in certain blobs. Yeah, that seems like ripe research for somebody to get stuck in with soon.

01:14:54.000 --> 01:15:03.000
But, It's not something we were able to do formally. It seems like quite a big, I don't know how you categorize that.

01:15:03.000 --> 01:15:12.000
It certainly varied.

01:15:12.000 --> 01:15:22.000
Does anybody have any questions? Raising, raising hands? I think. Some of the recent questions I haven't quite read yet.

01:15:22.000 --> 01:15:34.000
Go ahead, anybody?

01:15:34.000 --> 01:15:44.000
I guess while we look for other questions, I can bring one question on my own. So I think I loved, kind of, I enjoyed the kind of reading kind of these different examples.

01:15:44.000 --> 01:15:50.000
Of aftersmith behavior. So it seems like in a lot of cases these apps are quite patchy looking.

01:15:50.000 --> 01:16:01.000
And absolute occurrence room places but not others. I wonder, I was kind of thinking that kind of on my mind like What explain what kind of produces?

01:16:01.000 --> 01:16:12.000
Or factors produce apps to a slip somewhere but not others. There's some ideas about why it's So patchy looking and why there's sometimes just a gap.

01:16:12.000 --> 01:16:18.000
Resembling a nap out. There are only 2 patches for a localized.

01:16:18.000 --> 01:16:25.000
If, yeah, maybe I can ask again, new inversions. Is there anything that you can say about places without after.

01:16:25.000 --> 01:16:32.000
Since you have these. In inverted parameters.

01:16:32.000 --> 01:16:45.000
Yes, I think. I think we Notice the difference between the fractional parameters and the I'll say, Smake and Sees Mic agents of default.

01:16:45.000 --> 01:16:58.000
Would like the start feeling behavior was much more pronounced in the portions of the fault that Well, as opposed to the Cosses.

01:16:58.000 --> 01:17:14.000
As to of course, our Dynamic model is still only a model and we don't have any Liquids and today and the additional physics over the dynamic model.

01:17:14.000 --> 01:17:26.000
Basically, so It's possible that what we see in the functional parameters is a projection of some other physical mechanism.

01:17:26.000 --> 01:17:37.000
It'll be interesting too. See that but we would need much more complicated model for that.

01:17:37.000 --> 01:17:41.000
So, Rachel, do you wanna ask a question?

01:17:41.000 --> 01:17:52.000
Yeah, can do. My question was really related to what was in the chat, which was just about constraining the depth extent of a seismic and post-sized mixlit.

01:17:52.000 --> 01:18:17.000
I mean there's a panel of experts here who know more than I do. Obviously the closer your observations the better you do it shallow depths, but I mean if you see creep you know on the ground Do we know whether it goes down 5 meters, 10 meters, a hundred a kilometer, when you have your GPS, You know, are we better at constraining whether it's?

01:18:17.000 --> 01:18:26.000
Reached great depth, shallow depths. Can we differentiate between the top 10 meters and the top 100 meters in the top kilometer?

01:18:26.000 --> 01:18:31.000
That, you know, how can non experts assess what's what's seen and what's typical? What do we know?

01:18:31.000 --> 01:18:45.000
What's typical data available? Thanks.

01:18:45.000 --> 01:18:46.000
Okay.

01:18:46.000 --> 01:18:55.000
Well, yes, yes, I know. We can, with the increasing death of course the resolution and our authorities are

01:18:55.000 --> 01:19:06.000
Becomes, the, the increases. With increasing depth but if you have a dense observation of in.

01:19:06.000 --> 01:19:18.000
Then yeah. You can you can Estimate the depth.

01:19:18.000 --> 01:19:26.000
I can't do that based on my experience of working on park field. So if the model is just constrained by GPS data.

01:19:26.000 --> 01:19:38.000
So you cannot bid, but I think you cannot really resolve better than the spacing of the stations. Right, even for Parkfields when the stations are so close to the faults.

01:19:38.000 --> 01:19:44.000
I think the resolution in my model, I think, confidently would say it's 3 to 4 kilometers.

01:19:44.000 --> 01:19:53.000
But that resolution itself depends on the data, space, station spacing, but also on your regularization, your constraints, your prior constraint that goes into it.

01:19:53.000 --> 01:19:56.000
You may increase the resolution a little bit. But it's hard to do much better with the GPS itself.

01:19:56.000 --> 01:20:07.000
With Insar, I think that with the much smaller pixel size. You can constraint on the near surface.

01:20:07.000 --> 01:20:21.000
Near fall, creep much better but that would be challenging for deep in a similar way as with the insert with the GPS data.

01:20:21.000 --> 01:20:38.000
Yeah, we've actually been talking about this exact question a lot here because with a creek meter you have an aperture of 10 meters or so and when there's a creep event or something seen in the creep meter the question becomes how deep does it go and without information from sort of the wings of the deformation field.

01:20:38.000 --> 01:20:47.000
We, don't know. That's where insar helps as long as it's coherent out, you know, several kilometers from the fault.

01:20:47.000 --> 01:20:50.000
So, Gordon, do you have a question?

01:20:50.000 --> 01:20:58.000
Yeah, I have a question. How does triggered slip? Factor into the the way you observe after slip.

01:20:58.000 --> 01:21:12.000
I mean it's it's a a seismic but My concern is like if you look at a rupture like the Napa rupture, it seems like all these parallel strands and also the northern part of the rapture.

01:21:12.000 --> 01:21:19.000
Was much smaller like the magnitude of slip was very small like Less than 10 cm.

01:21:19.000 --> 01:21:36.000
Was are we overestimating the rupture length if that's just triggered slip?

01:21:36.000 --> 01:21:44.000
Yeah, wondering do any of our panelists wanna give a perspective on that?

01:21:44.000 --> 01:21:49.000
And are we also not?

01:21:49.000 --> 01:22:04.000
Observing or identifying a triggered slip. As triggered slip a lot of time so we just consider it part of a rupture.

01:22:04.000 --> 01:22:05.000
Well, I think it's an open question.

01:22:05.000 --> 01:22:19.000
Yeah, I guess that having existing instrumentation is a key for Actually resolving. What's happening immediately after the Kozizmic raptures that That's I think it's a unexplored territory.

01:22:19.000 --> 01:22:29.000
Where there are very few observation actually constraining the stop of cozismic rapture and the beginning of after slip.

01:22:29.000 --> 01:22:41.000
And I think this is a really And really. Inj's on availability of near field observations already.

01:22:41.000 --> 01:22:44.000
Established.

01:22:44.000 --> 01:22:45.000
When?

01:22:45.000 --> 01:22:57.000
Right, but what I guess my There is a difference like that you can distinguish between after slip and triggered slip right because after slip only occurs where you've had primary slip.

01:22:57.000 --> 01:23:07.000
Co-size makes slip, right? And triggered slip can be only triggered. But not seismic.

01:23:07.000 --> 01:23:10.000
Not co-size me.

01:23:10.000 --> 01:23:11.000
Yes.

01:23:11.000 --> 01:23:24.000
Oh, that's an interesting, that's an interesting question. I, I hadn't, I didn't realize or like I didn't think that some of that after slip needed or required to happen in a place where co-size makes slip happened.

01:23:24.000 --> 01:23:38.000
Yes, well, for the, you see the cause I think, the information is captured by inside very well and there is almost no slip at the surface.

01:23:38.000 --> 01:23:49.000
All the costs asked me into programs shows that it's it's clear there's no discontinued at the surface But so.

01:23:49.000 --> 01:24:09.000
The passage of the second satellite may be in after a few days so that nothing happened during the this week or so but when we process the data after we see that this sleep reaches to the surface very clearly, observed and measured by inside also.

01:24:09.000 --> 01:24:20.000
So there is a delay. Of silic to the surface. Turing can observe hundreds, of away from the fault and Roger is not here, but you know.

01:24:20.000 --> 01:24:31.000
The, A crip, 500 commisser away on this ledpasha.

01:24:31.000 --> 01:24:37.000
Cripping. And yeah, it's.

01:24:37.000 --> 01:24:46.000
But sometimes, of course, but is mapped as, postsax, is partially, it's after sleep.

01:24:46.000 --> 01:24:55.000
We cannot. Separate unless you have good dense temporal observation.

01:24:55.000 --> 01:25:00.000
Okay.

01:25:00.000 --> 01:25:05.000
Ruth?













01:25:05.000 --> 01:25:20.000
How might I wanna say thank you so much to the speakers today, bunch of great talks. So I wanted to return us to a question that, also in the chat and that has to do with, creeping faults, creeping parts of thoughts.

01:25:20.000 --> 01:25:29.000
Stopping large. Ruptures. So when I wrote the, 2017 review article on large earthquakes and creeping flops.

01:25:29.000 --> 01:25:37.000
I hadn't known yet that the eastern Anatolian fault was creeping and then at AGU that year, I'm someone came up to me and said, oh yeah, he's done a toian font also is.

01:25:37.000 --> 01:25:45.000
And of course we know that really well now. The standard Tolian fault is also had the largest earthquake.

01:25:45.000 --> 01:25:53.000
On a continental creeping vault that 6 6.8 and then of course the most recent devastating earthquakes.

01:25:53.000 --> 01:26:00.000
So what are what is our thinking now now that we have more information? Do we think that creeping sections are false?

01:26:00.000 --> 01:26:07.000
Can stop large earthquakes? Or does it depend on how fast the font is creeping relative to the long term, slip rates?

01:26:07.000 --> 01:26:13.000
So I was just wondering what, how you are thinking about, given that we have new knowledge.

01:26:13.000 --> 01:26:18.000
Can we use this say when we're doing, seismic, modeling or mapping?

01:26:18.000 --> 01:26:23.000
Can we, can we just say that, well, this here is a major prepping section.

01:26:23.000 --> 01:26:29.000
It's gonna stop the next large earthquake. Or are we still in the state that we were before and that we're not sure?

01:26:29.000 --> 01:26:37.000
So just wondering when all of you are thinking about this. Thank you.

01:26:37.000 --> 01:26:46.000
Well, both, the, the, last year, February, the sixth and the, 2,000, and 20, a clearly show that, yeah.

01:26:46.000 --> 01:26:55.000
Fault crib can, that can stop rupture, prerogations of larger cakes.

01:26:55.000 --> 01:27:10.000
But I'm not seismologist, maybe, the rupture propagation or I don't know do some parameters the speed or the rate of creeping.

01:27:10.000 --> 01:27:23.000
Types of frogs may have some role in this. But the certainly the rapture, the 2 point, about 2 and a half metres of.

01:27:23.000 --> 01:27:40.000
Centered like at the depth of like 3 kilometers but not reached the surface. So it's it's very clear and there just above the hypo center of the resort take the crypt rate reaches the bread velocity so it kind of full almost full right.

01:27:40.000 --> 01:27:48.000
But alone other sections the corporate rate decreases. So and after sleep kind of fills the gap.

01:27:48.000 --> 01:28:06.000
But then I just noticed that, you know, one of the, and we said, Solutions indicate high sleep peak is noticed by summer, some one year, that there's a high city peak, ending up between the 2 earthquakes.

01:28:06.000 --> 01:28:18.000
It's it's it looks also quite dip so yeah maybe that high creep Deep, maybe, also.

01:28:18.000 --> 01:28:26.000
The reason why the The first both quick stop to the east.

01:28:26.000 --> 01:28:34.000
You know, I just have a comment that relates to this. That's the On the San Andreas at Shalom.

01:28:34.000 --> 01:28:43.000
We had 4 meters of slip in 1857, but we also had a small earthquake in 66 with about.

01:28:43.000 --> 01:28:56.000
15 cm. And it's also creepy. So I think. The the amount of creep at on the fault section is probably.

01:28:56.000 --> 01:29:04.000
One of the key factors, cause it's not creeping that much down there, maybe. 3 to 5 a year.

01:29:04.000 --> 01:29:09.000
Or less maybe even. So.

01:29:09.000 --> 01:29:11.000
Tom, do you want to ask a question?

01:29:11.000 --> 01:29:20.000
Yes, this one is for Celeste. And in a number of her red and blue plots for the.

01:29:20.000 --> 01:29:32.000
East Anatolian fault. But not all of them. In the red zone. In the those, site flights.

01:29:32.000 --> 01:29:48.000
In most of the red zones. There is a a small blue zone orthogonal to the East Anatolian fault.

01:29:48.000 --> 01:29:55.000
And I just wonder what caused that. But again, it's not in all of the.

01:29:55.000 --> 01:29:58.000
Those plots just in some of them.

01:29:58.000 --> 01:30:02.000
Landslides, I would say.

01:30:02.000 --> 01:30:03.000
Yeah, that's right. Where

01:30:03.000 --> 01:30:12.000
So that's by the way, sorry, to interrupt that there is, the earthquake, aftershock there was 5.6 and there are interferograms very clearly.

01:30:12.000 --> 01:30:18.000
So it's, I don't think that's a, triggered or see the prevent.

01:30:18.000 --> 01:30:26.000
It's an aftershock and probably driven by the afterslip. Because it is located.

01:30:26.000 --> 01:30:36.000
Yes, at the tip of the rapture. And after sleep probably kind of pushed it and triggered that aftershock with Magnated 5 Points in August.

01:30:36.000 --> 01:30:39.000
The same same period.

01:30:39.000 --> 01:30:43.000
Okay.

01:30:43.000 --> 01:30:44.000
Sorry, yes.

01:30:44.000 --> 01:30:48.000
So you know what I'm talking about that I wasn't. I wasn't quite. And it's quite dramatic.

01:30:48.000 --> 01:31:00.000
In those plots. Is it in orthogonal fault? You know, perpendicular to The, you stand totally in fault.

01:31:00.000 --> 01:31:05.000
I think it's more along. I'm not sure how well you can see it and the.

01:31:05.000 --> 01:31:11.000
The displacement maps, but, it's follows the Euphrates river.

01:31:11.000 --> 01:31:14.000
So it's more likely the landslides.

01:31:14.000 --> 01:31:17.000
Can you not share a screen?

01:31:17.000 --> 01:31:41.000
Yeah. Well, now let me, I don't have the thing.

01:31:41.000 --> 01:31:51.000
Go ahead and share screen.

01:31:51.000 --> 01:31:59.000
Okay, so I thank you or sorry, let me see here. That's interesting. Are you talking about these?

01:31:59.000 --> 01:32:00.000
Never others.

01:32:00.000 --> 01:32:01.000
Kind of Yeah.

01:32:01.000 --> 01:32:11.000
No, it's It's for it's really near the middle of the plot. But it's not showing up on this You know, red and blue plot.

01:32:11.000 --> 01:32:12.000
Okay, hold on.

01:32:12.000 --> 01:32:19.000
It's, it's near the center of the, the fault. Right in, well, you can't see my cursor, yes.

01:32:19.000 --> 01:32:29.000
But it would be farther to the southwest.

01:32:29.000 --> 01:32:30.000
Right here.

01:32:30.000 --> 01:32:38.000
Okay, that, yeah, this thing right there. So. Okay, and it shows up. You know stronger on some of the other but this is basically it.

01:32:38.000 --> 01:32:39.000
So these are landslides along the river.

01:32:39.000 --> 01:32:43.000
Yeah, this should be, yeah.

01:32:43.000 --> 01:32:50.000
Yeah, most like I've seen I have similar. Results, they show up every time.

01:32:50.000 --> 01:32:53.000
Okay.

01:32:53.000 --> 01:33:04.000
Thank you. That kind of confused me for a little while.

01:33:04.000 --> 01:33:10.000
No, I've got a question for Jan. Which is, do you, how do you think your, your study or your results could be extended to other earthquakes?

01:33:10.000 --> 01:33:25.000
This this consistent framework between the co size make and post size mix lip do you have thoughts about other earthquakes that might be.

01:33:25.000 --> 01:33:33.000
Sufficiently well instrumented, and a plausible extension of the study.

01:33:33.000 --> 01:33:41.000
Yes, yes, yes, that's a good question. Right now we are actually working on extending.

01:33:41.000 --> 01:33:51.000
This study, by the 2,000 for Parkfield earthquake, we are preparing a paper on that.

01:33:51.000 --> 01:33:56.000
The park field is.

01:33:56.000 --> 01:34:14.000
Somewhat more interesting. We see much more after sleep over the whole fold. It's not just concentrated and 2 patches for us so let's that's interesting, but we will definitely need more earthquakes to make some Morning general.

01:34:14.000 --> 01:34:23.000
General. To have a more general idea about what's happening. With this joint models.

01:34:23.000 --> 01:34:36.000
You should talk to Julia about that.

01:34:36.000 --> 01:34:42.000
We have a way stand from RC.

01:34:42.000 --> 01:34:53.000
It's nice to have such a panel of experts, but. To ask this question. Doesn't afterslip.

01:34:53.000 --> 01:35:04.000
And after shocks. Makes sense to all of you. As far as Once you have slip. You' an elastic medium.

01:35:04.000 --> 01:35:21.000
That you would expect. Over time to see these phenomena. I just like your comments on that.

01:35:21.000 --> 01:35:33.000
I mean, like in the turkey. Earthquake. It seems to me you have slip at depth.

01:35:33.000 --> 01:35:34.000
Right.

01:35:34.000 --> 01:35:41.000
And then it over time it grows to the surface. That to me. Makes perfect sense.

01:35:41.000 --> 01:35:55.000
Just any of your comments on. Just the physical reality of these earthquakes.

01:35:55.000 --> 01:36:11.000
Well, I don't know if it takes, Exactly. With them. This, the scripting sections, are on soft rocks, let's say, and when the rapture basically cannot propagate into.

01:36:11.000 --> 01:36:23.000
So it's a Windows is strengthening. It's so stop and then the stress due to the cost as specific adept kind of.

01:36:23.000 --> 01:36:44.000
Can the only accommodated you know closer Hey, seismic slip. Alright, that surface all around the across Aspenk, city patch.

01:36:44.000 --> 01:36:45.000
Yes.

01:36:45.000 --> 01:36:49.000
So the the earthquake occurs in the strong rock at depth. And then I can't come to the surface in the softer rock.

01:36:49.000 --> 01:36:50.000
So much.

01:36:50.000 --> 01:37:05.000
Yeah, maybe I don't know the thickness of the cripping section maybe is important. Maybe root knows, can explain better, maybe the, yeah, the thickness of the, cooking section or.

01:37:05.000 --> 01:37:22.000
The rate. Because The decripping is not kind of, it could. Degree sections are not fully, releasing the, the far fit stress in the sleep.

01:37:22.000 --> 01:37:33.000
So there's still some accumulation. So it's not like flowing 100%. There is still some accumulation and this could be broken cause seismically I assume.

01:37:33.000 --> 01:37:42.000
But, but so when you have slip it depth. And I've been on the field a lot and It comes it.

01:37:42.000 --> 01:37:51.000
It slips afterwards or. Towards the surface. And the non seismogenic kind of part of the earth.

01:37:51.000 --> 01:37:59.000
But it reoccurs right on old scarves. So it seems to, it seems to have a behavior that.

01:37:59.000 --> 01:38:08.000
It comes to the surface. In very distinct. Repeating areas. Can you?

01:38:08.000 --> 01:38:14.000
Exactly because we have all the morphology we see default that that means that dropter is a surface if that morphology is totally produced by ASIC. We don't know.

01:38:14.000 --> 01:38:29.000
And it's unlikely. Sometimes it I think it probably gets true or sometimes that maybe sometimes it I think it probably gets true or sometimes that maybe depends true or sometimes that maybe it is true or sometimes that maybe it is true or sometimes that maybe depends on direct. I don't know.

01:38:29.000 --> 01:38:32.000
I want to know.

01:38:32.000 --> 01:38:34.000
Thank you.

01:38:34.000 --> 01:38:38.000
Alright, so, we'll take one more question from Oz.

01:38:38.000 --> 01:38:45.000
Hi, thank you very much. First I made, I want to reiterate one of my comments in the chat that, you know, We are always looking for a yes or no questions.

01:38:45.000 --> 01:38:49.000
Does it always stop the, rupture if there's a creeping section or not.

01:38:49.000 --> 01:39:04.000
I don't think this is a fair question because these are evolving conditions from earthquake to earthquake or even within the earthquake sometimes.

01:39:04.000 --> 01:39:22.000
And they may not be persistent. So that's something to keep in mind. The other question, actually the question I have is the this is a really interesting observation that the slip starts on the you know hard rocket depth and then you know it's it's either dampened or delayed through the software or vehicle rock towards the surface.

01:39:22.000 --> 01:39:27.000
Do we have a geophysical or any other kind of confirmation of this rock strength profile at these locations.

01:39:27.000 --> 01:39:36.000
If not, that would be a really interesting study to follow up with. Thank you.

01:39:36.000 --> 01:39:57.000
We asked for that. We don't have the direct GPS, your physical observation measurements, but we have the rocks, the geology itself, and when we look at the geology, we have sheaths and gn and otholites, zapentines, we know that these rocks contain minerals with law friction.

01:39:57.000 --> 01:40:06.000
How far off that? To the depth that it calls, we don't know, but they are the rocks.

01:40:06.000 --> 01:40:16.000
And I want to say something that you know after the 19 ninety-nineties with, okay, okay, it's 150.

01:40:16.000 --> 01:40:25.000
Kermits long enough, only the small part of the fold. Is exhibiting after the state after I don't know how many is not 25 years.

01:40:25.000 --> 01:40:41.000
And when we check the data before those corrects and there was no, no, no, no, creep or anything, but the, the, that area, that section central section of the ismit structure started moving a seismically and still moving today.

01:40:41.000 --> 01:40:46.000
And when we look at drugs again at the surface, patchy, we see some volcanic rocks.

01:40:46.000 --> 01:41:02.000
Most of the drugs are buried by Qatar. I said, we don't know. But, it seems that they are related with the, Depends on what kind of rocks, faults are 

01:41:02.000 --> 01:41:04.000
located.

01:41:04.000 --> 01:41:07.000
Thank you, Zia. And cheers to.

01:41:07.000 --> 01:41:15.000
Hello, from Istanbul.

01:41:15.000 --> 01:41:22.000
Yeah, so I thought I would I would comment on that also with 2 examples one from the East. And.

01:41:22.000 --> 01:41:29.000
Roger, Bill, and very kindly sent us some samples of the fault goug within the PETERG metamorphics.

01:41:29.000 --> 01:41:38.000
And with a lot of help from Tamara Jeppson, we tested those in the lab at different temperature conditions.

01:41:38.000 --> 01:41:48.000
Kind of trained to mimic different depth intervals of the crust. And so what we found is that for the shallow intervals from basically room temperature to 100o C.

01:41:48.000 --> 01:41:56.000
The materials velocity strengthening. Where is that? 200o C, it was velocity weakening.

01:41:56.000 --> 01:42:04.000
So If the, metamorphic rocks were to extend that deep in the crust, we might expect earthquakes to nucleate, propagate.

01:42:04.000 --> 01:42:13.000
Below you know around 5km depth and then reach velocity strengthening behavior more shallowly.

01:42:13.000 --> 01:42:27.000
And then for the South Napa sequence, I think there's a pretty good evidence as Yon pointed out for the different geologic units that may have control of the coseismic and postseismic behavior.

01:42:27.000 --> 01:42:43.000
So. And as you said, the southern portion is an alluvial basin. With really clay rich materials and so i think that's another kind of as clear as we can say, case of there being kind of bedrock at depth.

01:42:43.000 --> 01:42:50.000
That hosted the co-size mixed set. And these shallow sediments that may get promoted afterslip.

01:42:50.000 --> 01:42:59.000
And one thing that I thought was really interesting from your results, Was that the? The rate parameter b minus a.

01:42:59.000 --> 01:43:07.000
Wasn't that different between the 2 sections of the rupture. They were both you know, nominally velocity strengthening.

01:43:07.000 --> 01:43:30.000
And so I was wondering, you know, if like, basically, how, are the dynamic effects in in your model, the maybe the directivity of the rapture in terms of reaching, you know, so much coast size, next step towards the surface, given that the frictional parameters seem to be Not that different actually.

01:43:30.000 --> 01:43:42.000
Yes, that's, that's a good question. Yeah, yeah, I think the directional it is Probably pretty important and also the Enhanced.

01:43:42.000 --> 01:43:58.000
Weakening. affect where the fault behaves. As we can. Even for negative or strengthening the minus I.

01:43:58.000 --> 01:44:11.000
If the slip-rate is high enough. So I think that also played a very large role and that plates are all probably in the

01:44:11.000 --> 01:44:12.000
Okay.

01:44:12.000 --> 01:44:18.000
Okay. Yeah, I think, you know, another thing that occurred to me kind of watching all of this and thinking about you know, spatial correlations between.

01:44:18.000 --> 01:44:32.000
Creep and inter seismic creep and where we expect afterslip. You know, Napa is another good example of where there was no documented inter seismic creep, at least as far as I know, and then it had this tremendous after slip signal.

01:44:32.000 --> 01:44:37.000
So We're thinking about, you know, California hazards and where we should expect. After slip to occur, I still think, there are a lot of unknowns about that.

01:44:37.000 --> 01:44:46.000
So exciting area of research and thank you all for your great talks.

01:44:46.000 --> 01:44:51.000
Yeah. Thank you to all of our speakers. We've reached the end of our allotted time.

01:44:51.000 --> 01:45:02.000
Great, great talks by everybody and I'll pass it off to Sarah. To wrap us up.

01:45:02.000 --> 01:45:09.000
Thank you so much to all of our wonderful speakers and to our wonderful moderators. That was really a fantastic session.

01:45:09.000 --> 01:45:18.000
Thank you so much. So this is all lunch break. I know it's a little bit early on the West Coast, but just think usually every day everyone east of us is hungry enough to eat one of those cute autos by the time we break.

01:45:18.000 --> 01:45:30.000
So this is this is for them. We will come back in 1 hour at, 12:15pm (Pacific Time).

01:45:30.000 --> 01:45:38.000
And as always, we invite the speakers in our next session to come at 12. Straight up so that we can help them get set up for the session.

01:45:38.000 --> 01:45:42.000
So that would be Nadine, Richard, Jean, and Marine, Emily, Rob, Jessica, Noah, and Kevin.

01:45:42.000 --> 01:45:56.000
So please come early if you want. We'll be here. Otherwise everybody else come at 1215 and have a fantastic lunch or dental or midnight snack depending on where you are around the world.

01:45:56.000 --> 01:45:57.000
Thank you so much.

01:45:57.000 --> 01:46:00.000
Bye