WEBVTT
Kind: captions
Language: en-US

00:00:01.200 --> 00:00:07.740
[Silence]

00:00:07.800 --> 00:00:13.120
[inaudible background conversations]

00:00:13.120 --> 00:00:17.140
All right. Hello, everybody,
and welcome to

00:00:17.140 --> 00:00:20.320
Earthquake Science
Center seminar.

00:00:20.320 --> 00:00:24.240
A reminder of an upcoming
special seminar on Friday.

00:00:24.250 --> 00:00:28.110
Tom Mitchell from University
College London, who’s visiting

00:00:28.110 --> 00:00:30.690
the rocks mechanic –
rocks mechanics labs is going to

00:00:30.690 --> 00:00:38.260
be giving a talk on fault zone –
fault zone pulverization. Sorry.

00:00:38.260 --> 00:00:41.910
And then we’re going to have
two weeks with no seminar.

00:00:41.910 --> 00:00:44.970
Next week during the SSA meeting,
and then the week after that

00:00:44.970 --> 00:00:48.510
is going to be the Northern
California Hazards Workshop.

00:00:48.510 --> 00:00:56.600
So, might as well get going, then, and 
introduce today’s speaker, Kang Wang.

00:00:56.600 --> 00:01:01.540
He got his Ph.D. from Scripps
working with Yuri Fialko.

00:01:01.540 --> 00:01:07.340
And then he came to the Bay Area
to work with Roland Burgmann

00:01:07.340 --> 00:01:13.220
at Scripps – sorry, Roland
Burgmann at Berkeley in 2017.

00:01:13.220 --> 00:01:19.020
And his work has focused on
looking at postseismic observations

00:01:19.020 --> 00:01:21.440
for many large earthquakes.
Okay.

00:01:21.440 --> 00:01:26.780
- All right. Thanks, Jeanne.
And thanks for the invitation.

00:01:26.780 --> 00:01:30.220
And it’s my great pleasure to
be here to talk about some of my

00:01:30.220 --> 00:01:35.260
recent work that is primarily
done in the past few months.

00:01:35.260 --> 00:01:38.799
And this work is still in progress,
so if you have any questions,

00:01:38.800 --> 00:01:42.980
comments, thoughts, at the end
of the talk, please do let me know.

00:01:44.000 --> 00:01:49.560
All right, so I would like to
divide my talk into four parts.

00:01:49.560 --> 00:01:56.799
The first part, I’ll give some very
basic introduction to the geodetic tools,

00:01:56.799 --> 00:02:01.020
primarily the InSAR
that I’m using for this study.

00:02:01.020 --> 00:02:07.060
And second, I’m going to show
some of the observations of

00:02:07.060 --> 00:02:11.910
surface deformations due to the Iran
earthquake – Iran-Iraq earthquake.

00:02:11.910 --> 00:02:18.020
In the third part, I’m going to focus
on the modeling of the postseismic

00:02:18.020 --> 00:02:21.760
deformation. This is kind of
very interesting part to me.

00:02:21.760 --> 00:02:26.920
And finally, I will briefly talk a little
about some of the implications.

00:02:27.780 --> 00:02:32.400
All right, so we know that earthquakes
generate surface deformation,

00:02:32.400 --> 00:02:36.580
and measuring the surface deformation
is a very important for us to understand

00:02:36.580 --> 00:02:40.520
the strain and the stress
removed during an earthquake.

00:02:42.200 --> 00:02:48.300
Nowadays, we usually use two
types of geodetic tools to

00:02:48.300 --> 00:02:52.610
measure the ground deformation.
One is GPS. The other is

00:02:52.610 --> 00:02:57.230
Synthetic Aperture – Interferometric
Synthetic Aperture Radar.

00:02:57.230 --> 00:03:02.740
So the way – I guess GPS is more or less
very familiar to us, but I would like to

00:03:02.740 --> 00:03:08.100
spend a few more – couple minutes
to go over some of the basics

00:03:08.100 --> 00:03:11.800
of InSAR because this is
very important to this study.

00:03:11.800 --> 00:03:18.209
So InSAR is a remote sensing technique.
So it sends out – the satellite sends out

00:03:18.209 --> 00:03:22.970
the magnetic – electromagnetic
waves to the ground.

00:03:22.970 --> 00:03:25.480
And it bounces back.
And second time,

00:03:25.480 --> 00:03:28.230
when the satellite passes by,
it connects a lot of data sites,

00:03:28.230 --> 00:03:34.140
so when do the interferometry, you
will get the measure of the ground.

00:03:35.180 --> 00:03:38.560
So there are some very nice
characteristics of InSAR.

00:03:38.560 --> 00:03:42.480
First, it’s passive remote sensing.
So you don’t have to deploy

00:03:42.480 --> 00:03:46.230
instruments on the ground.
So it connects data – you know,

00:03:46.230 --> 00:03:51.320
all the signal is bounced back by the
lateral scatters on the ground.

00:03:51.320 --> 00:03:56.019
And second, it has a wide coverage.
It can be up to a few hundred

00:03:56.019 --> 00:04:03.569
kilometers, as you will see in the talk.
And third, it’s operating in a microwave

00:04:03.569 --> 00:04:10.209
band, so it works day and night,
and it can penetrate clouds.

00:04:10.209 --> 00:04:14.280
But there are some – there are
some kind of limitations for InSAR.

00:04:15.460 --> 00:04:20.240
For example, InSAR can only
give you one direction observation,

00:04:20.250 --> 00:04:26.650
which is along the line of sight.
And usually InSAR has a relatively

00:04:26.650 --> 00:04:32.560
long revisit time, depending on
orbit design, but it’s typically around –

00:04:32.560 --> 00:04:35.500
it’s up to
[times to] months.

00:04:35.500 --> 00:04:44.080
And also, InSAR is is very prone to
many noises – like atmosphere noises,

00:04:44.080 --> 00:04:49.880
which is a very significant problem
for very small signal detection.

00:04:50.440 --> 00:04:58.720
Nevertheless, InSAR has been
very important for crustal deformation

00:04:58.720 --> 00:05:03.440
studies, especially for studies
related to earthquake deformation.

00:05:03.440 --> 00:05:09.140
For example, this is a very famous
interferogram – or, I guess the first

00:05:09.150 --> 00:05:14.800
interferogram is using InSAR to detect
ground deformation due to earthquake.

00:05:14.800 --> 00:05:17.820
This is a interferogram
of Landers earthquake.

00:05:17.820 --> 00:05:22.820
And, in the past two decades,
we have seen – actually, I should

00:05:22.820 --> 00:05:30.090
say 30 decades – three decades, there
were a lot of observation of Earth –

00:05:30.090 --> 00:05:33.520
Earth observation missions.

00:05:33.520 --> 00:05:40.330
And, as you can see, there are many
satellites – many SAR satellites.

00:05:40.330 --> 00:05:46.300
But there are two unique satellites,
or very special satellites called

00:05:46.300 --> 00:05:52.260
Sentinel 1 and Sentinel B –
Sentinel 1A and Sentinel 1B,

00:05:52.260 --> 00:05:57.800
which are very
kind of special.

00:05:57.800 --> 00:06:04.400
And Sentinel 1 was launched in 2014,
and Sentinel 1B was launched in 2016.

00:06:04.400 --> 00:06:09.000
They were both operated
by ESA and has

00:06:09.009 --> 00:06:14.979
C-band wavelengths, which means the
wavelengths is about 6 centimeters,

00:06:14.979 --> 00:06:20.560
and has a unique observational mode.
It’s called TOPS mode.

00:06:20.560 --> 00:06:28.470
So first, it’s a scan SAR, but so it
has a very wide coverage on the ground.

00:06:28.470 --> 00:06:34.800
It can be up to 250 kilometers.
And it has – because of the

00:06:34.800 --> 00:06:39.520
wide coverage, the revisit time
of the system is very short.

00:06:39.520 --> 00:06:44.080
It’s only – it can be up to six days
for some parts of the world.

00:06:44.080 --> 00:06:47.900
And, for the rest of the world,
for the rest of the continents,

00:06:47.900 --> 00:06:49.530
it’s 12 days everywhere.

00:06:49.530 --> 00:06:51.980
And the most important thing,
I think, is the open access data.

00:06:51.980 --> 00:07:00.500
So everybody can get the data for free.
So this is some very nice features.

00:07:01.420 --> 00:07:07.860
And I just did a
Google search yesterday.

00:07:07.860 --> 00:07:12.880
As of yesterday, there has been
more than 2,000 papers published

00:07:12.889 --> 00:07:19.160
using Sentinel InSAR. And this is
just for earthquake-related study.

00:07:19.160 --> 00:07:21.820
So very impressive data set.

00:07:21.820 --> 00:07:26.640
All right. With that being said,
I think it’s about ready to move on

00:07:26.650 --> 00:07:33.259
to the study I’m
going to show today.

00:07:33.259 --> 00:07:40.180
So this work, I’m focusing on
the deformation due to an earthquake

00:07:40.180 --> 00:07:45.590
that occurred around the Iran
and the Iraq border in 2017.

00:07:45.590 --> 00:07:50.090
The earthquake occurred
on November 12th.

00:07:50.090 --> 00:07:57.699
And, as you can see on this map,
both Iran and Iraq has

00:07:57.700 --> 00:08:05.360
experienced very strong ground shaking
and has caused more than 700 deaths.

00:08:08.180 --> 00:08:15.940
So the earthquake – geologically,
the earthquake occurred in a mountain

00:08:15.940 --> 00:08:22.970
belt called Zagros Fold and Thrust Belt.
This mountain range is caused by

00:08:22.970 --> 00:08:30.020
the continental collision between
the Arabian Plate and the Europe plates.

00:08:30.020 --> 00:08:36.160
And this region is seismically
very active, as you can see

00:08:36.160 --> 00:08:45.660
by these green dots.
So it’s – has been – the seismic –

00:08:45.660 --> 00:08:51.350
this is one of those seismically
active origins in the world.

00:08:51.350 --> 00:08:58.020
However, if you look at the historic
seismicities in the Zagros mountains,

00:08:58.020 --> 00:09:04.360
there is – there is almost no
big earthquakes in the past.

00:09:04.370 --> 00:09:09.300
There was no big earthquakes
of magnitude greater than 7 around

00:09:09.300 --> 00:09:13.100
the whole mountain range.
And most of the earthquakes seem to

00:09:13.100 --> 00:09:20.070
occur at relatively shallow depths – 
probably less than 15 kilometers,

00:09:20.070 --> 00:09:28.030
as you can see from this profile.
So the 2017 earthquake was –

00:09:28.030 --> 00:09:33.890
again, it was magnitude 7.3.
So that was a very exceptional event.

00:09:33.890 --> 00:09:40.959
So we want to kind of look at – and if
you look at the aftershock distributions,

00:09:40.959 --> 00:09:47.760
they seems to have a kind of north-south-
trending distribution of the aftershocks.

00:09:47.760 --> 00:09:51.700
However, the faults – all the mapped –
geologically mapped faults in this –

00:09:51.700 --> 00:09:58.200
in this region seem to trend at a
strike of about probably 330 degrees.

00:09:59.840 --> 00:10:02.360
And of course, this earthquake
didn’t produce any surface ruptures.

00:10:02.360 --> 00:10:07.760
So we wanted to know where the
earthquake is occurring, at what depth,

00:10:07.760 --> 00:10:11.500
and also what – you know,
how the faults are responding

00:10:11.500 --> 00:10:14.080
to this stress change.

00:10:15.160 --> 00:10:24.940
So, to answer these questions, we used
InSAR to map the surface deformation.

00:10:25.680 --> 00:10:30.620
So here’s a map to show the coseismic
interferograms due to this earthquake.

00:10:30.630 --> 00:10:39.440
As you can see, there were up to
probably 1 meter of surface deformation

00:10:39.440 --> 00:10:45.329
around a very wide range.
So here, one fringe, one cycle –

00:10:45.329 --> 00:10:49.420
one color cycle represents
10 centimeters of deformation.

00:10:49.420 --> 00:10:55.030
So it’s – the earthquake produced
up to 1 meter of surface deformation.

00:10:55.030 --> 00:11:00.230
And if you look at these interferograms,
you, you know, see – well, you would

00:11:00.230 --> 00:11:02.680
think the fault is probably
going this way, right?

00:11:02.680 --> 00:11:09.260
Especially when you compare with the
faults and the – fault and – with the –

00:11:09.260 --> 00:11:12.860
kind of the strike of the faults.
You would think of the – the fault

00:11:12.860 --> 00:11:20.140
is going through this line, or at least
parallel to this direction, right?

00:11:20.140 --> 00:11:23.610
But you will see later,
it turns out to be not the case.

00:11:23.610 --> 00:11:27.170
So then what’s the
geometry of the fault?

00:11:27.170 --> 00:11:30.180
To answer this, we first did a –
kind of inversion for the geometry

00:11:30.180 --> 00:11:34.720
using just a single dislocation
to assume the earthquake was

00:11:34.720 --> 00:11:42.790
accommodated by a single dislocation.
And we did this kind of inversion of –

00:11:42.790 --> 00:11:47.690
in a Bayesian framework.
And we solved for the location

00:11:47.690 --> 00:11:51.839
and the size, strike, dip, rake,
and slip of this single distribution.

00:11:51.839 --> 00:12:02.090
What we found was that, you know,
the strike is almost 360 degrees,

00:12:02.090 --> 00:12:06.620
which is almost north-south-trending.
It’s a very striking feature.

00:12:06.620 --> 00:12:09.910
It’s quite different from what
you see from the interferograms

00:12:09.910 --> 00:12:13.160
and also the surface
expressions of the faults.

00:12:13.160 --> 00:12:19.550
And the other feature is that the
depth range – the depth –

00:12:19.550 --> 00:12:24.120
the central depth is around
17 degrees – 17 kilometers,

00:12:24.120 --> 00:12:28.660
which is much deeper than many
of the historical earthquakes.

00:12:31.600 --> 00:12:34.880
And this model, as you can expect,
fits the data pretty well.

00:12:34.880 --> 00:12:38.800
So, even with a single
dislocation model, this –

00:12:38.800 --> 00:12:42.040
all the deformation
can be fit very nicely.

00:12:42.780 --> 00:12:49.320
However, as I said, we want to
answer kind of – we want to explore

00:12:49.320 --> 00:12:54.279
the detailed slip distribution
and also the slip evolution.

00:12:54.280 --> 00:12:57.660
We want to derive a
kind of finite fault model.

00:12:57.660 --> 00:13:02.150
So, based on the geometry we just saw,
we derived a finite fault slip, and this is

00:13:02.150 --> 00:13:08.360
a very straightforward exercise.
As you can see on here, the coseismic

00:13:08.360 --> 00:13:13.720
slip is primarily concentrated at
depths between 15 to 20 kilometers.

00:13:13.720 --> 00:13:19.340
There was no clear evidence –
there was no evidence of any slip

00:13:19.340 --> 00:13:26.460
shallower than 15 kilometers.
The vectors represents the direction

00:13:26.460 --> 00:13:33.860
of the slip, and, as you can see,
it’s very oblique.

00:13:33.860 --> 00:13:41.620
It’s a very oblique thrust.
So the – so here, this is the up-dip,

00:13:41.620 --> 00:13:46.880
and to the east, this is down-dip,
just to remind you because

00:13:46.880 --> 00:13:50.000
this is very important.
I will say it many times,

00:13:50.000 --> 00:13:52.960
up-dip and down-dip,
in the next few slides.

00:13:55.620 --> 00:13:59.440
If you compare – there has – by the way,
there has been a few studies published.

00:13:59.440 --> 00:14:06.600
They, more or less, use the same data
set, and they got very similar results

00:14:06.600 --> 00:14:11.080
in terms of slip distribution
for the coseismic rupture.

00:14:12.860 --> 00:14:18.400
Again, if you compare with the –
kind of the geological faults –

00:14:18.410 --> 00:14:22.970
geologically mapped faults,
they seem to kind of have a exception –

00:14:22.970 --> 00:14:26.460
a very different location.

00:14:27.860 --> 00:14:31.600
It’s not matching the
geologically mapped faults.

00:14:33.520 --> 00:14:37.900
So this is all about
coseismic deformation.

00:14:39.350 --> 00:14:43.720
So far, we have obtained the – we have
determined the geometry of the fault –

00:14:43.720 --> 00:14:49.820
geometry of the rupture and also
very detailed distribution of the slip.

00:14:49.820 --> 00:14:53.040
But this is not the complete story.
So we know the earthquake –

00:14:53.041 --> 00:14:59.520
the, you know, earthquake cycle,
it contains both coseismic and

00:14:59.520 --> 00:15:05.519
postseismic deformation, as shown by
this kind of very simple figure.

00:15:05.519 --> 00:15:08.339
Right after the earthquake, you see –
very often, you see very

00:15:08.339 --> 00:15:13.209
rapid surface deformation.
And if you subtract the interseismic

00:15:13.209 --> 00:15:19.779
deformation, you will get a kind of
postseismic transient signal.

00:15:19.779 --> 00:15:24.490
So postseismic deformation contains
a lot of information about the

00:15:24.490 --> 00:15:30.310
mechanical properties of the Earth.
For example, for, you know, rheology of

00:15:30.310 --> 00:15:36.520
the fault zone and also the bulk rheology
of the – of the crust and lithosphere.

00:15:36.520 --> 00:15:43.640
So, for this purpose, we want to have a –
we want to have kind of tensors

00:15:43.649 --> 00:15:48.681
of the surface deformation.
So initially, we don’t have any GPS or

00:15:48.681 --> 00:15:54.550
any other ground-based measurements.
So what we do is use kind of

00:15:54.550 --> 00:15:58.639
InSAR time series analysis.
So how do we obtain the InSAR

00:15:58.639 --> 00:16:03.050
time series from single –
from interferograms?

00:16:03.050 --> 00:16:08.860
So here’s a slide to show you kind of,
once you have the interferograms made

00:16:08.860 --> 00:16:19.710
between different image acquisitions,
and you can form a kind of time series

00:16:19.710 --> 00:16:22.410
using these interferograms.
So here, each line represents

00:16:22.410 --> 00:16:27.459
one interferogram between
any given two images.

00:16:27.459 --> 00:16:31.209
So if you have a whole network of
interferograms, you can – you can

00:16:31.209 --> 00:16:37.360
generate the time series, or time history,
of the ground deformation.

00:16:38.960 --> 00:16:44.639
This sounds very cool, right?
However, this is not always that easy

00:16:44.639 --> 00:16:49.180
for the very low-amplitude deformation
detection, especially on, for example,

00:16:49.180 --> 00:16:54.869
for interseismic or for
postseismic deformation studies.

00:16:54.869 --> 00:16:58.670
Here I put some – this figure,
put some clouds here.

00:16:58.670 --> 00:17:02.610
So all of the magnetic waves
propagating through the clouds will

00:17:02.610 --> 00:17:09.070
be delayed. So that causes some losses
in the measurement for InSAR.

00:17:09.070 --> 00:17:16.419
This can be up to many centimeters,
as shown by this figure.

00:17:16.420 --> 00:17:21.300
I kind of did – I put out –
I plot out other surface deformation –

00:17:21.300 --> 00:17:24.160
other interferograms
between a very short period.

00:17:24.169 --> 00:17:29.360
So each panels represents the
interferogram between 12 or 24 days.

00:17:29.360 --> 00:17:33.650
So, during this time period, you
should expect almost no deformation.

00:17:33.650 --> 00:17:37.490
However, what you see here is that,
for some interferograms, it has

00:17:37.490 --> 00:17:42.780
very strong range change, which
can be up to 30 centimeters,

00:17:42.780 --> 00:17:49.470
as you can see the color bar here.
This is a area in south Pakistan.

00:17:49.470 --> 00:17:56.540
You know, this is a very typical
structure of atmospheric delays.

00:17:56.540 --> 00:18:01.970
So the message – take-home message
is that InSAR is often contaminated

00:18:01.970 --> 00:18:03.800
by strong atmosphere delays.

00:18:03.800 --> 00:18:09.020
We have to find a way to deal with
that for the small signal detection.

00:18:09.020 --> 00:18:16.420
So how do we do that? Very often,
we just, you know, add one more

00:18:16.420 --> 00:18:21.340
extra term, assuming the deformation,
you know, follows certain function.

00:18:21.340 --> 00:18:28.740
Or, you know, kind of add
some temporal smoothing.

00:18:28.740 --> 00:18:32.670
Which works fine for some cases.
If you know the deformation –

00:18:32.670 --> 00:18:35.720
you know, the evolution –
the temporal function of the

00:18:35.720 --> 00:18:39.460
deformation, but most of the time,
we don’t know for sure.

00:18:39.460 --> 00:18:46.140
So what we did is kind of – called
estimating the atmospheric delays using

00:18:46.140 --> 00:18:52.260
a method called common-scene stacking.
The way this method works is that they

00:18:52.260 --> 00:18:58.920
take advantage of the fact that each
InSAR acquisition – each image

00:18:58.920 --> 00:19:02.400
acquisition contains the same –
each interferogram contains the

00:19:02.400 --> 00:19:09.660
same contribution from –
of the atmospheric delays

00:19:09.660 --> 00:19:17.940
that show the same thing.
So, in this equation, delta represents

00:19:17.940 --> 00:19:22.370
the deformation between any two –
any interferogram – any given

00:19:22.370 --> 00:19:29.549
interferogram, and tau represents the
atmospheric delays in that given scene.

00:19:29.549 --> 00:19:38.170
So, as you can see – so the observed
phase of the interferogram 1-2 is

00:19:38.170 --> 00:19:44.010
equal to the deformation plus the
atmospheric delay from 1 and 2.

00:19:44.010 --> 00:19:49.720
And similarly, you can form an
equation for the interferogram 2 to 3.

00:19:49.720 --> 00:19:54.360
If you subtract – if you subtract these
two interferograms, you will get –

00:19:54.360 --> 00:19:57.950
oh, of course, the delta-tau is the
same because the atmosphere delay

00:19:57.950 --> 00:20:00.680
in the second acquisition is the same.
So when you subtract these

00:20:00.680 --> 00:20:06.400
interferograms, you will essentially
kind of remove the deformation

00:20:06.410 --> 00:20:16.020
between 1-2 and 2-3. And what’s left
is the atmosphere delay from 1 and 3.

00:20:16.020 --> 00:20:19.330
If you assume atmosphere delay
from the acquisitions are more or less

00:20:19.330 --> 00:20:25.610
random in time, then you can estimate
the atmospheric delay for a given scene.

00:20:25.610 --> 00:20:28.760
And this is just a case
for two interferograms.

00:20:28.760 --> 00:20:31.730
If you have many interferograms,
you can – you can improve

00:20:31.730 --> 00:20:36.900
this accuracy significantly.
So next, I’m going to show example.

00:20:36.900 --> 00:20:42.100
I know you probably are confused
by this, and, you know, don’t worry.

00:20:42.100 --> 00:20:47.120
I’m going to show example how that –
how well this method works.

00:20:47.120 --> 00:20:49.700
All right. So this is
a time series of the

00:20:49.700 --> 00:20:54.580
postseismic deformation
one year after the Iran earthquake.

00:20:54.580 --> 00:21:00.560
As you can see – sorry – the mouse.
Where’s the mouse?

00:21:02.340 --> 00:21:06.500
Here’s the coseismic rupture.
In this interferogram – in this

00:21:06.500 --> 00:21:08.960
time series, I didn’t correct
for atmosphere delays.

00:21:08.960 --> 00:21:17.840
So you’re going to see a lot of kind of
randomly distributed red or blue.

00:21:17.840 --> 00:21:22.640
So – by the way, in my –
in all my figures, right represents

00:21:22.640 --> 00:21:25.900
the motion toward
the satellite, or uplift.

00:21:25.900 --> 00:21:29.230
And blue represents the
motion away from the satellite.

00:21:29.230 --> 00:21:35.059
So, as you can see – already you
can see, from this figure, in all the

00:21:35.059 --> 00:21:41.890
acquisitions, the deformation is
characterized by some kind of uplift

00:21:41.890 --> 00:21:43.810
west of the rupture,
or up-dip of the rupture.

00:21:43.810 --> 00:21:48.400
However, in some cases, the, you know,
atmosphere delay is so strong

00:21:48.400 --> 00:21:54.220
that you cannot see the signal.
Or it’s completely obstructed.

00:21:54.220 --> 00:21:57.280
And some other times,
you see a much smaller signal.

00:21:57.289 --> 00:22:02.790
This is – this is a feature of atmosphere
delay if you don’t correct for it.

00:22:02.790 --> 00:22:08.820
So let’s compare what we got
using the method I just introduced.

00:22:08.820 --> 00:22:14.640
Boom. It’s so clean. You know, it’s
completely – most of atmospheres just,

00:22:14.640 --> 00:22:19.640
you know, magically disappeared.
So, again, if you see – if you compare

00:22:19.640 --> 00:22:25.020
this, the coseismic rupture,
now you can see a very coherent

00:22:25.020 --> 00:22:29.710
deformation signal at both up-dip
and down-dip of the rupture.

00:22:29.710 --> 00:22:36.970
So, again, this small band is
corresponding to the down-dip.

00:22:36.970 --> 00:22:40.480
And this very broad red zone
is the up-dip region.

00:22:40.480 --> 00:22:44.040
So you see a very
significant change.

00:22:44.740 --> 00:22:48.660
All right, so this is
just one track example.

00:22:48.660 --> 00:22:51.220
By the way, this is a
automatic processing system.

00:22:51.220 --> 00:22:54.120
Because the Sentinel – as I said,
Sentinel data is completely free.

00:22:54.120 --> 00:22:57.160
So you can get the
data with no problems.

00:22:57.160 --> 00:22:59.800
You can easily get the data.
So I have a automatic processing

00:22:59.809 --> 00:23:04.570
system, so once the new data available,
my system goes to process the data.

00:23:04.570 --> 00:23:07.750
Very automatically.
You don’t have to touch anything.

00:23:07.750 --> 00:23:11.220
And this works for anywhere.
So you give a region, my system can do

00:23:11.220 --> 00:23:15.760
it, including the time series – including
the atmosphere delay corrections.

00:23:16.590 --> 00:23:18.880
So this just
an example.

00:23:18.880 --> 00:23:27.020
So I have a movie to show –
to show this.

00:23:29.320 --> 00:23:32.940
So you can see – what you see in the –
in the left is the – kind of the map view

00:23:32.950 --> 00:23:36.610
of the ground deformation due to
the postseismic deformation.

00:23:36.610 --> 00:23:41.360
And what you see in the left –
to the left is the –

00:23:41.360 --> 00:23:46.210
is the surface deformation at up-dip –
at the point up-dip of the rupture

00:23:46.210 --> 00:23:51.700
and also a point at the down-dip.
And also, I plot out the evolution

00:23:51.700 --> 00:23:58.140
of the aftershocks within this
box and also within this box.

00:23:58.140 --> 00:24:03.900
As you can see, at both up-dip and –
at both up-dip and down-dip,

00:24:03.900 --> 00:24:07.490
the surface deformation seemed
to involve in kind of, you know,

00:24:07.490 --> 00:24:10.180
similar fashion
as the aftershocks.

00:24:10.180 --> 00:24:17.260
So, also, in the time series analysis,
I didn’t apply any temporal smoothing

00:24:17.260 --> 00:24:22.669
or any function fitting.
This is just out of data after the

00:24:22.669 --> 00:24:27.950
correction of the atmosphere delays.
And, to me, the time series looks as

00:24:27.950 --> 00:24:32.700
good as some of the GPS time series,
probably. But this is from InSAR.

00:24:34.760 --> 00:24:38.380
All right, so this is just
data from one track.

00:24:38.390 --> 00:24:42.270
And I have observation
from all four tracks.

00:24:42.270 --> 00:24:47.299
So, again, the contours
are the coseismic rupture.

00:24:47.299 --> 00:24:52.659
And, as you can see, the two ascending
tracks are characterized by the range

00:24:52.659 --> 00:25:02.010
decrease at the up-dip of the rupture and
also some range change at the down-dip.

00:25:02.010 --> 00:25:05.679
And for the two descending tracks,
the observation – the signal

00:25:05.679 --> 00:25:09.430
is kind of different.
It’s primarily dominated by range

00:25:09.430 --> 00:25:14.289
decrease – range increase at the –
right beneath the coseismic rupture.

00:25:14.289 --> 00:25:21.150
The different signs of near-site
deformation from ascending or

00:25:21.150 --> 00:25:25.090
descending tracks indicate
that the postseismic deformation

00:25:25.090 --> 00:25:27.730
of this earthquake is primarily –
not primarily –

00:25:27.730 --> 00:25:31.240
has a strong component
of horizontal motion.

00:25:32.080 --> 00:25:40.140
And I also point out there are
two very strong aftershocks.

00:25:40.150 --> 00:25:47.430
One is October 25th.
And also there is on November 25th.

00:25:47.430 --> 00:25:50.960
So they are all
captured by the InSAR.

00:25:50.960 --> 00:25:53.600
By the postseismic
InSAR analysis.

00:25:54.870 --> 00:25:56.820
So this is some –
this is just observations.

00:25:56.830 --> 00:26:00.850
What we want to – what we
want to do is to kind of explore

00:26:00.850 --> 00:26:04.400
what mechanism are causing
this surface deformation.

00:26:06.380 --> 00:26:10.670
Commonly considered models
for the postseismic deformation

00:26:10.670 --> 00:26:16.190
includes three mechanisms.
One is viscoelastic relaxation,

00:26:16.190 --> 00:26:21.470
which is primarily due to the increase of
temperature and pressure with depth.

00:26:21.470 --> 00:26:26.620
So the material behaves
ductily under stress.

00:26:26.620 --> 00:26:29.610
Under stress change –
so when you have a earthquake,

00:26:29.610 --> 00:26:35.760
the stress will be
relieved slowly with time.

00:26:35.760 --> 00:26:38.700
And the second mechanism
is poroelastic rebound.

00:26:38.700 --> 00:26:45.100
This is due to the change of the pore
pressure due to the coseismic stress.

00:26:45.100 --> 00:26:49.920
And the third mechanism is
called afterslip, which is basically

00:26:49.929 --> 00:26:57.060
a continuation of the slip on the fault.
So for this study, I will only consider

00:26:57.060 --> 00:27:00.799
afterslip, and you will see this
mechanism works so well.

00:27:00.800 --> 00:27:04.420
So you don’t probably have to
consider the other two mechanisms.

00:27:06.080 --> 00:27:12.920
So we want to kind of explore where the
afterslip is occurring and to what extent

00:27:12.920 --> 00:27:16.660
the afterslip can be
explained by current existing –

00:27:16.660 --> 00:27:19.920
kinds of existing
mechanical models.

00:27:20.720 --> 00:27:27.040
So to start with, we have to –
can explore the geometry because

00:27:27.059 --> 00:27:29.850
we don’t know for sure what –
where the afterslip is occurring,

00:27:29.850 --> 00:27:33.400
so we have to explore the
geometry of the afterslip.

00:27:33.400 --> 00:27:39.260
So, in this – in this study, I can explore
the up-dip fault – up-dip region of the

00:27:39.260 --> 00:27:46.860
fault, and I allow for the dip angle
of the shallow afterslip to change.

00:27:46.860 --> 00:27:55.799
And, as you can see, in this –
in this figure shows the misfit of the –

00:27:55.799 --> 00:28:00.230
misfit of the model as a function
of different dip angles of the

00:28:00.230 --> 00:28:04.090
shallow afterslip plane.
So the dashed line shows the

00:28:04.090 --> 00:28:10.360
dip angle of the coseismic rupture.
As you can see, it’s very clear that, for –

00:28:10.360 --> 00:28:15.340
if you want to make a model that
fits the data, the dip angle of the

00:28:15.350 --> 00:28:20.950
shallow part has to be small.
The dip angle – the model prefers

00:28:20.950 --> 00:28:25.870
a very small dip angle for the –
for the up-dip region.

00:28:25.870 --> 00:28:30.800
So and the geometry
would be like this.

00:28:32.260 --> 00:28:36.640
All right, so this is the geometry.
If you – the thing that you can do,

00:28:36.650 --> 00:28:41.090
the inversion for the slip distribution of
the afterslip, and that’s what you get.

00:28:41.090 --> 00:28:49.549
So very clearly, the afterslip seems
to occur at the up-dip – up-dip of the

00:28:49.549 --> 00:28:54.350
fault and down-dip – also with some
contribution from the down-dip.

00:28:54.350 --> 00:28:58.540
And right beneath the coseismic
rupture, there was almost no afterslip.

00:28:58.540 --> 00:29:06.220
This is a very clear signal.
I didn’t mask any slip from this plot.

00:29:06.220 --> 00:29:09.130
This is just
from the inversion.

00:29:09.130 --> 00:29:14.700
So if you plot out the seismicity –
the aftershocks, you see that most of

00:29:14.700 --> 00:29:24.880
aftershocks also tend to occur at the up-
dip region of the fault – of the rupture.

00:29:25.800 --> 00:29:29.120
All right, this is kind of –
kinematic inversion.

00:29:29.130 --> 00:29:34.310
This is just to invert for the data and
doesn’t have any physics involved yet.

00:29:34.310 --> 00:29:43.990
So because of the – as I said,
atmosphere delay is a very significant

00:29:43.990 --> 00:29:48.679
problem because there has been
one published study, which didn’t

00:29:48.680 --> 00:29:53.640
capture the down-dip afterslip at all.
But our model has very clear

00:29:53.640 --> 00:29:59.120
down-dip for afterslip.
All right, so that’s – next I’m going to

00:29:59.120 --> 00:30:01.960
move to the modeling
of the afterslip using kind of

00:30:01.970 --> 00:30:06.990
very simple rate-and-state
friction laws.

00:30:06.990 --> 00:30:12.370
So I assumed the evolution
of the afterslip follows the

00:30:12.370 --> 00:30:19.400
rate-strengthening law.
So then the slip rate at onset of the

00:30:19.400 --> 00:30:22.880
afterslip is given
by this equation.

00:30:22.880 --> 00:30:27.130
And there are two parameters
that we have to explore.

00:30:27.130 --> 00:30:30.070
One is the V-zero – the reference
slip rate – and also the a-sigma.

00:30:30.070 --> 00:30:35.909
Because the delta-tau can be calculated
from the coseismic slip model we just

00:30:35.909 --> 00:30:41.360
have derived. So there are two
parameters we have to explore.

00:30:45.460 --> 00:30:50.100
From both – from the observations –
from the postseismic observations,

00:30:50.110 --> 00:30:55.880
we see – we seem to see that the
up-dip – at the down-dip,

00:30:55.880 --> 00:30:59.320
the surface deformation
seems to decay faster.

00:30:59.320 --> 00:31:04.060
And this is also the case for the
evolution of the aftershocks.

00:31:05.880 --> 00:31:10.440
Also, if you compare the magnitude
of the – of the afterslip, the up-dip

00:31:10.440 --> 00:31:15.140
afterslip seems to be much larger
compared to the down-dip.

00:31:15.140 --> 00:31:18.880
So, you know, the different
relaxation times and different

00:31:18.880 --> 00:31:24.440
magnitude of afterslip, do they
suggest different fault rheologies?

00:31:25.580 --> 00:31:32.080
Maybe. So to answer this,
we can cut the model into three parts.

00:31:32.090 --> 00:31:36.059
At the up-dip, we assume
one set of friction, and down-dip,

00:31:36.059 --> 00:31:40.730
we assume another set for friction.
In between, we assume it’s

00:31:40.730 --> 00:31:45.920
velocity-weakening, so it’s doesn’t move
during the postseismic time period.

00:31:45.920 --> 00:31:52.700
All right, so here’s the result.
So we solved this problem in

00:31:52.700 --> 00:31:56.100
a Bayesian – again, a Bayesian
inversion framework.

00:31:56.100 --> 00:32:02.750
And, as you can see, after a few
hundred iterations, all the parameters

00:32:02.750 --> 00:32:11.300
converge to a very narrow band.
And, you know, don’t know –

00:32:11.300 --> 00:32:16.510
you know, wondering what
those parameters are indicating.

00:32:16.510 --> 00:32:26.470
So the point is that we found a very
different – very different, for friction,

00:32:26.470 --> 00:32:31.049
up-dip and down-dip.
As you can see, for the up-dip region,

00:32:31.049 --> 00:32:39.000
the – for friction, the a-sigma is –
oh, it’s 2.7 megapascals.

00:32:39.000 --> 00:32:48.860
And this is – this is very different
from the 0.07 for the down-dip region.

00:32:48.870 --> 00:32:56.890
So, and here’s the comparison
between the kinematic inversion

00:32:56.890 --> 00:33:02.540
and the stress-driven afterslip models.
As you can see, both magnitude and

00:33:02.540 --> 00:33:07.360
the spatial pattern,
they match pretty well.

00:33:07.360 --> 00:33:13.080
So that indicates the model
does a very good job.

00:33:14.560 --> 00:33:19.680
And if you compare this – if you
compare the surface deformation,

00:33:19.680 --> 00:33:23.510
on the left, it’s the observation
from different tracks

00:33:23.510 --> 00:33:28.669
And this column is the model
prediction from the kinematic inversion.

00:33:28.669 --> 00:33:32.590
So – and on third column,
it’s the prediction from the

00:33:32.590 --> 00:33:36.380
rate-strengthening model.
As you can see, the rate-strengthening

00:33:36.380 --> 00:33:42.080
model – rate-strengthening model
predicts the data fitting as good as

00:33:42.090 --> 00:33:46.840
the kinematic inversion,
except at a few places,

00:33:46.840 --> 00:33:49.980
probably when you
are close to the rupture.

00:33:51.300 --> 00:33:54.320
That’s probably due to the fault
friction heterogeneity because

00:33:54.320 --> 00:34:01.280
we assume the same friction
at up-dip – at up-dip region.

00:34:01.940 --> 00:34:06.140
You can also look at the
data fitting for the time series.

00:34:06.820 --> 00:34:17.280
So here, the red curve represents the
time series at the down-dip region.

00:34:17.290 --> 00:34:20.750
And the gray curves are for the
surface deformation time series

00:34:20.750 --> 00:34:23.899
at the up-dip region.
As you can see, both – the model

00:34:23.899 --> 00:34:28.601
fits the data – both the temporal
evolution and the surface deformation

00:34:28.601 --> 00:34:34.740
pattern are predicted very well by this
very simple rate-and-state model.

00:34:38.180 --> 00:34:42.320
And this is – this is
for the ascending track.

00:34:42.329 --> 00:34:45.310
You can also see – for the descending
track, I did – because I didn’t show you

00:34:45.310 --> 00:34:48.330
time series for the descending track.
As you can see, the descending tracks

00:34:48.330 --> 00:34:53.180
also has a very nice decaying
feature from the InSAR time series.

00:34:53.180 --> 00:34:55.250
And, again,
the models fits the

00:34:55.250 --> 00:34:59.220
descending track data
also reasonably well.

00:35:02.820 --> 00:35:08.320
I guess I’m going to skip this slide, but –
to come to this slide because we also

00:35:08.320 --> 00:35:12.140
tried to kind of force the model
to have different – to have the

00:35:12.140 --> 00:35:19.390
same fault friction. And what you find
is that there was no way to predict

00:35:19.390 --> 00:35:24.200
the data – to fit the data well.
If you assume the friction is

00:35:24.200 --> 00:35:28.980
homogeneous – is uniform everywhere,
then there is no way to fit the data.

00:35:30.280 --> 00:35:36.900
So I guess the Iran earthquake is
a very rare case where you see

00:35:36.910 --> 00:35:41.630
clear separation between
postseismic and afterslip.

00:35:41.630 --> 00:35:48.830
So this seems to fit the very standard
standard rate-and-state model well.

00:35:48.830 --> 00:35:52.960
Because in this – in this framework,
you expect that afterslip tends to

00:35:52.960 --> 00:36:01.420
occur at the – surrounding
of coseismic slip patch.

00:36:02.580 --> 00:36:05.480
And also there’s a model
to show, you know,

00:36:05.480 --> 00:36:09.260
what you should expect
for the afterslip.

00:36:10.280 --> 00:36:16.540
The right patch is the coseismic
asperities, and the right – yellow is the –

00:36:16.540 --> 00:36:23.040
for the kind of region you should
expect afterslip or interseismic creep.

00:36:23.760 --> 00:36:28.099
However, this is not – as I said,
this is not – even though this model

00:36:28.099 --> 00:36:33.790
predicts this behavior, but it’s not
always the case, especially for –

00:36:33.790 --> 00:36:41.140
for example, for the – for the
earthquakes along the Himalaya,

00:36:41.140 --> 00:36:48.099
what you find is that the –
you don’t see any shallow afterslip

00:36:48.100 --> 00:36:51.460
from the 2015
Gorkha earthquake.

00:36:52.530 --> 00:37:00.760
So what’s causing the kind of
very unique feature that you see

00:37:00.770 --> 00:37:06.630
a clear separation between
the coseismic and the afterslip?

00:37:06.630 --> 00:37:11.300
This is some question
we are still trying to explore.

00:37:11.300 --> 00:37:15.700
But I think it could be related to the
kind of lithology across the Zagros.

00:37:15.700 --> 00:37:21.680
Because – you know, across much
of the Zagros, there is a very thick

00:37:21.680 --> 00:37:24.790
sedimentary cover. And the base
of the – at the base of the

00:37:24.790 --> 00:37:33.640
sedimentary cover, there is – there is –
it’s been suggested there is a

00:37:33.640 --> 00:37:40.920
very mechanically weak salt layer,
which may accommodate kind of the

00:37:40.920 --> 00:37:48.740
shear – the stress change
so that the – after the stress change,

00:37:48.740 --> 00:37:54.869
the stress can be easily
relieved across this salt layer.

00:37:54.869 --> 00:38:05.540
So with that, I think I will stop here.
And so the conclusion is that the

00:38:05.540 --> 00:38:11.140
dense catalog of Sentinel InSAR
provides a very unprecedented

00:38:11.140 --> 00:38:15.700
opportunity to study the fault zone
rheology in greater details.

00:38:15.700 --> 00:38:22.720
And the 2017 earthquake occurred –
ruptured almost – a nearly

00:38:22.720 --> 00:38:25.660
north-south-trending
fault in the basement.

00:38:25.660 --> 00:38:30.460
And the postseismic deformation
one year following the main shock was

00:38:30.460 --> 00:38:36.940
dominated by afterslip both up-dip
and down-dip of the coseismic rupture.

00:38:39.000 --> 00:38:43.800
All right, so with that, I will take
questions or comments or suggestions.

00:38:43.800 --> 00:38:44.860
Thanks.

00:38:44.860 --> 00:38:48.560
[Applause]

00:38:48.560 --> 00:38:52.460
- All right. Thank you very much.
Do we have any questions?

00:38:56.260 --> 00:39:00.050
- Thanks for a really interesting talk.
I had a couple questions.

00:39:00.050 --> 00:39:04.520
When you showed the
interferograms for the coseismic case –

00:39:04.520 --> 00:39:09.140
or, the coseismic time period,
it looked like they spanned something

00:39:09.150 --> 00:39:11.630
from about a week to 10 days.
- Correct.

00:39:11.630 --> 00:39:16.300
- So it seems like probably there is
some immediate postseismic that’s

00:39:16.300 --> 00:39:17.960
getting mapped into coseismic.
- Right.

00:39:17.960 --> 00:39:20.240
- So I wondered if you could
just comment on how that

00:39:20.240 --> 00:39:24.180
influences your results.
- That’s a – that’s a good question.

00:39:24.180 --> 00:39:28.470
For example, for this – I didn’t show
this slide, but anyway – so I’ll talk this –

00:39:28.470 --> 00:39:34.310
so, right. So all the coseismic
interferograms span probably

00:39:34.310 --> 00:39:40.900
a few days after the earthquake.
And, yes, it definitely contained

00:39:40.900 --> 00:39:46.329
some early postseismic deformation.
With this model, I guess I can compute

00:39:46.329 --> 00:39:51.440
how much deformation is being
produced during this short time period,

00:39:51.440 --> 00:39:55.020
but I haven’t done that yet.
You know, from the model,

00:39:55.020 --> 00:39:58.900
it seems to be significant from
the rate-and-state model that I got.

00:39:59.780 --> 00:40:06.040
- The other question I had is to do
with the atmospheric corrections.

00:40:06.040 --> 00:40:09.920
And you showed a slide – I think it
was actually an example from Pakistan,

00:40:09.920 --> 00:40:18.339
but it – I think it kind of showed
that the atmospheric signals can also

00:40:18.339 --> 00:40:23.110
be correlated with topography,
which itself would be correlated

00:40:23.110 --> 00:40:26.470
with the fault structure, probably.
So I’m just interested in

00:40:26.470 --> 00:40:31.720
how you tease those apart.
- Oh. I guess this slide.

00:40:34.300 --> 00:40:38.410
So for this region, actually the
topography doesn’t change

00:40:38.410 --> 00:40:43.369
very dramatically. So all you see
is not primarily due to topography.

00:40:43.369 --> 00:40:46.880
Everything is due to
precipitation for this case.

00:40:46.880 --> 00:40:51.650
But in my code – in my method,
we have already considered this part.

00:40:51.650 --> 00:40:57.560
So we have two components.
One is to correct for the – kind of

00:40:57.560 --> 00:41:02.319
the generally stratified
component of atmosphere delay.

00:41:02.319 --> 00:41:05.599
And the rest is done by the
common-scene stacking,

00:41:05.599 --> 00:41:09.290
which assumes that atmosphere
delay is more or less random in time.

00:41:09.290 --> 00:41:13.590
So, yes, we are considering that.
But, for this case, all you see is

00:41:13.590 --> 00:41:19.540
pretty much due to the very turbulent
component of atmosphere delay.

00:41:22.020 --> 00:41:25.620
- Could you – could you go back to the –
you’re kinematic inversion of afterslip

00:41:25.620 --> 00:41:32.160
slide, where you show the geometry
of the – of the lower –

00:41:32.160 --> 00:41:36.520
or, the second dislocation?
- The afterslip?

00:41:36.520 --> 00:41:40.140
- Yeah. The kinematic
inversion of afterslip.

00:41:43.260 --> 00:41:45.640
- All right.
- Or the one in cross-section where

00:41:45.640 --> 00:41:49.300
you show the – your favored plane.
- Oh, okay, okay.

00:41:49.300 --> 00:41:51.460
- Yeah.
- Yeah.

00:41:54.400 --> 00:41:55.640
- Right.
- Right.

00:41:55.650 --> 00:41:57.819
- And if you could go forward.
Because you show the plane –

00:41:57.819 --> 00:42:01.880
the shallow plane that’s
shallower – yeah, right.

00:42:01.880 --> 00:42:08.540
So I’m curious about that.
So that’s your best-fitting shallow plane.

00:42:08.540 --> 00:42:10.740
- Right.
- Family of planes.

00:42:10.740 --> 00:42:15.800
So the question is, do you really
believe that there’s another fault plane

00:42:15.800 --> 00:42:21.160
that the shallow afterslip is occurring
on, which would make it a ramp/flat

00:42:21.160 --> 00:42:24.340
system, or a sort of – yeah,
a sort of ramp/flat system?

00:42:24.349 --> 00:42:26.690
And then the question is, if you believe
that, have you looked at the topography,

00:42:26.690 --> 00:42:29.250
and is there long-term
topography over that bend?

00:42:29.250 --> 00:42:34.369
- Very good question.
So – but first let me clarify this.

00:42:34.369 --> 00:42:37.290
Are you saying there’s –
if there’s a better kind of ramp …

00:42:37.290 --> 00:42:40.870
- No. No. I’m saying at the
bend right between the two.

00:42:40.870 --> 00:42:43.570
- Oh, right between the two.
- If that’s a long-term feature,

00:42:43.570 --> 00:42:45.029
there ought to be
some topography there.

00:42:45.029 --> 00:42:49.420
- Oh, for sure, there’s topography.
Yes. Yes, there is topography.

00:42:49.420 --> 00:42:54.690
And, in fact, this has been proposed
as a mountain frontal fracture,

00:42:54.690 --> 00:42:59.160
which is kind of – kind of sudden
change of the topography here.

00:42:59.160 --> 00:43:03.780
But, as I said, there is no
geologically mapped faults

00:43:03.780 --> 00:43:06.680
along this north-south-
trending direction.

00:43:07.420 --> 00:43:08.400
- Right.
- Yeah.

00:43:08.400 --> 00:43:10.691
- But …
- Definitely the earthquake

00:43:10.691 --> 00:43:13.221
produces some uplift in the –
in the mountain range.

00:43:13.221 --> 00:43:16.089
- Right. But you have oblique slip
on your fault plane, which would

00:43:16.089 --> 00:43:19.080
be pretty much perpendicular
to the trends of the folds there.

00:43:19.080 --> 00:43:22.240
- Right.
- So the structure – if it’s going over

00:43:22.250 --> 00:43:28.280
that bend, ought to – if that happens
repeatedly, there ought to be

00:43:28.280 --> 00:43:35.360
a structure there, over the long run.
- I didn’t – I didn’t observe any

00:43:35.360 --> 00:43:40.839
deformation kind of associated
with what you said about kind of ramp

00:43:40.840 --> 00:43:47.840
and a connection at the junction.
But we do see some secondary faults –

00:43:47.840 --> 00:43:54.099
secondary deformation.
As you see from this interferogram,

00:43:54.099 --> 00:43:59.869
there seems to be a very –
a very large offset further to the south.

00:43:59.869 --> 00:44:01.920
But, you know, in this region,
we don’t see it.

00:44:01.920 --> 00:44:07.099
- But, so just to clarify, then.
So you believe that there’s

00:44:07.099 --> 00:44:12.900
a dipping fault – coseismic dipping
fault and then another upper fault plane

00:44:12.900 --> 00:44:16.710
that’s less steeply dipping that’s …
- Right.

00:44:16.710 --> 00:44:18.240
- … more shallow.
- Right.

00:44:18.240 --> 00:44:20.630
- Upon which the shallow
afterslip is occurring.

00:44:20.630 --> 00:44:22.940
- Right, right. Right, right, right.
- Okay, thanks.

00:44:22.940 --> 00:44:28.180
- At least from the geodetic perspective,
we don’t see any evidence of

00:44:28.180 --> 00:44:33.710
continuation of a ramp connecting
to the – to the main shock.

00:44:33.710 --> 00:44:36.880
- So it’s kind of interesting then
because then the coseismic –

00:44:36.880 --> 00:44:40.600
the up-dip portion of the coseismic
rupture stopped at that bend.

00:44:40.619 --> 00:44:42.359
- Right.
- With that interpretation.

00:44:42.360 --> 00:44:48.720
- So we kind of interpret that as
due to the mechanical weak layer.

00:44:48.720 --> 00:44:50.120
- The layer.
- Because the earthquake

00:44:50.120 --> 00:44:53.440
nucleating in the basement.
But because of the layer,

00:44:53.440 --> 00:44:58.859
kind of decouples the deformation
from bottom, so it doesn’t probably

00:44:58.860 --> 00:45:01.520
get all the way to the –
to the shallow depth.

00:45:02.220 --> 00:45:05.320
- Thanks.
- Got more questions?

00:45:06.560 --> 00:45:08.740
Were you thinking about
asking a question?

00:45:09.140 --> 00:45:15.680
- If this is really a weak decoupled layer,
wouldn’t you see expression of that in

00:45:15.680 --> 00:45:30.120
the long-term secular deformation field?
- I think, from what I write, the GPS –

00:45:30.120 --> 00:45:35.620
kind of the – the seismic – if you take
account the seismic moment release,

00:45:35.620 --> 00:45:41.230
it doesn’t account all the deformation –
all the strain that is determined

00:45:41.230 --> 00:45:45.510
from the geodetic measurements.
I guess that there has some –

00:45:45.510 --> 00:45:54.980
some kind of interseismic creep
or shearing across that layer.

00:45:56.060 --> 00:45:58.260
Not sure if I answered
your question, but …

00:46:02.680 --> 00:46:08.180
- Yeah, that – very interesting talk.
I think, in response to Ben’s question,

00:46:08.180 --> 00:46:16.261
it might be nice just to, you know,
draw the topography over your fault –

00:46:16.261 --> 00:46:20.700
your fault structure and put the
topography on the top of that and see

00:46:20.700 --> 00:46:25.280
for yourself as to what, if any,
structures, might be …

00:46:25.280 --> 00:46:27.220
- Right.
- … related to the fault at depth.

00:46:27.220 --> 00:46:32.680
- Mm-hmm.
- And the question I have is, is there any

00:46:32.680 --> 00:46:42.569
surface faulting or surface deformation
that is observable or was mappable the –

00:46:42.569 --> 00:46:46.369
this is kind of a rough place to do –
be in the field.

00:46:46.369 --> 00:46:50.060
- Right.
- But is there any …

00:46:51.049 --> 00:46:54.100
- You know, I had this question, I …
- … thoughts on that front?

00:46:54.109 --> 00:46:57.630
- Right. I asked a colleague, and he said,
well, all the geological faults

00:46:57.630 --> 00:47:02.220
are kind of very rough.
All the locations are pretty bad.

00:47:02.220 --> 00:47:06.440
I don’t know how they did the mapping
here, but it seems to – yeah, as you said,

00:47:06.440 --> 00:47:09.130
it’s difficult to map
the faults in this region.

00:47:09.130 --> 00:47:14.470
Even though it’s a very active place
for oil and gas exploration.

00:47:14.470 --> 00:47:18.220
- Because – well, if you go back to
one of your very first slides,

00:47:18.220 --> 00:47:22.460
it looked like it was kind of
significant ground deformation

00:47:22.460 --> 00:47:28.200
in one of the – one of the slides.
Right up front.

00:47:28.200 --> 00:47:30.220
- Hold a second.

00:47:32.760 --> 00:47:39.060
Because I’m not very used to this
kind of – the presenter mode. [chuckles]

00:47:42.880 --> 00:47:46.570
This one? This one?
- No, no, no.

00:47:46.570 --> 00:47:55.180
A picture of a demolished city and –
I think ground deformation.

00:47:55.180 --> 00:47:57.080
- Oh. Oh, okay, okay.
- Right up front.

00:47:57.080 --> 00:47:59.360
Right at the beginning of your talk.
- This one?

00:47:59.360 --> 00:48:02.660
- Yeah. So what is – what is
all that cracking up …

00:48:02.660 --> 00:48:08.560
- Oh, actually, in this – in this blog,
they proposed as landslides.

00:48:08.560 --> 00:48:11.650
This is some landslides.
- That’s from a landslide.

00:48:11.650 --> 00:48:14.170
- This is from landslide.
This is not the ground – this is

00:48:14.170 --> 00:48:18.740
not the rupture due to earthquake.
This is kind of very shallow feature.

00:48:19.460 --> 00:48:28.280
But I do see some – again, I do see
some kind of triggered slip,

00:48:28.290 --> 00:48:31.290
both coseismic and postseismically,
as you see here.

00:48:31.290 --> 00:48:35.640
In the coseismic interferogram,
you see a very sharp offset.

00:48:35.640 --> 00:48:42.240
And this is corresponding to about
6 centimeters’ offset from the

00:48:42.240 --> 00:48:46.300
InSAR observation – from the
coseismic InSAR observation.

00:48:46.300 --> 00:48:52.520
And coseismically, it seems that
this very small structure

00:48:52.520 --> 00:48:57.000
is still moving with time. As you
can see, the color represents time,

00:48:57.000 --> 00:49:02.620
so with time, it’s still building up
more deformation across the fault.

00:49:04.820 --> 00:49:07.480
[Silence]

00:49:08.080 --> 00:49:13.220
- Any more questions?
All right. If not, let’s give

00:49:13.220 --> 00:49:15.200
our speaker another
round of applause.

00:49:15.200 --> 00:49:19.040
[Applause]

00:49:19.040 --> 00:49:21.740
And if you’d like to join us
for lunch, or if you would like to

00:49:21.740 --> 00:49:23.140
get a spot on the speaker’s …