WEBVTT
Kind: captions
Language: en-US

00:00:00.720 --> 00:00:02.640
I’m going to give you
a quick overview of

00:00:02.640 --> 00:00:06.696
what we did in the last couple
of years during the pandemic.

00:00:06.720 --> 00:00:10.856
Let’s consider two types
of creeping fault.

00:00:10.880 --> 00:00:16.800
The end member on the left is a
creeping fault driven from below.

00:00:16.800 --> 00:00:22.216
Seismicity causes creep
to propagate to the surface.

00:00:22.240 --> 00:00:29.496
The one on the right is a creeping fault
that is driven by anti-symmetric shear.

00:00:29.520 --> 00:00:33.736
And the seismogenic zone
is potentially locked.

00:00:33.760 --> 00:00:39.416
And each creep event propagates to a
shallow depth or to a greater depth but

00:00:39.440 --> 00:00:44.720
eventually loads the seismogenic zone.
So this is potentially more interesting

00:00:44.720 --> 00:00:48.880
than this kind of creep,
which is interesting,

00:00:48.880 --> 00:00:53.896
but it only tells you about
the processes after seismicity.

00:00:53.920 --> 00:00:56.960
Here’s an example of useful
data from creepmeter.

00:00:56.960 --> 00:01:02.536
In fact, there was an earthquake
in San Juan Bautista in 1998,

00:01:02.560 --> 00:01:06.080
surrounded by a cluster of creepmeters.
And, if you sum three of those

00:01:06.080 --> 00:01:09.840
creepmeters and detrend the
summed series, you finish up with

00:01:09.840 --> 00:01:15.520
this graph on the top, which shows
rather clearly that there was a massive

00:01:15.520 --> 00:01:20.560
retardation on the creep record
preceding that earthquake.

00:01:20.560 --> 00:01:25.920
And, of course, this is a process
that will occur again and again

00:01:25.920 --> 00:01:29.520
near this triple junction.
So, in principle, if we could

00:01:29.520 --> 00:01:34.616
understand this, we have a way of
forecasting moderate seismicity.

00:01:34.640 --> 00:01:37.840
On the Hayward Fault,
there are five creepmeters.

00:01:37.840 --> 00:01:42.960
We’d like greater density, especially
the absence of creepmeters on the

00:01:42.960 --> 00:01:47.120
Calaveras Fault is a bit of a problem
at the moment, but what you can see

00:01:47.120 --> 00:01:54.320
is that, near Berkeley, two creepmeters
show a concave-up distribution of slip,

00:01:54.320 --> 00:01:59.656
which implies acceleration
starting around about 2007.

00:01:59.680 --> 00:02:04.560
At the same time, there was a
deceleration in creep rate at Palisades

00:02:04.560 --> 00:02:12.320
in Hayward, and this creepmeter
to the south near Fremont has sort of

00:02:12.320 --> 00:02:16.720
decadal variation in creep,
which could be due to

00:02:16.720 --> 00:02:18.240
interaction with
the Calaveras Fault.

00:02:18.240 --> 00:02:20.960
Unfortunately we don’t have
the creepmeters to tell us that.

00:02:20.960 --> 00:02:25.816
And the alignment arrays, of course,
are much too noisy.

00:02:25.840 --> 00:02:29.496
Let’s look at the central
creeping zone briefly.

00:02:29.520 --> 00:02:35.040
Here are data from Melendy
Ranch, XMR, and you can

00:02:35.040 --> 00:02:38.160
see that there are large numbers
of creep events – these grays things,

00:02:38.160 --> 00:02:41.920
and a smaller number of larger ones.
In fact, there’s a bimodal distribution

00:02:41.920 --> 00:02:46.080
of small ones and big ones.
And, if you sum the slip from just the

00:02:46.080 --> 00:02:50.560
small ones, you get a large creep rate,
from the big ones, a smaller creep rate,

00:02:50.560 --> 00:02:56.080
and you can plot it in a graph like
this showing what is a proxy

00:02:56.080 --> 00:02:59.680
for slip at depth.
And I think you can see

00:02:59.680 --> 00:03:05.520
fairly clearly that the rapid slip
near the surface has resulted in

00:03:05.520 --> 00:03:10.216
75% of the slip deficit being
removed, but leaving 25%.

00:03:10.240 --> 00:03:13.440
But, at deeper depths, somewhere
between 3 and 5 kilometers,

00:03:13.440 --> 00:03:16.320
we may have
a 50% slip deficit.

00:03:16.320 --> 00:03:23.160
Here, again, here are creep data that
are telling us about seismic hazards.

00:03:23.600 --> 00:03:26.080
John Langbein did this
for the entire creeping zone.

00:03:26.080 --> 00:03:29.200
You can’t see the plots,
but let me explain what you

00:03:29.200 --> 00:03:32.776
could do with
this kind of data.

00:03:32.800 --> 00:03:37.440
Middle Mountain data are
shown without editing here.

00:03:37.440 --> 00:03:43.360
It’s a fairly steady linear creep rate
up to the Parkfield earthquake

00:03:43.360 --> 00:03:47.600
at about 16 millimeters a year.
If you only look at the creep events,

00:03:47.600 --> 00:03:52.376
you get a lower creep rate,
6.6 millimeters a year,

00:03:52.400 --> 00:03:57.256
and if you detrend those two data,
you can see fairly clearly that the

00:03:57.280 --> 00:04:04.000
creep rate from creep events is far
less contaminated by surface noise.

00:04:04.000 --> 00:04:08.720
In other words, one can clean up the
data, suppress the noise considerably,

00:04:08.720 --> 00:04:12.160
by just looking at creep events.
Another way of looking at these data

00:04:12.160 --> 00:04:16.160
is to look at the long-term trends.
And they’re emphasized by these

00:04:16.160 --> 00:04:22.136
black lines, where you can actually
see changes of rate influenced

00:04:22.160 --> 00:04:27.280
at the time of regional earthquakes.
For example, the Loma Prieta

00:04:27.280 --> 00:04:31.040
earthquake caused a significant change
in rate in the northern creepmeters.

00:04:31.040 --> 00:04:34.456
The Parkfield earthquake, of course,
influenced most of them.

00:04:34.480 --> 00:04:37.520
But we can think of
the creeping zone as

00:04:37.520 --> 00:04:42.536
a very sensitive strainmeter
to regional stress changes.

00:04:42.560 --> 00:04:46.880
Now, what we did in the last year
or so was to add orthogonal

00:04:46.880 --> 00:04:50.587
measurements to strainmeters.
So a normal strainmeter just –

00:04:50.587 --> 00:04:54.880
a creepmeter measures merely
an oblique component.

00:04:54.880 --> 00:05:00.080
But it’s sort of ambiguous
if one assumes that block-like

00:05:00.080 --> 00:05:03.360
motion does not apply.
So various geometries were tried,

00:05:03.360 --> 00:05:07.840
and the results are shown here.
Dilation is plotted vertically,

00:05:07.840 --> 00:05:14.960
dextral slip horizontally. And there
are seven faults here, and we’ll

00:05:14.960 --> 00:05:18.800
look at the extreme results now.
The first one is from the Dead Sea.

00:05:18.800 --> 00:05:25.520
Second from Salt Creek on the Salton
Sea segment of the San Andreas Fault.

00:05:25.520 --> 00:05:28.160
And the third one is XMR,
which we just looked at.

00:05:28.160 --> 00:05:34.880
If you look at the Dead Sea vector,
it’s across a normal fault, and you

00:05:34.880 --> 00:05:39.440
would expect there to be considerable
opening should it move, and it did.

00:05:39.440 --> 00:05:42.880
It started – the first six months of
data didn’t do very much, and then,

00:05:42.880 --> 00:05:48.376
in September, it started slipping
slowly, 24 millimeters a year.

00:05:48.400 --> 00:05:52.880
However, the first six months of data
were accompanied by dextral slip.

00:05:52.880 --> 00:05:56.320
And only when it started moving
rapidly did you get a sinistral

00:05:56.320 --> 00:06:01.920
motion across the fault.
Second example is from Durmid Hill

00:06:01.920 --> 00:06:07.256
on the southern San Andreas.
The creepmeter array looks like this.

00:06:07.280 --> 00:06:10.080
You have three
simultaneous equations to solve.

00:06:10.080 --> 00:06:14.400
And when you do that, you finish up
with an observation that the fault zone

00:06:14.400 --> 00:06:21.920
is closing slowly at about
0.2 millimeters per year,

00:06:21.920 --> 00:06:27.360
a very small amount, but it’s
closing both in a secular sense

00:06:27.360 --> 00:06:30.320
and in creep events –
individual creep events.

00:06:30.320 --> 00:06:34.560
So here is the creep during the
Chiapas event. It closes at 7.9 degrees.

00:06:34.560 --> 00:06:37.680
During Ridgecrest, at 5.
During a recent creep event,

00:06:37.680 --> 00:06:44.160
it’s still ongoing at 7.5.
All these angles of transpressional

00:06:44.160 --> 00:06:51.896
closure are very similar to that
inferred for the entire fault system.

00:06:51.920 --> 00:06:57.600
Now we can look at the XMR data.
This is from the northern creeping zone.

00:06:57.600 --> 00:07:03.896
And you can see a very
complicated fault slip vector.

00:07:03.920 --> 00:07:09.416
First of all, periods of heavy rain
make the fault go completely nuts.

00:07:09.440 --> 00:07:15.680
But, in between, during the dry season,
you can see that the fault zone opens

00:07:15.680 --> 00:07:22.080
quite uniformly at about 12 degrees.
Here’s an example of a single

00:07:22.080 --> 00:07:26.880
creep event. And, during the
inter-event period, it also closes

00:07:26.880 --> 00:07:30.240
at about 12 degrees.
And the difference in velocities

00:07:30.240 --> 00:07:35.416
between these two are about 24,000.
So this is a very unexpected result.

00:07:35.440 --> 00:07:38.800
We have a sort of explanation for that.
If you look at the fault zone,

00:07:38.800 --> 00:07:45.920
you can find that it consists of these
curious clasts of clay that are obviously

00:07:45.920 --> 00:07:49.360
rotating relative to each other
and interlinked in some way.

00:07:49.360 --> 00:07:54.856
So, during slip, we suppose that these –
during a creep event – rapid creep,

00:07:54.880 --> 00:07:59.256
we suppose dilation occurs through
rotation or wedging of these clasts.

00:07:59.280 --> 00:08:02.936
But, during the inter-event
period, they rotate back.

00:08:02.960 --> 00:08:07.280
And during saturation conditions,
when heavy rain occurs,

00:08:07.280 --> 00:08:12.880
they completely relax into a tight
packing arrangement. We’re not sure.

00:08:12.880 --> 00:08:18.616
Now, what about gas?
During the Sivrice earthquake

00:08:18.640 --> 00:08:25.520
in 2021, there was a huge amount –
or, sorry, 2020 the earthquake occurred.

00:08:25.520 --> 00:08:31.336
In the first few days of that event,
when afterslip was just beginning,

00:08:31.360 --> 00:08:35.200
we saw a large amount of gas
coming out of the fault zone.

00:08:35.200 --> 00:08:39.280
So the natural question is, does that
occur on the San Andreas system?

00:08:39.280 --> 00:08:43.520
Is it caused by ventilation –
the fault moving – or is it actually

00:08:43.520 --> 00:08:47.840
causing the fault to move through
some kind of lubrication effect?

00:08:47.840 --> 00:08:51.920
Well, the answer is that gases are
indeed coming out of the fault zone,

00:08:51.920 --> 00:08:58.296
but largely because of
biogenic processes.

00:08:58.320 --> 00:09:02.000
Microbes become active
when the fault zone is wet.

00:09:02.000 --> 00:09:06.000
But what we did notice was that,
after creep events, not during them,

00:09:06.000 --> 00:09:12.936
we had bursts of gas coming out.
And we suppose that’s caused by

00:09:12.960 --> 00:09:16.880
en echelon crack formation,
since here is an example of

00:09:16.880 --> 00:09:22.800
crack formation – we saw these
cracks appear soon after we

00:09:22.800 --> 00:09:25.920
witnessed gas coming out.
Radon seems to come out of the fault

00:09:25.920 --> 00:09:31.440
with carbon dioxide as a kind of carrier.
Now, looking at the creep events

00:09:31.440 --> 00:09:37.200
in detail, this is the opening
velocity stacked for five events.

00:09:37.200 --> 00:09:42.400
And the carbon dioxide vented,
one-minute samples for five events,

00:09:42.400 --> 00:09:45.600
also stacked. And what you can
see is that, when the fault opens,

00:09:45.600 --> 00:09:50.080
there is a reduction is the
concentration of carbon dioxide.

00:09:50.080 --> 00:09:53.760
This would be atmospheric levels,
and this would be the fault zone levels.

00:09:53.760 --> 00:09:56.960
So there is – this reduction means
the fault probably did indeed

00:09:56.960 --> 00:10:01.256
dilate sufficiently to suck
air in from the atmosphere.

00:10:01.280 --> 00:10:04.536
The fault sucks
during creep events.

00:10:04.560 --> 00:10:08.960
This example I’m going to show you
is from the Calaveras Fault,

00:10:08.960 --> 00:10:10.800
and it shows something
much more interesting.

00:10:10.800 --> 00:10:18.000
What we’re seeing here is clearly
a gas venting before a creep event and

00:10:18.000 --> 00:10:22.240
turning off a few weeks afterwards.
Again, same thing happened here.

00:10:22.240 --> 00:10:25.040
Very large creep event –
these are 16 millimeters.

00:10:25.040 --> 00:10:29.416
And it switches off afterwards.
And there’s, you know,

00:10:29.440 --> 00:10:35.280
normal oscillations of gas in between.
But these are off-scale, both for radon

00:10:35.280 --> 00:10:40.216
and for carbon dioxide. What we
think may be going on there is

00:10:40.240 --> 00:10:46.560
that fault valving of some sort nucleates
gas near microearthquakes that are

00:10:46.560 --> 00:10:50.800
occurring in the seismogenic zone,
propagating to the surface slowly,

00:10:50.800 --> 00:10:56.640
and coming to the surface with creep
events. Since they precede creep

00:10:56.640 --> 00:11:05.256
events, we infer that maybe they are
lubricating the fault in some way.

00:11:05.280 --> 00:11:08.616
So gas venting
actually brackets slip.

00:11:08.640 --> 00:11:14.240
So, in summary – and we found
various new novel ways of

00:11:14.240 --> 00:11:19.040
looking at creeping faults.
And, if they are to be incorporated in

00:11:19.040 --> 00:11:24.936
the next century of fault measurements,
we need to ensure data continuity.

00:11:24.960 --> 00:11:27.760
In other words, right now,
they are a bit of an outlier.

00:11:27.760 --> 00:11:30.856
They’re maintained
almost by accident.

00:11:30.880 --> 00:11:35.416
We need more creepmeters,
and we need multiple sensors.

00:11:35.440 --> 00:11:38.390
Thanks very much
for listening.