WEBVTT
Kind: captions
Language: en-US

00:00:02.046 --> 00:00:06.160
[silence]

00:00:06.160 --> 00:00:08.640
Hi, everyone.
My name is Noha Farghal.

00:00:08.640 --> 00:00:12.800
I’m a Mendenhall postdoctoral fellow
at the USGS Earthquake Science Center

00:00:12.800 --> 00:00:16.720
in Moffett Field, California.
Today, I will talk about the potential

00:00:16.720 --> 00:00:20.056
of fiber optic sensing
in earthquake early warning.

00:00:20.080 --> 00:00:22.640
But first I would like to
acknowledge and thank

00:00:22.640 --> 00:00:27.120
my collaborators,
Jessie Saunders and Grace Parker.

00:00:31.440 --> 00:00:35.840
Recently, there has been a focus on
using data from borehole strainmeters

00:00:35.840 --> 00:00:39.520
in earthquake early warning and
creating different strain seismology

00:00:39.520 --> 00:00:44.616
tools to calculate various source and
ground motion parameters from strain.

00:00:44.640 --> 00:00:49.920
Using records from the July 2019
Ridgecrest sequence magnitude 7.1

00:00:49.920 --> 00:00:54.320
main shock, we demonstrated that
reliable magnitude estimates –

00:00:54.320 --> 00:00:58.480
the blue and yellow lines –
can be derived from strain

00:00:58.480 --> 00:01:01.440
through this empirical relation
between earthquake moment

00:01:01.440 --> 00:01:06.320
magnitudes, peak dynamic strains,
and hypocentral distances.

00:01:06.320 --> 00:01:10.000
And we showed that incorporating
strain would have possibly helped the

00:01:10.000 --> 00:01:14.880
ShakeAlert system reach the threshold
number of stations for alerting faster

00:01:14.880 --> 00:01:20.310
and possibly contributed to a more
accurate system performance.

00:01:21.920 --> 00:01:26.640
In another recent work, we derived
a magnitude scale entirely based on

00:01:26.640 --> 00:01:31.576
strain data with corrections
for source and site effects.

00:01:31.600 --> 00:01:35.200
Which was also tested using
the Ridgecrest main shock records

00:01:35.200 --> 00:01:38.400
and further confirms
the utility of strain data

00:01:38.400 --> 00:01:42.645
in providing robust strain
magnitude updates.

00:01:43.680 --> 00:01:47.520
It is also worth mentioning that this
strain-based magnitude scale is

00:01:47.520 --> 00:01:52.151
currently used by UNAVCO
for strain magnitude estimation.

00:01:54.808 --> 00:01:58.000
[silence]

00:01:58.000 --> 00:02:02.400
We also did a proof of concept for
calculating peak ground velocities,

00:02:02.400 --> 00:02:07.336
or PGVs, from strain for several
recent earthquakes in California

00:02:07.360 --> 00:02:12.480
and found that strain-derived PGVs
are practically indistinguishable

00:02:12.480 --> 00:02:17.440
from seismic-derived ones.
And we are currently working on

00:02:17.440 --> 00:02:21.360
calculating peak ground
accelerations, or PGAs, from strain.

00:02:21.360 --> 00:02:24.400
And so far, we are getting
promising results comparing

00:02:24.400 --> 00:02:28.582
strain-derived and
seismic-derived PGAs.

00:02:33.040 --> 00:02:36.480
In deriving all these relations,
we have been using data from

00:02:36.480 --> 00:02:40.560
borehole strainmeters of the USGS,
a Network of the Americas networks

00:02:40.560 --> 00:02:45.200
along the U.S. West Coast,
which are sparse, and it’s quite unlikely

00:02:45.200 --> 00:02:49.200
that more such strainmeters will be
installed on a large scale in the future

00:02:49.200 --> 00:02:53.709
to complement traditional seismic
data in earthquake early warning.

00:02:55.200 --> 00:02:59.600
So we looked to fiber optic distributed
acoustic sensing, or fiber optic DAS,

00:02:59.600 --> 00:03:05.600
for its potential for scalability, real-time
telemetry, cost-effective dense sensing.

00:03:05.600 --> 00:03:10.400
It has been used for years through
arrays of different sizes and layouts,

00:03:10.400 --> 00:03:16.776
vertical and horizontal arrays, for
various applications all over the world.

00:03:16.800 --> 00:03:21.360
Dark-fiber-type installations also
successfully recorded earthquakes

00:03:21.360 --> 00:03:25.680
with high signal-to-noise ratios,
and there is a growing discussion now

00:03:25.680 --> 00:03:29.440
about how to leverage the
extensive network of infrastructure

00:03:29.440 --> 00:03:33.896
subsea cables for improving
offshore hazard monitoring.

00:03:33.920 --> 00:03:38.080
Google is currently working on
tracking state of polarization of light

00:03:38.080 --> 00:03:41.920
to detect seismic activity over
tens of thousands of kilometers

00:03:41.920 --> 00:03:47.120
using infrastructure cables.
They detected the January 28th, 2020,

00:03:47.120 --> 00:03:51.440
magnitude 7.7 earthquake
off of Jamaica and were able to

00:03:51.440 --> 00:03:55.496
compare their data to those
from traditional seismic stations.

00:03:55.520 --> 00:03:59.360
Although this is not fiber optic DAS,
I found it to be an interesting

00:03:59.360 --> 00:04:03.200
recent demonstration of the
potential of infrastructure cables

00:04:03.200 --> 00:04:06.045
in hazard monitoring
in general.

00:04:09.440 --> 00:04:12.480
Just a quick reminder of
how fiber optic DAS works.

00:04:12.480 --> 00:04:16.800
It starts with an interrogator unit
generating and sending short laser

00:04:16.800 --> 00:04:21.120
pulses down the optical fiber.
Some of this energy is backscattered

00:04:21.120 --> 00:04:24.560
by the naturally occurring scatters
in the fiber, which are basically

00:04:24.560 --> 00:04:28.856
imperfections or inconsistencies
in the refractive index.

00:04:28.880 --> 00:04:32.160
When the backscattered energy
reaches the interrogator unit,

00:04:32.160 --> 00:04:35.440
interferometry is performed,
and if the fiber is subjected to

00:04:35.440 --> 00:04:40.240
some acoustic field that causes strain
or deformation somewhere on the fiber,

00:04:40.240 --> 00:04:43.520
there will be a resulting phase
difference between the pulses.

00:04:43.520 --> 00:04:45.600
And, depending on
the optical setup,

00:04:45.600 --> 00:04:49.656
you can end up either
measuring strain or strain rate.

00:04:49.680 --> 00:04:53.440
And, of course, the two-way travel
time of the received pulse is used

00:04:53.440 --> 00:04:56.776
to determine the location of
measurement down the fiber.

00:04:56.800 --> 00:05:02.376
This location is called channel
location, or sensing point.

00:05:02.400 --> 00:05:05.920
And the length of fiber in which the
strain or strain rate measurement

00:05:05.920 --> 00:05:10.800
is averaged is called the
gauge length, which is typically

00:05:10.800 --> 00:05:14.696
8 to 16 meters for
earthquake surface recordings

00:05:14.720 --> 00:05:19.929
and 4 to 7 meters for microseismic
borehole recordings.

00:05:24.320 --> 00:05:29.336
DAS has many applications,
including vertical seismic profiling;

00:05:29.360 --> 00:05:33.040
earthquake detection and monitoring;
soil property estimation and

00:05:33.040 --> 00:05:36.720
hydrological characterization;
structural monitoring;

00:05:36.720 --> 00:05:41.680
CO2 monitoring; hydraulic fracturing
process monitoring in oil and gas plays;

00:05:41.680 --> 00:05:45.520
and long-term strain sensing,
such as melting of the permafrost

00:05:45.520 --> 00:05:49.176
and oceanic properties,
to mention a few.

00:05:49.200 --> 00:05:51.920
And there are many works that
compare the performance of DAS

00:05:51.920 --> 00:05:56.000
to seismometers and geophones that
demonstrate the good signal-to-noise

00:05:56.000 --> 00:06:00.080
ratio that DAS is capable of providing.
And, while DAS measures strain

00:06:00.080 --> 00:06:04.000
or strain rate, particle velocity
can be derived from DAS data

00:06:04.000 --> 00:06:06.880
if apparent slowness in the
cable direction is known.

00:06:09.200 --> 00:06:12.560
It is also worth mentioning that
one interrogator can be used

00:06:12.560 --> 00:06:19.016
to analyze fiber cables that are
50 or even 100 kilometers long.

00:06:19.040 --> 00:06:22.560
Moreover, an interrogator unit
can be placed onshore

00:06:22.560 --> 00:06:26.000
with a marine cable or be
placed underwater if needed.

00:06:29.760 --> 00:06:33.040
So now let’s think about the role
that fiber optic DAS can play in

00:06:33.040 --> 00:06:37.336
enhancing the performance of various
earthquake early warning systems.

00:06:37.360 --> 00:06:41.440
When it comes to earthquake locations
and arrival times, the current DAS

00:06:41.440 --> 00:06:45.760
capabilities can provide high
signal-to-noise ratios for accurate

00:06:45.760 --> 00:06:50.080
source location and origin time
estimation, which are useful for

00:06:50.080 --> 00:06:52.640
point-source source
parameter algorithms,

00:06:52.640 --> 00:06:56.383
like the ShakeAlert
system EPIC algorithm.

00:06:58.800 --> 00:07:02.240
We can further enhance the
signal-to-noise ratio if needed

00:07:02.240 --> 00:07:07.819
by using interrogators with higher
sensitivities and lower noise floors.

00:07:08.640 --> 00:07:12.400
While data with good signal-to-noise
ratios were obtained using standard

00:07:12.400 --> 00:07:18.000
telecom fibers, dedicated installations
that require high fiber sensitivity and

00:07:18.000 --> 00:07:23.416
exception signal-to-noise ratios
utilize special or engineered fibers.

00:07:23.440 --> 00:07:27.520
Engineered fibers provide higher
backscattering capabilities without

00:07:27.520 --> 00:07:31.680
significantly increasing fiber losses,
which allows for the detection of

00:07:31.680 --> 00:07:36.880
very small signals at distances of about
1 kilometer from the interrogator.

00:07:36.880 --> 00:07:42.560
They can offer 100 times’ improvement
in fiber sensitivity compared to telecom

00:07:42.560 --> 00:07:47.280
fiber varieties and up to 30-decibel
enhancement in noise levels.

00:07:47.280 --> 00:07:52.000
However, the cost of engineered fibers
can be significantly higher than those

00:07:52.000 --> 00:07:56.480
of standard telecom fibers,
which cost about $1 to 2 per meter.

00:07:56.480 --> 00:08:00.800
But we’re not considering only the
fibers but the whole cable bundle –

00:08:00.800 --> 00:08:07.421
engineered fiber cables can cost up to
15 times more than telecom fiber cables.

00:08:12.080 --> 00:08:15.920
We also need to make sure that the
cables are oriented in such a way

00:08:15.920 --> 00:08:21.896
as to have optimal azimuthal or
directional sensitivity to P waves.

00:08:21.920 --> 00:08:26.240
So, if the fiber is oriented this way –
horizontally – when a seismic wave

00:08:26.240 --> 00:08:29.840
of wavelength much longer than the
gauge length propagates through the

00:08:29.840 --> 00:08:34.720
fiber, the fiber will be highly sensitive
to longitudinal waves propagating

00:08:34.720 --> 00:08:40.240
along the fiber and to transverse
waves at 45 degrees through the fiber.

00:08:40.240 --> 00:08:42.240
And I will show
an example fiber layout

00:08:42.240 --> 00:08:45.530
for multi-components
acquisition in a minute.

00:08:48.480 --> 00:08:52.880
One issue that there isn’t an update
on is how a DAS array would perform

00:08:52.880 --> 00:08:57.440
when installed very close to a large
rupture or a large earthquake and

00:08:57.440 --> 00:09:01.200
whether there is an equivalent of
clipping in DAS systems as there is

00:09:01.200 --> 00:09:05.520
in seismometers, for example,
as well as the effect of high strain

00:09:05.520 --> 00:09:09.097
and high strain rates
on the fiber cables themselves.

00:09:11.040 --> 00:09:15.920
Clipping in the optical system may be
thought of as a phase wraparound as a

00:09:15.920 --> 00:09:21.360
result of the signal rising too quickly.
This can presumably be adjusted

00:09:21.360 --> 00:09:26.640
by pre-setting the phase quantization
increment in the hardware if the

00:09:26.640 --> 00:09:31.440
expected strain rates are known.
However, this increment will

00:09:31.440 --> 00:09:35.440
become a hard limit,
as it is not user-selectable

00:09:35.440 --> 00:09:39.183
and will not be easy
to change afterwards.

00:09:44.720 --> 00:09:47.920
When talking about P wave
displacements for rapid magnitude

00:09:47.920 --> 00:09:52.400
estimation, DAS capabilities
to provide good signal-to-noise ratio

00:09:52.400 --> 00:09:55.920
is well-established.
But several studies show that

00:09:55.920 --> 00:10:00.000
DAS strain amplitude accuracy
highly depends on the quality

00:10:00.000 --> 00:10:03.280
of strain coupling between
the ground and the fiber

00:10:03.280 --> 00:10:05.884
and the conditions
of the installation.

00:10:08.000 --> 00:10:12.240
For example, the top figure shows
how the installation conditions,

00:10:12.240 --> 00:10:17.760
being a single conduit, a cased conduit,
or an attached conduit, can affect

00:10:17.760 --> 00:10:21.576
recorded data, possibly
due to variations in coupling.

00:10:21.600 --> 00:10:26.000
This work by Nate Lindsey and others
shows that, while the phase response

00:10:26.000 --> 00:10:30.480
remained practically flat for the entire
range of seismic frequencies they

00:10:30.480 --> 00:10:35.200
studied, there is an elevated amplitude
response at higher frequencies,

00:10:35.200 --> 00:10:39.280
which the authors think might be
due to coupling issues of the cable

00:10:39.280 --> 00:10:43.896
that leads to such
frequency-dependent response.

00:10:43.920 --> 00:10:48.160
Even in dedicated installations,
poor strain coupling has resulted

00:10:48.160 --> 00:10:51.840
in mismatches between DAS data
and co-located geophones

00:10:51.840 --> 00:10:55.976
and uncemented fiber segments
were thought to be problematic.

00:10:56.000 --> 00:11:00.160
A few studies of cable packaging have
shown some minor differences in

00:11:00.160 --> 00:11:03.840
cable response compared to the
more prominent effect of

00:11:03.840 --> 00:11:08.000
the strength of strain coupling.
So this is also a need for further

00:11:08.000 --> 00:11:11.920
research for different array
installations and to investigate

00:11:11.920 --> 00:11:15.120
whether infrastructure or
dark fiber installations

00:11:15.120 --> 00:11:19.286
would be suitable for earthquake
early warning applications.

00:11:21.040 --> 00:11:25.440
PGVs or PGAs used in ground
motion-based finite fault parameter

00:11:25.440 --> 00:11:29.920
estimation algorithms, such as FinDer,
or wavefield propagation algorithms

00:11:29.920 --> 00:11:34.776
such as PLUM, can be
derived from dynamic strains.

00:11:34.800 --> 00:11:38.720
But they additionally require
multi-component DAS acquisition,

00:11:38.720 --> 00:11:42.536
which can be achieved
with helically wound fibers.

00:11:42.560 --> 00:11:47.200
And you can reconstruct the full strain
tensor from this layout of five helically

00:11:47.200 --> 00:11:51.256
wound fibers and one straight fiber
that isn’t shown in the figure.

00:11:51.280 --> 00:11:55.040
And it’s been confirmed in field tests
that at least the P wave response

00:11:55.040 --> 00:11:59.280
of the cable was angle independent.
I know the cable looks large in the

00:11:59.280 --> 00:12:03.120
figure, but the next-generation
cables have diameters as small as

00:12:03.120 --> 00:12:07.360
1.8 centimeter, and you can
get multi-component response

00:12:07.360 --> 00:12:09.840
without complicated layouts.

00:12:14.400 --> 00:12:18.560
Finally, since we derived a number
of useful relations between dynamic

00:12:18.560 --> 00:12:22.160
strains and earthquake magnitudes
and ground motion parameters,

00:12:22.160 --> 00:12:26.240
if at some point in the future,
we can develop a new algorithm

00:12:26.240 --> 00:12:30.400
based on strain data from fiber
optic sensors or strainmeters,

00:12:30.400 --> 00:12:33.920
it can be used for calculating
different earthquake source parameters

00:12:33.920 --> 00:12:37.440
and ground motion metrics
directly from strain and sent to

00:12:37.440 --> 00:12:38.960
the earthquake
early warning system.

00:12:41.731 --> 00:12:45.760
[silence]

00:12:45.760 --> 00:12:49.360
In summary, there are many ways
that DAS data can contribute to

00:12:49.360 --> 00:12:52.720
different types of earthquake
early warning algorithms through

00:12:52.720 --> 00:12:56.640
contributing ground motion
observations directly to existing

00:12:56.640 --> 00:13:00.800
earthquake early warning algorithms
after conversion to the data types

00:13:00.800 --> 00:13:04.880
ingested by these algorithms and
through augmenting the current

00:13:04.880 --> 00:13:08.720
earthquake early warning systems by
supplying additional earthquake source

00:13:08.720 --> 00:13:14.790
parameters through the development of
an independent strain-based algorithm.

00:13:16.160 --> 00:13:20.400
In our upcoming paper, we highlighted
a few DAS-related issues in need

00:13:20.400 --> 00:13:24.080
of further investigation,
such as amplitude accuracy

00:13:24.080 --> 00:13:27.520
in weak or non-uniform
coupling installations and

00:13:27.520 --> 00:13:31.719
array performance
very close to large ruptures.

00:13:34.240 --> 00:13:38.480
I would like to thank Martin
Karrenbach, Sarah Minson,

00:13:38.480 --> 00:13:44.456
Andrew Barbour, Scott DeWolf, Evelyn
Roeloffs, John Lanbein, Julian Bunn,

00:13:44.480 --> 00:13:48.320
and Elizabeth Cochran for their
feedback and help with this work.

00:13:48.320 --> 00:13:50.880
And, with that, I thank you
very much for your attention,

00:13:50.880 --> 00:13:54.440
and I look forward to your
questions and the discussion.

00:13:56.624 --> 00:14:02.340
[silence]