WEBVTT
Kind: captions
Language: en-US

00:00:00.560 --> 00:00:03.920
So I’m Kate Allstadt from the
USGS in Golden, Colorado.

00:00:03.920 --> 00:00:07.680
And today I’m going to talk about the
landslides that were triggered by the

00:00:07.680 --> 00:00:12.160
2021 Nippes earthquake in Haiti
and the rapid remote response effort

00:00:12.160 --> 00:00:15.760
that myself and my co-authors,
who are listed on the slide here,

00:00:15.760 --> 00:00:20.376
undertook in order to provide
situational awareness support.

00:00:20.400 --> 00:00:23.520
So, though this is a northern California
earthquake hazards workshop,

00:00:23.520 --> 00:00:26.800
and Haiti is a different place, I’m hoping
that you’ll see some points of relevance

00:00:26.800 --> 00:00:30.080
with regards to, first, what tools we
now have available for situational

00:00:30.080 --> 00:00:33.440
awareness after major landslide-
triggering events that we didn’t have

00:00:33.440 --> 00:00:37.040
a few years ago, and second,
what changes to the landslide hazard

00:00:37.040 --> 00:00:41.120
outlook to expect both right after and for
years after a major landslide-triggering

00:00:41.120 --> 00:00:45.016
event, which have been known
to happen in California.

00:00:45.040 --> 00:00:48.880
And much of the work that I’m going to
present on today was recently published

00:00:48.880 --> 00:00:54.536
in a Open-File report along with a
data release of the landslide inventory.

00:00:54.560 --> 00:00:57.760
So thanks to Sabrina Martinez
for seeing that through so quickly.

00:00:57.760 --> 00:01:00.160
And so, if you want more details about
what I’m going to talk about today,

00:01:00.160 --> 00:01:03.816
I suggest that you take
a look at these publications.

00:01:03.840 --> 00:01:06.880
So, before we jump into what happened
in Haiti, here’s a quick overview

00:01:06.880 --> 00:01:11.040
of cascading hazards related to
earthquake-triggered landslides.

00:01:11.040 --> 00:01:14.560
And, so in this diagram by Fan et al.,
it shows the immediate effects

00:01:14.560 --> 00:01:17.120
of the earthquake in the center.
So those are the landslides triggered

00:01:17.120 --> 00:01:21.040
by the earthquake as well as damage
to the slopes caused by the shaking.

00:01:21.040 --> 00:01:24.480
But, in the blue section, you see shorter-
term hazards, such as debris flows

00:01:24.480 --> 00:01:29.680
mobilized from the landslide deposits
and also landslide dams that can form

00:01:29.680 --> 00:01:34.296
and impound water, and they can fail
catastrophically and cause flooding.

00:01:34.320 --> 00:01:38.720
Then, in the yellow section, you see
longer-term hazards, such as changing

00:01:38.720 --> 00:01:41.920
in the – changes in the way that the
rivers behave and their

00:01:41.920 --> 00:01:45.440
flooding behavior and so on.
And we also see – typically,

00:01:45.440 --> 00:01:50.216
for several years after a earthquake,
you see an increase in the overall

00:01:50.240 --> 00:01:54.536
landslide activity because of the
damaged slopes from the earthquake.

00:01:54.560 --> 00:01:57.920
So here I’m showing a screenshot of
the interactive map from the USGS

00:01:57.920 --> 00:02:02.080
event page, and I’m showing the
USGS finite faults as well as

00:02:02.080 --> 00:02:05.040
the contours of the ShakeMap.
And you can see that the earthquake

00:02:05.040 --> 00:02:08.800
affected mainly the western
half of the Tiburon Peninsula,

00:02:08.800 --> 00:02:12.456
and it ruptured
from east to west.

00:02:12.480 --> 00:02:14.720
And so the rest of the maps,
we’re going to be zooming in

00:02:14.720 --> 00:02:17.656
on the western part
of the peninsula.

00:02:17.680 --> 00:02:22.720
And, in this map here, kind of
summarizes the event and its landslides.

00:02:22.720 --> 00:02:27.840
So you can see the epicenter in the east,
and so the rupture started with more

00:02:27.840 --> 00:02:30.560
reverse motion in the east, and then,
as it progressed westward,

00:02:30.560 --> 00:02:34.000
it’s thought to have transitioned
to strike-slip motion primarily.

00:02:34.000 --> 00:02:36.320
And it ended in the
mountainous area where

00:02:36.320 --> 00:02:39.576
there’s intense landsliding
shown on this map here.

00:02:39.600 --> 00:02:45.896
And so you can see that, in the east, the
landslides are more sparsely distributed.

00:02:45.920 --> 00:02:50.080
They kind of narrow towards the center,
and then most of the intense landsliding

00:02:50.080 --> 00:02:53.416
actually happened on
the western side of the rupture.

00:02:53.440 --> 00:02:58.080
And so some of that is thought to be
probably due to directivity effects

00:02:58.080 --> 00:03:03.520
because of the east-to-west rupture,
but this area to the west is also highly

00:03:03.520 --> 00:03:10.160
susceptible – very steep slopes, and so
multiple factors were likely at play.

00:03:10.160 --> 00:03:13.160
But I’m sure that we’ll find out
more about the details of the

00:03:13.160 --> 00:03:16.856
earthquake as research comes
out in the coming months.

00:03:16.880 --> 00:03:20.480
So, at left here, we kind of have
a timeline of the sequence of

00:03:20.480 --> 00:03:23.520
events and our response to it.
So the earthquake happened

00:03:23.520 --> 00:03:28.400
on the 14th. And, as for all major
earthquakes, the USGS produced

00:03:28.400 --> 00:03:33.120
its rapid near-real-time products,
and that now includes a ground failure

00:03:33.120 --> 00:03:36.080
product, so that’s the model
shown in the background here.

00:03:36.080 --> 00:03:38.456
And I’ll talk about
that more in a minute.

00:03:38.480 --> 00:03:41.680
Two days after the earthquake,
a tropical cyclone hit the island,

00:03:41.680 --> 00:03:46.616
dropping about 5 to 10 inches of rain.
And it likely triggered more landslides,

00:03:46.640 --> 00:03:50.480
but we don’t have the imagery
resolution to separate which were from

00:03:50.480 --> 00:03:54.616
the cyclone and which were from the
earthquake from the map at right.

00:03:54.640 --> 00:03:57.680
However, based on the
characteristics of the landslides,

00:03:57.680 --> 00:04:00.536
most of them were caused
by the earthquake.

00:04:00.560 --> 00:04:04.720
So, starting right after the earthquake,
and through the end of August,

00:04:04.720 --> 00:04:09.576
the team of seven of us worked on
this rapid response and mapping effort

00:04:09.600 --> 00:04:13.760
where we mapped collaboratively
and we provided situational reports

00:04:13.760 --> 00:04:17.760
to the USAID’s Disaster Assistance
Response teams that were on the

00:04:17.760 --> 00:04:22.560
ground here pretty much every day.
And, once that rapid effort had kind of

00:04:22.560 --> 00:04:26.400
wound down, then we worked on
refining the inventory and looking

00:04:26.400 --> 00:04:30.720
more carefully at kind of
ongoing hazards of concern.

00:04:30.720 --> 00:04:38.856
And we prepared a preliminary
assessment report internally for BHA.

00:04:38.880 --> 00:04:42.880
And then finally, we cleaned that up,
and cleaned up the inventory

00:04:42.880 --> 00:04:45.760
and prepared it all for public release,
and that’s what’s available on

00:04:45.760 --> 00:04:49.369
that Open-File report
I mentioned earlier.

00:04:50.880 --> 00:04:56.080
So going back to the USGS ground
failure product, here I’m just

00:04:56.080 --> 00:04:58.880
showing you where it lives.
This is the USGS event page

00:04:58.880 --> 00:05:01.920
for the Haiti earthquake, and you
can see that the ground failure product

00:05:01.920 --> 00:05:08.376
has its own card that summarizes
the product on the event page.

00:05:08.400 --> 00:05:10.960
And you can see that there’s
information on both landslides

00:05:10.960 --> 00:05:13.840
and liquefaction. If you click
on the card, you can – you go to

00:05:13.840 --> 00:05:19.336
a more detailed page where you
can view maps and other statistics.

00:05:19.360 --> 00:05:22.080
And I’m not going to talk
about liquefaction very much,

00:05:22.080 --> 00:05:26.720
but it is available. And the landslide
model is shown at the top here,

00:05:26.720 --> 00:05:31.840
and you can see that the summary
information assigned this event

00:05:31.840 --> 00:05:34.880
an orange alert for landslides
based on both its hazard and

00:05:34.880 --> 00:05:41.600
population exposure, which indicates a
pretty significant event. There is fewer

00:05:41.600 --> 00:05:47.988
than three orange alerts that happen per
year for the ground failure product.

00:05:49.120 --> 00:05:53.440
Another product that is not public yet
that is still in testing mode, but that

00:05:53.440 --> 00:05:59.200
we did run for this event, and we
shared with responders, is this product

00:05:59.200 --> 00:06:03.520
that takes the landslides probability
model and the road network and

00:06:03.520 --> 00:06:06.800
kind of estimates where
roads might be blocked.

00:06:06.800 --> 00:06:11.040
And so this is the result of that,
and you can see some yellow areas

00:06:11.040 --> 00:06:14.720
that were thought to have been
at higher risk of being obstructed.

00:06:14.720 --> 00:06:20.400
And this area here was a major issue
because this is the main connector

00:06:20.400 --> 00:06:26.136
between Les Cayes and Jeremie
and these other areas to the north.

00:06:26.160 --> 00:06:28.720
And so this is what
that area looks like.

00:06:28.720 --> 00:06:31.920
This is in some imagery
from Google Earth.

00:06:31.920 --> 00:06:36.376
Basically, this whole stretch of road
was affected by intense landsliding.

00:06:36.400 --> 00:06:42.320
There’s some – there was one drone
video posted on Twitter showing them

00:06:42.320 --> 00:06:44.400
trying to clear this road,
but it took quite a while,

00:06:44.400 --> 00:06:49.119
and it was a big issue for getting –
for moving supplies over land.

00:06:50.960 --> 00:06:55.440
So, as I said earlier, this was the first
time that we did a entirely remote

00:06:55.440 --> 00:07:00.320
response, and we were able to take
advantage of some new tools through

00:07:00.320 --> 00:07:04.480
ArcGIS Pro and ArcGIS Online,
where the seven of us who were

00:07:04.480 --> 00:07:09.120
involved in this effort were able to
map collaboratively, you know,

00:07:09.120 --> 00:07:12.080
in the same files, and we were
able to share those updates

00:07:12.080 --> 00:07:16.480
in real time with the public.
And we mainly did this so that

00:07:16.480 --> 00:07:19.816
responders could see this map
as we were updating it,

00:07:19.840 --> 00:07:25.176
but it was available to anybody
to see as preliminary information.

00:07:25.200 --> 00:07:29.200
And so this is showing a screenshot of
what we were able to share publicly.

00:07:29.200 --> 00:07:33.280
And we mapped about
3,600 landslides within

00:07:33.280 --> 00:07:36.136
about two weeks of the
earthquake happening.

00:07:36.160 --> 00:07:40.240
And, as I mentioned earlier,
we provided daily updates to USAID

00:07:40.240 --> 00:07:44.984
on what we were seeing and
how many landslides and so on.

00:07:46.800 --> 00:07:53.176
So one area of early concern was
the Pic Macaya National Park.

00:07:53.200 --> 00:07:56.480
So this is imagery of the
Pic Macaya National Park from right –

00:07:56.480 --> 00:08:00.880
a few days before the earthquake.
And this is imagery – this is all

00:08:00.880 --> 00:08:04.480
Sentinel 2 imagery –
from right after the earthquake.

00:08:04.480 --> 00:08:08.800
And you can see that it was just
completely devastated by landslides

00:08:08.800 --> 00:08:13.280
in this gorge. And this was a big
concern because this gorge is upstream

00:08:13.280 --> 00:08:17.280
from many of the most populated
areas in the region, so Camp Perrin and

00:08:17.280 --> 00:08:21.760
Les Cayes are downstream from this.
And there was concern that some of

00:08:21.760 --> 00:08:24.720
these landslides might have formed
landslide dams that we couldn’t see

00:08:24.720 --> 00:08:29.040
in this low-resolution imagery,
and they could fail catastrophically

00:08:29.040 --> 00:08:30.800
and cause rapid
flooding downstream.

00:08:30.800 --> 00:08:34.216
And there was a citizen report about
something like this happening.

00:08:34.240 --> 00:08:38.856
But we did not get cloud-free imagery
for quite a while after the earthquake.

00:08:38.880 --> 00:08:42.640
Though, fortunately, USAID was able
to arrange an overflight, and some

00:08:42.640 --> 00:08:46.536
Haitian colleagues were able to get
on that flight, and they flew

00:08:46.560 --> 00:08:52.560
up the gorge and did not see
anything of major concern.

00:08:53.600 --> 00:08:56.320
And we later learned from some
imagery that came in that there

00:08:56.320 --> 00:09:01.096
had been a landslide dam there
with a pond about 75 meters long,

00:09:01.120 --> 00:09:06.668
but that had breached by the
time that they had flown over it.

00:09:07.920 --> 00:09:12.000
So this map here shows kind of the
overview of all of the landslides

00:09:12.000 --> 00:09:15.600
that we mapped and the ones that
we flagged of potential concern.

00:09:15.600 --> 00:09:19.016
So the flagged ones
are shown with red triangles.

00:09:19.040 --> 00:09:23.840
And those include some landslide
dams with ponding, but also some

00:09:23.840 --> 00:09:29.096
that are close to populated
areas and close to roads.

00:09:29.120 --> 00:09:33.120
You can see we also flagged lots of
other landslides as landslide dams in

00:09:33.120 --> 00:09:39.416
yellow and road hazards in purple that
we did not flag as of potential concern.

00:09:39.440 --> 00:09:42.320
But interestingly, even though
most landslides were in the west,

00:09:42.320 --> 00:09:46.720
a lot of the flagged landslides were
in the eastern 2/3 of the affected area,

00:09:46.720 --> 00:09:50.503
which is also where more
people happen to live.

00:09:51.600 --> 00:09:56.856
So it’s pretty difficult to assess human
impacts from satellite imagery,

00:09:56.880 --> 00:10:02.800
so we do know from news reports
that there was undoubtedly a large toll

00:10:02.800 --> 00:10:04.560
due to the landslides
triggered by this earthquake.

00:10:04.560 --> 00:10:08.720
There were definitely some fatalities,
and a lot of people lost their livelihoods,

00:10:08.720 --> 00:10:11.360
lots their farms,
lost their buildings, and so on.

00:10:11.360 --> 00:10:14.640
But only some of this is really visible
in the satellite imagery,

00:10:14.640 --> 00:10:18.880
like the example at right here.
So one way to just roughly estimate

00:10:18.880 --> 00:10:22.880
where the human impacts were greatest
is to basically find the landslides that

00:10:22.880 --> 00:10:27.360
have populations living near them.
So this plot here is showing landslides

00:10:27.360 --> 00:10:32.800
that have more than 50 people within
a 200-meter radius of the landslide.

00:10:32.800 --> 00:10:37.920
So it’s just a proxy for human impacts,
but this kind of shows that the human

00:10:37.920 --> 00:10:41.920
impacts – even though most landslides
were in the west, actually most of

00:10:41.920 --> 00:10:44.240
the human impacts were
probably more towards the center

00:10:44.240 --> 00:10:48.296
and the eastern parts
of the affected area.

00:10:48.320 --> 00:10:52.376
So the human impacts from this
event from the landslides triggered by –

00:10:52.400 --> 00:10:55.120
you know, may not be over.
And one of the reasons is because

00:10:55.120 --> 00:10:58.400
of these cascading hazards.
So one of them that I’ve mentioned

00:10:58.400 --> 00:11:01.520
a few times is the landslide dams.
And there were quite a few landslide

00:11:01.520 --> 00:11:04.800
dams formed by this event. And we
know there were at least 20 that have

00:11:04.800 --> 00:11:09.120
observed pooling behind them.
And these can catastrophically fail,

00:11:09.120 --> 00:11:12.000
and that can cause flooding –
rapid flooding without warning

00:11:12.000 --> 00:11:15.016
downstream if these
aren’t monitored.

00:11:15.040 --> 00:11:19.760
And so this is one ongoing
hazard to keep an eye on.

00:11:20.560 --> 00:11:26.080
And then, as I alluded to earlier as well,
a lot of these channels, especially in

00:11:26.080 --> 00:11:29.520
Pic Macaya, but also elsewhere,
are just filled with sediment from

00:11:29.520 --> 00:11:33.040
these landslides now. And so that
really changes how the river behaves,

00:11:33.040 --> 00:11:38.160
and it can increase not only flooding
risk by changing how the water flows,

00:11:38.160 --> 00:11:42.640
but also this debris can be mobilized
into debris flows, and that can reach

00:11:42.640 --> 00:11:48.776
areas downstream and can cause
damage and other problems.

00:11:48.800 --> 00:11:53.360
And so we didn’t do any, like,
sophisticated modeling or anything,

00:11:53.360 --> 00:11:56.640
but this plot here is just basically
taking the river network and

00:11:56.640 --> 00:12:00.233
showing the areas that have
the most landslides along them.

00:12:00.233 --> 00:12:04.056
So the areas of river with
the most sedimentation.

00:12:04.080 --> 00:12:10.160
And so you can see that, you know,
the western third is probably the highest

00:12:10.160 --> 00:12:14.560
area of concern, but there’s actually,
you know, channels in the

00:12:14.560 --> 00:12:17.680
eastern 2/3 that may also
have increased debris flow

00:12:17.680 --> 00:12:20.928
and flooding hazards
for the years to come.

00:12:22.480 --> 00:12:28.000
So, as you all know, there was
a major earthquake in 2010 in Haiti

00:12:28.000 --> 00:12:32.640
that was further east, and that
also triggered a lot of landslides.

00:12:32.640 --> 00:12:36.720
And so this is just a quick
comparison between the two.

00:12:36.720 --> 00:12:40.000
It’s difficult to compare them directly
because we have a research-grade

00:12:40.000 --> 00:12:45.440
inventory of almost 24,000 landslides
mapped for the 2010 earthquake,

00:12:45.440 --> 00:12:49.176
and we don’t have that yet
for the 2021 earthquake.

00:12:49.200 --> 00:12:53.120
But, to more directly compare them,
I’m showing just the large landslides

00:12:53.120 --> 00:12:56.560
from the 2010 inventory
at the right compared to

00:12:56.560 --> 00:13:00.856
our inventory from
the 2021 earthquake.

00:13:00.880 --> 00:13:04.800
You can see that they’re, you know,
fairly similar in number when you

00:13:04.800 --> 00:13:10.320
consider only the large landslides,
and so they’re pretty similar, you know,

00:13:10.320 --> 00:13:14.640
earthquake-triggered landslide events,
except you can see that the

00:13:14.640 --> 00:13:20.080
2021 event is – kind of has areas
of higher concentrations

00:13:20.080 --> 00:13:24.928
of landslides than the
2020 earthquake did.

00:13:25.600 --> 00:13:30.240
So, in conclusion, the 2021
Haiti earthquake was a major

00:13:30.240 --> 00:13:34.936
landslide-triggering event, and
it had substantial human impacts.

00:13:34.960 --> 00:13:39.680
And basically, the hazard outlook
in the region has been changed

00:13:39.680 --> 00:13:44.936
because of these landslides,
likely for years to come.

00:13:44.960 --> 00:13:48.640
We found that sharing information
in these daily reports with the

00:13:48.640 --> 00:13:54.960
humanitarian groups and aid agencies
was really important, and especially

00:13:54.960 --> 00:13:58.160
writing those in plain English
were really useful, and we really

00:13:58.160 --> 00:14:03.336
benefited because Stephen Slaughter,
one of the co-authors of this work,

00:14:03.360 --> 00:14:09.520
is half employed by USAID. And so
he’s kind of embedded with them,

00:14:09.520 --> 00:14:12.536
and that made it a lot
easier to share information.

00:14:12.560 --> 00:14:15.040
We also saw that, you know,
there’s a lot that you can learn

00:14:15.040 --> 00:14:18.800
from a remote response, but it really
leaves a lot to be desired because

00:14:18.800 --> 00:14:22.696
you don’t really know exactly
what’s going on a lot of the time.

00:14:22.720 --> 00:14:27.040
You really need field investigations,
and I know that some Haitian

00:14:27.040 --> 00:14:34.320
colleagues and also a team organized
by the Geotechnical Extreme Events

00:14:34.320 --> 00:14:40.296
Reconnaissance group have been doing
some additional field investigations.

00:14:40.320 --> 00:14:45.200
And then I also hope that you see
how the near-real-time ground failure

00:14:45.200 --> 00:14:50.240
products that are now available can
be used to kind of help inform what

00:14:50.240 --> 00:14:52.216
happens right after
an earthquake.

00:14:52.240 --> 00:14:57.680
And so, with that, I’ll conclude
and I’ll take any questions.