1
00:00:00,000 --> 00:00:04,560
all right hi everyone my name is Gigi
Pavur and I'm an undergraduate senior

2
00:00:04,560 --> 00:00:09,300
at Georgia Tech majoring in Earth that
Atmospheric Sciences I'm so excited to

3
00:00:09,300 --> 00:00:13,019
be here today with you all at AGU
to tell you about a project I did this

4
00:00:13,020 --> 00:00:17,609
summer through NASA DEVELOP, so NASA
DEVELOP is a program within the Applied

5
00:00:17,609 --> 00:00:22,940
Sciences program at NASA which is
designed to increase participants like

6
00:00:22,949 --> 00:00:27,150
me. Our capacity to use NASA earth
observations so we do 10-week

7
00:00:27,150 --> 00:00:32,460
feasibility projects, so with the help of
my teammates Chelsey Dandridge, Robyn Kim

8
00:00:32,460 --> 00:00:37,680
and Sarah Aldama, um, we were able to use
NASA earth observations to improve

9
00:00:37,680 --> 00:00:41,340
landslides situational awareness in the
Dominican Republic

10
00:00:49,960 --> 00:00:52,880
okay so imagine that

11
00:00:52,890 --> 00:00:56,460
we are three thousand three hundred and
eighty nine miles away from where you're

12
00:00:56,460 --> 00:01:01,170
sitting right now at AGU you'll find
a small country on the island of

13
00:01:01,170 --> 00:01:06,299
Hispaniola and the Caribbean Sea the
Dominican Republic so with its location

14
00:01:06,299 --> 00:01:11,280
in the hurricane belt of the tropics the
Dominican Republic is frequently exposed

15
00:01:11,280 --> 00:01:15,030
to very heavy rainfall events which can
trigger landslides

16
00:01:15,030 --> 00:01:19,950
so my DEVELOP team partnered with a
geological service in the Dominican

17
00:01:19,950 --> 00:01:24,270
Republic to help improve their
situational awareness of landslides and

18
00:01:24,270 --> 00:01:27,960
our project partners were really
interested in understanding how to limit

19
00:01:27,960 --> 00:01:32,369
the loss of life and protect
infrastructure for landslides and then

20
00:01:32,369 --> 00:01:36,450
also our project supported an
international agreement between the NASA

21
00:01:36,450 --> 00:01:42,210
earth science division and the Central
American integration system and that

22
00:01:42,210 --> 00:01:46,439
agreement was designed to increase
collaborations between NASA and Central

23
00:01:46,439 --> 00:01:50,340
American countries, so I was really
exciting to get to work on a project of

24
00:01:50,340 --> 00:01:55,259
this international scale and calibre and
then also I'd like to point out that I

25
00:01:55,259 --> 00:02:03,210
too went bananas when I saw that fruit
also has landslide risk so NASA has

26
00:02:03,210 --> 00:02:07,619
about 30 satellites that are constantly
orbiting Earth and collecting

27
00:02:07,619 --> 00:02:11,250
environmental data and this data is
really useful to study environmental

28
00:02:11,250 --> 00:02:15,730
challenges like landslides
so for a project we use two earth

29
00:02:15,730 --> 00:02:20,620
observations the first was from the
shuttle radar topography mission or SRTM

30
00:02:20,620 --> 00:02:26,650
which provides elevation data and we use
that to determine slope and then we also

31
00:02:26,650 --> 00:02:30,880
used global precipitation measurement
from GPM which is the satellite right

32
00:02:30,880 --> 00:02:32,220
here

33
00:02:33,860 --> 00:02:39,060
okay so landslide experts at Goddard
Research Center at Space Flight Research

34
00:02:39,080 --> 00:02:44,340
Center created a model called LHASA
and yes you guessed it LHASA is another

35
00:02:44,340 --> 00:02:48,610
NASA acronym so it stands for a
Landslide Hazard Assessment for

36
00:02:48,610 --> 00:02:53,769
Situational Awareness so this model is
publicly available online through github

37
00:02:53,769 --> 00:03:00,579
as both a Python script and an R script
so my team worked to use this model and

38
00:03:00,579 --> 00:03:04,630
take it from a global scale and cater it
specifically for the Dominican

39
00:03:04,630 --> 00:03:08,590
Republic and we did this by
incorporating local data from our

40
00:03:08,590 --> 00:03:10,340
project partners

41
00:03:12,960 --> 00:03:16,500
so a really important
component of the LHASA model is a

42
00:03:16,500 --> 00:03:21,660
susceptibility map so this shows the
landslide susceptibility in the

43
00:03:21,670 --> 00:03:27,340
Dominican Republic and my team created
this by using a fuzzy overlay model to

44
00:03:27,340 --> 00:03:32,519
determine the correlations between
landslide factors like elevation slope

45
00:03:32,519 --> 00:03:36,519
distance to active fault lines clay
percentage lithology

46
00:03:36,519 --> 00:03:42,700
and forest covered loss and so when the
static map is combined in LHASA with

47
00:03:42,700 --> 00:03:52,720
precipitation measurements it will
output near real-time nowcast so when

48
00:03:52,720 --> 00:03:57,819
the rainfall the historical rainfall
exceeds the 95th percentile it will

49
00:03:57,819 --> 00:04:01,569
issue a high hazard nowcast
if your susceptibility map says it's a

50
00:04:01,569 --> 00:04:08,049
very susceptible region to landslides
so since GPM rainfall data is available

51
00:04:08,049 --> 00:04:12,220
every 30 minutes you could realistically
make new outputs for this every half

52
00:04:12,220 --> 00:04:21,099
hour so you might be wondering
landslides are a natural process for

53
00:04:21,099 --> 00:04:25,000
altering topography so why are they
considered a natural disaster what makes

54
00:04:25,000 --> 00:04:28,220
them a natural disaster
and what I learned through this project

55
00:04:28,220 --> 00:04:32,720
is that it really becomes a natural
disaster when it affects people so once

56
00:04:32,720 --> 00:04:35,990
again our project partners were really
interested in knowing where landslides

57
00:04:35,990 --> 00:04:40,640
impact people so we created this
bivariate map on which depicts the

58
00:04:40,640 --> 00:04:48,080
intersection of landslide susceptibility
and and population so the regions but

59
00:04:48,080 --> 00:04:53,840
you want to pay attention to are this
deep red color up there which represent

60
00:04:53,840 --> 00:04:57,680
high population and high landslide
susceptibility and now fortunately

61
00:04:57,680 --> 00:05:00,920
you'll see there aren't actually that
many pixels of that color which is a

62
00:05:00,920 --> 00:05:07,130
good thing and then you can see as well
but Santiago and Santo Domingo and then

63
00:05:07,130 --> 00:05:10,880
Bayer hona all have high populations
because those are their main cities but

64
00:05:10,880 --> 00:05:15,980
they aren't located in susceptible
regions the central area has very high

65
00:05:15,980 --> 00:05:19,280
landslide susceptibility but not a lot
of people living there because it's

66
00:05:19,280 --> 00:05:24,560
actually a national park so the idea is
that this bivariant map can be used fore

67
00:05:24,560 --> 00:05:29,300
quick heuristic identifications of
regions on where people are at risk

68
00:05:32,620 --> 00:05:37,600
all right then lastly a big final outcome of
our project was we wanted to increase

69
00:05:37,620 --> 00:05:42,440
our project partners capacity to
actually use NASA the earth observations

70
00:05:42,440 --> 00:05:46,280
so we wrote a tutorial so that they
could do everything that we did over the

71
00:05:46,280 --> 00:05:52,640
summer and one thing that we also showed
them was how mapping landslides from

72
00:05:52,640 --> 00:05:57,080
space can help them better understand
where landslides occur even if they

73
00:05:57,080 --> 00:06:01,940
don't have the budget necessarily to do
ground mapping or like field research so

74
00:06:01,940 --> 00:06:07,220
the three images on the right wait
left so an interesting story about a

75
00:06:07,220 --> 00:06:11,270
landslide that happened so we have a
forest with the road going through it in

76
00:06:11,270 --> 00:06:18,950
2012 and then by 2014 wait 2016 all the
trees are gone and then by 2018 we have

77
00:06:18,950 --> 00:06:23,420
evidence of a landslide so once again
our project partners can use this to go

78
00:06:23,420 --> 00:06:27,350
back and improve their historical
landslide inventory and then our

79
00:06:27,350 --> 00:06:29,780
tutorial also went through how to use
LHASA

80
00:06:29,780 --> 00:06:34,520
and how to make their own susceptibility
map and then lastly it was translated

81
00:06:34,520 --> 00:06:38,419
into Spanish and shared with other
Central American countries so that they

82
00:06:38,420 --> 00:06:42,840
can hopefully use this as a start
point to use LHASA in their country

83
00:06:45,420 --> 00:06:46,760
okay

84
00:06:47,840 --> 00:06:54,580
,um so I'd like to end with a quote that
was in the NASA calendar last year which

85
00:06:54,590 --> 00:06:58,720
judging from the line over there I'm
sure you guys are all familiar with it

86
00:06:58,720 --> 00:07:03,350
but the quote I think it perfectly
captures the value of applying NASA

87
00:07:03,350 --> 00:07:07,370
earth observations to combat
environmental challenges so the quote is

88
00:07:07,370 --> 00:07:12,400
from Nelson Mandela and it says action
without vision is only passing time

89
00:07:12,400 --> 00:07:17,600
vision with action is merely daydreaming
but vision with action can change the

90
00:07:17,600 --> 00:07:20,300
world
so quite literally earth observations

91
00:07:20,300 --> 00:07:24,320
give us the perspective and vision we
need to take action and according to

92
00:07:24,320 --> 00:07:28,370
Nelson Mandela that's how we change the
world so here we can see the night

93
00:07:28,370 --> 00:07:33,050
lights of Earth and we know that from
this perspective of space. Disasters

94
00:07:33,050 --> 00:07:37,760
don't they aren't confined by country
borders and from my DEVELOP project this

95
00:07:37,760 --> 00:07:42,080
summer I learned that the applications
of earth observations can also transcend

96
00:07:42,080 --> 00:07:51,560
these boundaries so lastly thank you to
the moon and back to all the people who

97
00:07:51,560 --> 00:07:56,180
helped me do this project including our
project partners in the Dominican

98
00:07:56,180 --> 00:07:59,210
Republic
our science advisors Dr. Ross and Dr.

99
00:07:59,210 --> 00:08:04,220
Dalia Kirschbaum and then of course to
my wonderful team Sarah and Robin and

100
00:08:04,220 --> 00:08:08,630
Chelsea they put up with a lot of
landslides jokes for me which as you

101
00:08:08,630 --> 00:08:12,850
know is a very slippery slope thank you