1
00:00:05,038 --> 00:00:06,039
If we were to see the

2
00:00:06,039 --> 00:00:08,475
world with polarization-sensitive eyes,

3
00:00:09,075 --> 00:00:10,643
the sky would not be blue.

4
00:00:11,611 --> 00:00:13,079
Grass would look gray.

5
00:00:13,380 --> 00:00:16,383
There would be all sorts of strange things
that would be happening.

6
00:00:16,616 --> 00:00:18,551
What we reveal about

7
00:00:18,551 --> 00:00:20,820
the environment with polarization

8
00:00:20,820 --> 00:00:23,156
is really kind
of another dimension of information.

9
00:00:25,091 --> 00:00:28,695
The PACE mission holds the keys
to unlock that dimension

10
00:00:28,695 --> 00:00:30,930
with two toaster-sized instruments

11
00:00:30,930 --> 00:00:32,832
called polarimeters,

12
00:00:32,932 --> 00:00:36,736
and polarimeters like these measure
the polarization of sunlight.

13
00:00:36,803 --> 00:00:41,341
Generally, sunlight has a combination
of different directions.

14
00:00:41,341 --> 00:00:46,479
Polarization is some preference
for an oscillation direction.

15
00:00:46,880 --> 00:00:49,449
The ability to detect the specific direction

16
00:00:49,449 --> 00:00:52,118
sunlight reflects back to PACE’s instruments

17
00:00:52,118 --> 00:00:53,553
will give us more information

18
00:00:53,553 --> 00:00:57,657
about clouds and tiny
atmospheric particles called aerosols.

19
00:00:58,158 --> 00:01:01,895
The aerosols are really important
to human health,

20
00:01:01,961 --> 00:01:05,131
so that's why we need to really quantify
what is out there,

21
00:01:05,131 --> 00:01:08,401
like what type of aerosols there are
and where they come from.

22
00:01:08,835 --> 00:01:12,372
Various interactions
with light in the environment,

23
00:01:12,372 --> 00:01:17,677
scattering events off of particles
or surfaces can impose some preference

24
00:01:17,677 --> 00:01:22,382
in the light that they reflect
in terms of the polarization nature.

25
00:01:23,516 --> 00:01:24,717
The two multi-angle

26
00:01:24,717 --> 00:01:27,187
polarimeters were built by NASA's partners

27
00:01:27,187 --> 00:01:29,022
both here and abroad.

28
00:01:29,155 --> 00:01:32,025
The Hyper-Angular Rainbow Polarimeter #2,

29
00:01:32,025 --> 00:01:33,159
or HARP2,

30
00:01:33,159 --> 00:01:34,994
will measure atmospheric particles

31
00:01:34,994 --> 00:01:36,729
in one of its spectral channels

32
00:01:36,729 --> 00:01:38,765
in up to 60 viewing angles.

33
00:01:39,199 --> 00:01:40,800
Why so many angles?

34
00:01:40,934 --> 00:01:43,169
This is like a camera,
like any other kind of camera.

35
00:01:43,169 --> 00:01:45,905
But instead of taking a picture at

36
00:01:45,905 --> 00:01:48,675
one particular geometry of what

37
00:01:48,675 --> 00:01:50,376
we would understand as light,

38
00:01:50,376 --> 00:01:53,179
it's looking at a scene from different angles.

39
00:01:53,246 --> 00:01:55,782
We will move the different angles

40
00:01:55,782 --> 00:01:58,251
to the one single location,

41
00:01:58,251 --> 00:01:58,918
and in that way

42
00:01:58,918 --> 00:02:02,555
we will collect the information at all
the different angles.

43
00:02:02,755 --> 00:02:06,326
And those different angles
contain information about

44
00:02:06,326 --> 00:02:08,228
what's present in the environment.

45
00:02:09,028 --> 00:02:11,965
For instance,
all these angles from HARP2 can analyze

46
00:02:11,965 --> 00:02:14,033
the elusive cloud bow.

47
00:02:19,105 --> 00:02:21,207
Cloud bows are slightly distinct
from a rainbow.

48
00:02:21,207 --> 00:02:24,544
Rainbow is light
scattering off of rain droplets.

49
00:02:24,544 --> 00:02:26,980
Cloud bows is light
scattering off of cloud droplets, which

50
00:02:26,980 --> 00:02:28,114
are a little bit smaller.

51
00:02:28,882 --> 00:02:29,916
By being able to

52
00:02:29,916 --> 00:02:33,019
observe cloud bows with polarization,

53
00:02:33,019 --> 00:02:36,723
if we very accurately measure
the geometry in which this happens,

54
00:02:36,723 --> 00:02:41,528
the exact position of that cloud bow with
respect to the Sun, in our observation,

55
00:02:41,528 --> 00:02:44,931
it tells us a lot about the size
distribution of the cloud droplets.

56
00:02:44,998 --> 00:02:47,100
If we understand
the size distribution of cloud droplets,

57
00:02:47,100 --> 00:02:50,270
we can understand things
about the formation of clouds and how long

58
00:02:50,270 --> 00:02:53,606
they will persist if they're going
to turn into precipitation or not.

59
00:02:53,673 --> 00:02:57,110
Polarization can also reveal the shape of sunglint,

60
00:02:57,110 --> 00:03:01,014
the pattern of sunlight reflecting
directly off the ocean's surface.

61
00:03:01,648 --> 00:03:05,785
Sunglint patterns can tell us
how rough or smooth the ocean's surface is,

62
00:03:06,152 --> 00:03:09,122
which can determine wind speed at the surface.

63
00:03:10,323 --> 00:03:12,358
Clouds also have an impact on climate.

64
00:03:12,358 --> 00:03:14,527
But the interaction between the two,

65
00:03:14,527 --> 00:03:15,995
there's many pathways in which aerosols

66
00:03:15,995 --> 00:03:17,397
can interact with clouds.

67
00:03:17,397 --> 00:03:20,400
Cloud droplets can form around
aerosol particles more easily

68
00:03:20,567 --> 00:03:23,203
and other things that are going on
in the local situation.

69
00:03:23,203 --> 00:03:26,673
That complexity of the interaction between
the two is one of the largest sources

70
00:03:26,673 --> 00:03:29,142
of uncertainty in understanding our global climate,

71
00:03:29,142 --> 00:03:31,411
and that's why we're making these measurements.

72
00:03:31,411 --> 00:03:35,515
The data from PACE will allow researchers
to tease out the species of aerosols,

73
00:03:35,515 --> 00:03:39,519
which will help fine tune climate models
so they make better predictions.

74
00:03:40,253 --> 00:03:44,457
PACE’s other polarimeter, SPEXone,
will tackle aerosol retrievals

75
00:03:44,557 --> 00:03:47,427
and give us precise measurements
of the angle, degree

76
00:03:47,427 --> 00:03:49,696
and intensity of polarization.

77
00:03:50,196 --> 00:03:54,200
But processing the sheer volume of data has been its own mission.

78
00:03:54,234 --> 00:03:59,572
Each pixel of data the polarimeters
measure covers about five kilometers square.

79
00:03:59,572 --> 00:04:04,077
In that space are hundreds—even thousands—
of observations at different angles,

80
00:04:04,077 --> 00:04:06,746
wavelengths and state of polarization.

81
00:04:06,746 --> 00:04:09,415
In the course of one full day of orbits

82
00:04:09,415 --> 00:04:11,017
those pixels pile up.

83
00:04:11,017 --> 00:04:12,685
If you put them together,

84
00:04:12,685 --> 00:04:16,122
there will be more than ten million pixels.

85
00:04:16,122 --> 00:04:21,094
That's a huge challenge on both storage
and the computational power.

86
00:04:21,094 --> 00:04:22,061
To meet that challenge,

87
00:04:22,061 --> 00:04:24,931
the PACE team has turned
to a kind of machine learning

88
00:04:24,931 --> 00:04:27,600
called a neural network emulator.

89
00:04:27,600 --> 00:04:29,802
Even before PACE gathers any data,

90
00:04:29,802 --> 00:04:31,371
the emulator has been trained with

91
00:04:31,371 --> 00:04:33,506
millions of simulations of the possible

92
00:04:33,506 --> 00:04:36,409
atmospheric conditions in that one pixel.

93
00:04:37,443 --> 00:04:40,146
With this emulator,
what would take an hour for one

94
00:04:40,146 --> 00:04:42,415
pixel is now a matter of milliseconds,

95
00:04:42,415 --> 00:04:44,717
allowing PACE to process a seemingly

96
00:04:44,717 --> 00:04:45,985
endless stream of data

97
00:04:45,985 --> 00:04:47,053
for the mission

98
00:04:47,053 --> 00:04:49,389
and atmospheric researchers all over.

99
00:04:50,056 --> 00:04:52,292
They will require a lot of measurement,

100
00:04:52,292 --> 00:04:56,562
especially if we can do that
from a global scale with satellites

101
00:04:56,562 --> 00:04:59,666
so we know where they're coming from
so we can trace their source.

102
00:04:59,666 --> 00:05:02,969
We probably can help to reduce
its impact on human health.