1
00:00:01,080 --> 00:00:02,250
Hello everybody.

2
00:00:02,250 --> 00:00:03,930
This week I'm gonna be introducing you

3
00:00:03,930 --> 00:00:05,970
to data types in remote sensing

4
00:00:05,970 --> 00:00:10,530
and going over some basics
of image visualization.

5
00:00:10,530 --> 00:00:12,720
Specifically, our goals
for the week are to be able

6
00:00:12,720 --> 00:00:17,280
to define what vector
and raster datasets are,

7
00:00:17,280 --> 00:00:18,960
and also to understand the difference

8
00:00:18,960 --> 00:00:22,170
between true color and false color images.

9
00:00:22,170 --> 00:00:25,260
You'll also learn how
you might make true color

10
00:00:25,260 --> 00:00:27,933
and false color images and
how you interpret them.

11
00:00:30,090 --> 00:00:32,910
Just as a review, last
week we talked about

12
00:00:32,910 --> 00:00:36,840
what tools humans use to
remotely sense our world.

13
00:00:36,840 --> 00:00:38,760
We reviewed different kinds of resolution

14
00:00:38,760 --> 00:00:41,910
including spatial resolution,
temporal resolution,

15
00:00:41,910 --> 00:00:45,270
radiometric resolution,
and spectral resolution.

16
00:00:45,270 --> 00:00:47,670
And I also describe the
difference between active

17
00:00:47,670 --> 00:00:49,320
and passive sensors.

18
00:00:49,320 --> 00:00:53,610
So remember that passive
sensors use the sun's energy

19
00:00:53,610 --> 00:00:57,360
to measure electromagnetic
energy or waves,

20
00:00:57,360 --> 00:01:01,290
and active sensors use artificial energy

21
00:01:01,290 --> 00:01:03,450
like laser pulses.

22
00:01:03,450 --> 00:01:05,130
In terms of passive sensors,

23
00:01:05,130 --> 00:01:08,913
I discussed the nine different
Landsat satellite missions,

24
00:01:09,750 --> 00:01:12,360
how commonly used they are,

25
00:01:12,360 --> 00:01:14,340
and I also discussed how they differ

26
00:01:14,340 --> 00:01:17,790
in terms of those four resolutions.

27
00:01:17,790 --> 00:01:19,800
And then I also mentioned the modus

28
00:01:19,800 --> 00:01:21,480
and Sentinel-2 satellites,

29
00:01:21,480 --> 00:01:23,910
which are passive sensors as well.

30
00:01:23,910 --> 00:01:27,960
And then finally, I talked
about one active sensor method,

31
00:01:27,960 --> 00:01:30,120
that's called LiDAR.

32
00:01:30,120 --> 00:01:32,580
And we can use LiDAR to
measure the height of buildings

33
00:01:32,580 --> 00:01:36,180
or vegetation, and it can
also be used to detect items

34
00:01:36,180 --> 00:01:41,180
that are underwater or
even under forest canopies.

35
00:01:43,800 --> 00:01:47,010
So now let's talk about different
kinds of geospatial data

36
00:01:47,010 --> 00:01:49,920
that we use in remote sensing and GIS.

37
00:01:49,920 --> 00:01:52,200
And this is probably
review for those of us

38
00:01:52,200 --> 00:01:54,183
who have taken GIS classes before.

39
00:01:55,470 --> 00:01:56,580
One kind of spatial data

40
00:01:56,580 --> 00:01:59,340
that we use a lot is called vector data.

41
00:01:59,340 --> 00:02:03,420
Vector data can be
points, lines or polygons,

42
00:02:03,420 --> 00:02:06,270
and they represent real world features.

43
00:02:06,270 --> 00:02:11,040
For example, a point can
represent the capital

44
00:02:11,040 --> 00:02:14,883
of a state or a point
could represent a tree.

45
00:02:15,750 --> 00:02:20,550
A point is a single vertex corresponding

46
00:02:20,550 --> 00:02:23,850
to a latitude and a longitude.

47
00:02:23,850 --> 00:02:25,620
For example, a line

48
00:02:25,620 --> 00:02:30,620
is two vertices, which are two points

49
00:02:32,970 --> 00:02:35,670
that are connected via this line.

50
00:02:35,670 --> 00:02:39,450
Lines can represent streets,
they can represent a distance

51
00:02:39,450 --> 00:02:41,100
between two trees.

52
00:02:41,100 --> 00:02:44,190
And then we have polygons,
which are made up

53
00:02:44,190 --> 00:02:48,990
of at least four vertices, and
they form an enclosed area.

54
00:02:48,990 --> 00:02:51,030
So polygons could represent

55
00:02:51,030 --> 00:02:52,650
political boundaries like a state.

56
00:02:52,650 --> 00:02:57,650
They could represent the
area of a tree canopy,

57
00:02:57,690 --> 00:03:00,210
they could represent a house.

58
00:03:00,210 --> 00:03:05,107
And vector features can
also have attribute data

59
00:03:06,390 --> 00:03:07,680
that describes it as well.

60
00:03:07,680 --> 00:03:12,090
So if you have 100 points,
you could connect data

61
00:03:12,090 --> 00:03:15,360
to those points to say
some of these points

62
00:03:15,360 --> 00:03:17,610
represent trees of this species,

63
00:03:17,610 --> 00:03:19,800
some of them represent
trees of this species,

64
00:03:19,800 --> 00:03:21,153
et cetera, et cetera.

65
00:03:23,070 --> 00:03:24,510
In terms of vector maps,

66
00:03:24,510 --> 00:03:29,510
so some maps are completely
made up of vector data.

67
00:03:29,580 --> 00:03:32,940
So, for example,
topographic maps, which show

68
00:03:32,940 --> 00:03:37,940
how elevated land is, those
are made up of vector data.

69
00:03:38,130 --> 00:03:39,900
Political maps, so for example,

70
00:03:39,900 --> 00:03:43,680
a political map would be
like a map of the world

71
00:03:43,680 --> 00:03:45,840
that shows country borders

72
00:03:45,840 --> 00:03:49,450
or I've shown here a map of
the United States that shows

73
00:03:50,490 --> 00:03:53,520
polygons for each state.

74
00:03:53,520 --> 00:03:55,530
And then also you can see point data

75
00:03:55,530 --> 00:03:58,530
representing state capitals.

76
00:03:58,530 --> 00:04:01,773
That's vector data, street
maps, that's also vector data.

77
00:04:04,770 --> 00:04:08,070
The other main kind of data
in remote sensing and GIS

78
00:04:08,070 --> 00:04:09,840
is called raster data.

79
00:04:09,840 --> 00:04:14,340
Raster organize information
into cells or pixels.

80
00:04:14,340 --> 00:04:16,710
Pixels are square like,

81
00:04:16,710 --> 00:04:19,560
and they're uniform in
size throughout the image.

82
00:04:19,560 --> 00:04:21,390
And they're also organized into rows

83
00:04:21,390 --> 00:04:24,450
and columns that form a grid.

84
00:04:24,450 --> 00:04:29,250
Remember that the size of the
pixels basically is the same

85
00:04:29,250 --> 00:04:31,170
as spatial resolution,

86
00:04:31,170 --> 00:04:33,120
which we talked about last week.

87
00:04:33,120 --> 00:04:36,870
In raster data, each
pixel contains a value

88
00:04:36,870 --> 00:04:39,360
that represents some information.

89
00:04:39,360 --> 00:04:42,810
The information that
that raster cell contains

90
00:04:42,810 --> 00:04:46,470
could be continuous data,
meaning that it's numerical

91
00:04:46,470 --> 00:04:50,520
and it represents how much of
a certain variable is present.

92
00:04:50,520 --> 00:04:52,740
An example of this would be

93
00:04:52,740 --> 00:04:55,620
if a raster represented
the percent snow coverage

94
00:04:55,620 --> 00:04:59,280
on the landscape or
temperature or elevation.

95
00:04:59,280 --> 00:05:02,850
Reflectance values are
also continuous data.

96
00:05:02,850 --> 00:05:07,740
So a raster can describe
how much reflectance

97
00:05:07,740 --> 00:05:11,160
in the red band there is on a landscape.

98
00:05:11,160 --> 00:05:13,830
Raster data can also be categorical,

99
00:05:13,830 --> 00:05:17,130
meaning that the numbers don't represent

100
00:05:17,130 --> 00:05:19,050
how much of something there is,

101
00:05:19,050 --> 00:05:22,440
but rather they represent certain classes.

102
00:05:22,440 --> 00:05:24,540
So a land cover map is an example

103
00:05:24,540 --> 00:05:27,900
of a categorical raster map.

104
00:05:27,900 --> 00:05:29,940
Even though the raster may be assigned

105
00:05:29,940 --> 00:05:34,940
numbers like 1, 2, 3, 4,
those numbers don't represent

106
00:05:35,580 --> 00:05:37,230
the actual numbers themselves

107
00:05:37,230 --> 00:05:38,880
or how much there is of something.

108
00:05:38,880 --> 00:05:42,270
But they represent classes
like one may represent water,

109
00:05:42,270 --> 00:05:44,610
two may represent forest land,

110
00:05:44,610 --> 00:05:46,680
three may represent urban land,

111
00:05:46,680 --> 00:05:49,473
and four may represent wetlands.

112
00:05:51,900 --> 00:05:56,250
Here are three examples of
raster maps that you could make.

113
00:05:56,250 --> 00:05:58,020
So the first is a surface map.

114
00:05:58,020 --> 00:06:02,400
And this map is interesting
because it is a raster map,

115
00:06:02,400 --> 00:06:05,070
but there's also some vector
data overlaid on top of it.

116
00:06:05,070 --> 00:06:07,770
So there's some vector line data,

117
00:06:07,770 --> 00:06:10,560
but the surface map underneath is a,

118
00:06:10,560 --> 00:06:11,820
contains continuous data.

119
00:06:11,820 --> 00:06:14,700
So it contains reflectance values

120
00:06:14,700 --> 00:06:17,400
and it's a true color map.

121
00:06:17,400 --> 00:06:20,310
So it contains, we'll talk
about this in a few slides,

122
00:06:20,310 --> 00:06:24,030
but it contains information
from the red, green,

123
00:06:24,030 --> 00:06:26,430
and blue bands.

124
00:06:26,430 --> 00:06:29,940
And then thematic maps can
be continuous or categorical.

125
00:06:29,940 --> 00:06:34,470
So this middle thematic
map represents elevation.

126
00:06:34,470 --> 00:06:35,970
And so the pixels represent

127
00:06:35,970 --> 00:06:39,270
how much elevation there is in each pixels

128
00:06:39,270 --> 00:06:41,400
and they're colored accordingly.

129
00:06:41,400 --> 00:06:44,700
And then finally, the map on
the right is a thematic map.

130
00:06:44,700 --> 00:06:48,240
It can also be called an
attribute map that is categorical.

131
00:06:48,240 --> 00:06:50,760
So the pixels represent certain classes.

132
00:06:50,760 --> 00:06:53,760
For example, the brown
pixels represent agriculture.

133
00:06:53,760 --> 00:06:57,510
The yellow pixels represent
bare ground, et cetera.

134
00:06:57,510 --> 00:07:00,450
In this class, you will learn how to make

135
00:07:00,450 --> 00:07:02,493
all three of these maps.

136
00:07:04,890 --> 00:07:07,410
Let's go back and talk in more detail

137
00:07:07,410 --> 00:07:09,600
about spatial resolution.

138
00:07:09,600 --> 00:07:11,100
So remember from last time

139
00:07:11,100 --> 00:07:16,100
that high spatial resolution
data has small pixel sizes

140
00:07:16,230 --> 00:07:20,343
and low spatial resolution
data has large pixel sizes.

141
00:07:21,450 --> 00:07:23,820
I also talked about
the spatial resolutions

142
00:07:23,820 --> 00:07:26,130
of some common satellite sensors.

143
00:07:26,130 --> 00:07:27,810
For example, Sentinel-2

144
00:07:27,810 --> 00:07:30,360
has a spatial resolution of 10 meters.

145
00:07:30,360 --> 00:07:33,240
Most of the Landsat sensors
have a spatial resolution

146
00:07:33,240 --> 00:07:36,180
of 30 meters, and MODIS
has a spatial resolution

147
00:07:36,180 --> 00:07:39,300
of 250 meters to 1,000 meters.

148
00:07:39,300 --> 00:07:43,410
These resolutions reflect the
capabilities of the sensors,

149
00:07:43,410 --> 00:07:46,110
but they also reflect
the different intentions

150
00:07:46,110 --> 00:07:47,610
of each of these satellites.

151
00:07:47,610 --> 00:07:51,480
For example, many
researchers using MODIS data

152
00:07:51,480 --> 00:07:55,170
are doing regional and global analyses.

153
00:07:55,170 --> 00:07:56,220
So I'm bringing this up

154
00:07:56,220 --> 00:07:58,050
because I wanted to talk
about the trade-offs

155
00:07:58,050 --> 00:08:03,050
between resolution or grain and extent.

156
00:08:03,180 --> 00:08:05,220
So spatial grain is kind of another way

157
00:08:05,220 --> 00:08:08,610
of talking about spatial resolution.

158
00:08:08,610 --> 00:08:12,000
So it's the smallest spatial
resolution of the data.

159
00:08:12,000 --> 00:08:13,800
It's kind of like a pixel size.

160
00:08:13,800 --> 00:08:16,890
And then extent is the
overall size of the landscape

161
00:08:16,890 --> 00:08:20,670
or area that you are
observing or measuring.

162
00:08:20,670 --> 00:08:23,130
And again, there's a
trade off between grain

163
00:08:23,130 --> 00:08:27,240
and extent that scientists or researchers

164
00:08:27,240 --> 00:08:30,390
or sensor engineers have to consider.

165
00:08:30,390 --> 00:08:32,370
So for example, satellite sensors,

166
00:08:32,370 --> 00:08:35,400
the extent is often regional or global

167
00:08:35,400 --> 00:08:38,760
depending on the orbit
path of the satellite.

168
00:08:38,760 --> 00:08:42,330
But for drones, the extent
might be much smaller,

169
00:08:42,330 --> 00:08:44,940
such as like one crop field.

170
00:08:44,940 --> 00:08:46,470
Therefore, for satellite sensors,

171
00:08:46,470 --> 00:08:49,800
an extremely high resolution,
spatial resolution,

172
00:08:49,800 --> 00:08:51,420
may be 10 meters.

173
00:08:51,420 --> 00:08:54,030
And to get a higher spatial
resolution than that

174
00:08:54,030 --> 00:08:57,810
would be extremely costly and
computationally difficult.

175
00:08:57,810 --> 00:09:01,560
But on the other hand, drones
which have a smaller extent,

176
00:09:01,560 --> 00:09:05,460
they could have a grain
size of just one centimeter.

177
00:09:05,460 --> 00:09:08,490
So therefore, what is
considered high resolution

178
00:09:08,490 --> 00:09:11,010
is all relative 'cause
you have to consider

179
00:09:11,010 --> 00:09:13,893
both the extent and the grain.

180
00:09:15,330 --> 00:09:17,160
This is something that you
should also keep in mind

181
00:09:17,160 --> 00:09:20,250
when you're choosing
ROIs for your projects.

182
00:09:20,250 --> 00:09:23,100
ROIs, regions of interests.

183
00:09:23,100 --> 00:09:25,800
If you choose too large of an extent,

184
00:09:25,800 --> 00:09:29,580
you may need to use
lower resolution imagery.

185
00:09:29,580 --> 00:09:33,780
And conversely, if you
choose a super small ROI,

186
00:09:33,780 --> 00:09:35,880
you may wanna make sure that your imagery

187
00:09:35,880 --> 00:09:38,190
has a small enough grain size

188
00:09:38,190 --> 00:09:41,883
that you can actually make
out features on the landscape.

189
00:09:43,500 --> 00:09:46,320
Going back to raster images,

190
00:09:46,320 --> 00:09:47,880
it's important to note that rasters

191
00:09:47,880 --> 00:09:50,970
can be a single layer or multi-layered.

192
00:09:50,970 --> 00:09:52,590
These layers are called bands,

193
00:09:52,590 --> 00:09:54,000
and as we discussed last week,

194
00:09:54,000 --> 00:09:58,980
each band from a satellite
sensor depicts reflectance values

195
00:09:58,980 --> 00:10:02,550
from a specific range in the
electromagnetic spectrum.

196
00:10:02,550 --> 00:10:04,740
So for example, from
the near infrared range

197
00:10:04,740 --> 00:10:07,530
or from the blue range.

198
00:10:07,530 --> 00:10:09,990
And you can think of multi-band images

199
00:10:09,990 --> 00:10:14,990
as three dimensional cubes or
like stacks of multiple grids.

200
00:10:15,210 --> 00:10:19,140
Individual bands are usually
displayed in gray scale.

201
00:10:19,140 --> 00:10:21,300
They could also be shown in pseudo color.

202
00:10:21,300 --> 00:10:24,750
But in order to get an image
that has different colors,

203
00:10:24,750 --> 00:10:26,700
we combine three bands

204
00:10:26,700 --> 00:10:29,520
and we visualize them with three colors,

205
00:10:29,520 --> 00:10:31,083
red, green, and blue.

206
00:10:31,980 --> 00:10:35,460
So red, green and blue are
the three primary colors

207
00:10:35,460 --> 00:10:38,730
and they can be combined
to make any color.

208
00:10:38,730 --> 00:10:41,520
As an example, when you're watching TV,

209
00:10:41,520 --> 00:10:43,890
you see 100 different colors, right?

210
00:10:43,890 --> 00:10:47,970
But actually the TV can only
display three different colors,

211
00:10:47,970 --> 00:10:50,280
red, green, and blue.

212
00:10:50,280 --> 00:10:54,180
But it displays them together
in different combinations

213
00:10:54,180 --> 00:10:56,250
and with different intensities.

214
00:10:56,250 --> 00:11:00,030
And that creates all of
the colors that we see.

215
00:11:00,030 --> 00:11:03,720
Color compositing in remote
sensing works the same way.

216
00:11:03,720 --> 00:11:05,760
We take three bands and we assign them

217
00:11:05,760 --> 00:11:08,430
to the colors red, green, and blue.

218
00:11:08,430 --> 00:11:12,090
And the reflectance values in each pixel

219
00:11:12,090 --> 00:11:14,250
in those three bands dictates

220
00:11:14,250 --> 00:11:17,643
how much of each of
those colors to display.

221
00:11:20,580 --> 00:11:24,990
A true color composite, also
called a natural composite,

222
00:11:24,990 --> 00:11:29,250
is an image that has the
red, green and blue bands

223
00:11:29,250 --> 00:11:32,520
assigned to the red,
green, and blue channels

224
00:11:32,520 --> 00:11:34,140
on the computer.

225
00:11:34,140 --> 00:11:36,150
Therefore, this composite looks like

226
00:11:36,150 --> 00:11:40,110
what we would observe
naturally with our human eyes

227
00:11:40,110 --> 00:11:42,090
where vegetation looks green

228
00:11:42,090 --> 00:11:45,780
and water looks dark,
maybe a little bit blue,

229
00:11:45,780 --> 00:11:49,263
bare ground might look brown.

230
00:11:50,880 --> 00:11:52,533
And then buildings look white.

231
00:11:53,460 --> 00:11:56,790
Many people prefer true color composites

232
00:11:56,790 --> 00:11:59,490
because they look natural.

233
00:11:59,490 --> 00:12:01,080
But it's important to
note that there's a lot

234
00:12:01,080 --> 00:12:04,680
of subtle differences in
the landscape that are hard

235
00:12:04,680 --> 00:12:08,880
to tell apart with true color composites.

236
00:12:08,880 --> 00:12:10,440
I also want you to be aware

237
00:12:10,440 --> 00:12:14,610
that a band numbers
change between satellites.

238
00:12:14,610 --> 00:12:19,610
So for example, some satellites,
red, green, and blue bands

239
00:12:19,710 --> 00:12:23,100
are labeled B3, B2 and B1,

240
00:12:23,100 --> 00:12:27,540
whereas in other satellites
they're labeled B4, B3, B2.

241
00:12:27,540 --> 00:12:29,340
This is just something
that you have to look up

242
00:12:29,340 --> 00:12:31,473
before you visualize your data.

243
00:12:33,210 --> 00:12:37,680
False color composites are
when we create color composites

244
00:12:37,680 --> 00:12:40,560
not assigning the red,
green and blue bands

245
00:12:40,560 --> 00:12:43,980
to the red, green, and blue
channels on a computer.

246
00:12:43,980 --> 00:12:46,770
I've shown here a table with
a bunch of different examples

247
00:12:46,770 --> 00:12:48,720
of different color composites

248
00:12:48,720 --> 00:12:52,260
that we can make using
the bands from Landsat 8.

249
00:12:52,260 --> 00:12:55,050
The first row shows the
natural color composite

250
00:12:55,050 --> 00:12:58,590
that we just talked about,
but underneath you can see

251
00:12:58,590 --> 00:13:01,050
many others with some of them labeled with

252
00:13:01,050 --> 00:13:04,710
which kinds of features
they may help distinguish.

253
00:13:04,710 --> 00:13:08,550
And I've shown visualizations
of two of these.

254
00:13:08,550 --> 00:13:11,220
The color infrared, false color composite,

255
00:13:11,220 --> 00:13:14,523
and then the short wave
infrared color composite.

256
00:13:15,900 --> 00:13:19,500
False color composites allow
us to visualize wavelengths

257
00:13:19,500 --> 00:13:22,140
that the human eye can't see

258
00:13:22,140 --> 00:13:24,480
like near infrared or short wave.

259
00:13:24,480 --> 00:13:27,870
And using these bands can help us

260
00:13:27,870 --> 00:13:31,560
increase the spectral separation
of objects on a landscape,

261
00:13:31,560 --> 00:13:34,563
and that helps us better
interpret the data.

262
00:13:37,830 --> 00:13:41,010
Finally, I wanted to give
a super concrete example

263
00:13:41,010 --> 00:13:44,040
about how these colors
or bands can be combined

264
00:13:44,040 --> 00:13:47,010
to make other colors.

265
00:13:47,010 --> 00:13:50,820
So in this example, I've shown bands

266
00:13:50,820 --> 00:13:52,200
and reflectance values.

267
00:13:52,200 --> 00:13:54,330
This would be for one pixel.

268
00:13:54,330 --> 00:13:58,140
So for example, in this
one pixel with four bands,

269
00:13:58,140 --> 00:14:01,110
the blue band pixel values 190,

270
00:14:01,110 --> 00:14:03,840
the green band pixel values 50,

271
00:14:03,840 --> 00:14:06,540
the red band pixel values 45,

272
00:14:06,540 --> 00:14:11,070
and the near infrared
band pixel value is 150.

273
00:14:11,070 --> 00:14:14,370
So if we assign the near infrared band

274
00:14:14,370 --> 00:14:18,900
to the color red on a computer,
the red band to green,

275
00:14:18,900 --> 00:14:21,213
and the green band to blue,

276
00:14:22,410 --> 00:14:25,800
what color do you think
it would show up as?

277
00:14:25,800 --> 00:14:27,840
So we can look at the reflectance values,

278
00:14:27,840 --> 00:14:31,590
which are also called DNs, which
stands for digital numbers.

279
00:14:31,590 --> 00:14:36,120
That's just sort of the raw
units of reflectance values.

280
00:14:36,120 --> 00:14:40,290
But you can see that near
infrared has the highest number.

281
00:14:40,290 --> 00:14:42,240
And so we can take that to mean

282
00:14:42,240 --> 00:14:46,290
that that pixel will
mostly show up as red.

283
00:14:46,290 --> 00:14:48,840
And just to demonstrate that even further,

284
00:14:48,840 --> 00:14:52,650
I used a website where you
can put in numbers for red,

285
00:14:52,650 --> 00:14:54,750
green, and blue and it'll show you exactly

286
00:14:54,750 --> 00:14:56,400
what color shows up.

287
00:14:56,400 --> 00:14:59,250
So I've done that with the values,

288
00:14:59,250 --> 00:15:01,770
and you can see indeed that the color

289
00:15:01,770 --> 00:15:03,573
that is created is red.