Data Models, Pixels, and Satellite Bands

Introduction to Digital Data and Imagery

Understand the differences between raster and vector data.

What are digital numbers (DNs) and what do they represent in remotely sensed data?

What are satellite bands?Be able to define spatial and spectral

resolution.What are some advantages and disadvantages

of high spatial resolution?What are some advantages and disadvantages

of high spectral resolution?

Learning Objectives

Data Models: Raster vs. Vector DataRaster Data

Images are grids of cells or “pixels”

Each pixel is represented by number called the digital number (DN)

A pixel represents an area on the ground

Vector DataPoints, lines, and

areasDefined by x,y

coordinates and info about “connectedness”

Attributes can describe the features they represent

100

10 250

0 122

35 75 255

5 90 92

112

12 11 2 3 200

78 200

20 75 93

187

202

35 87 90 90 140

144

52 68 89

2 3 23 76 56 62 178

142

24 53 1

0 0 10 35 57 63 242

137

36 25 0

Raster Grid w/ DNs

Point

Line or Arc

Area or Polygon

Vector Features

Data ModelImaginary matrix (row & column format) is

placed on the feature (e.g., the ground)Some phenomenon (e.g. amount of light) is

measuredA value (digital number) representing the

amount of light is assigned to each grid cell (pixel).

Somewhere on earth

Raster grid is placed

Raster data

32 47 67 93 11

105 79 35

23

11

56

43

89

21

213

245

201

179

136 155

55203

163 63

211 189

145

109122202

Real world >>> Raster value

For remote sensing, the total amount of EMR from the area of a pixel on the ground is recorded as a digital number (DN).

Depending on the intensity of the EMR, a numeric value is assigned to each pixelLow or None - Lowest value (dark)High - Maximum value (bright)Others - Scaled in between (gray)

Digital Image DataDigital images are matrices of digital

numbers (DNs).Satellite bands capture light from different

wavelength regions.There is one layer (or matrix) for each

satellite band.Each band covers the same area on the

groundEach DN corresponds to one pixel in one

bandIf there are 6 bands each pixel will be

described by 6 DNs, one for each bandThe DNs control how a computer displays an

image on your computer screen

Images are presented as 2-d grids. Each pixel (one square) has a location (x,y).

Position of pixel (x,y) often describe in terms of rows and columns but can be translated (projected) into other coordinate systems (e.g., latitude/longitude, UTM, etc.)

0 0 0 2

0 2 3 4

2 2 6 9

2 2 9 12

F(2,3)

F(4,1)

What are digital numbers (DNs)?DNs are relative measures of radianceDNs are NOT reflectanceDNs can be converted to ground reflectance

if you know atmospheric properties, etc.The range of DNs depends on the radiometric

resolution of the recording instrumentA common range of DNs is 0 – 255.

Resolution

Types of ResolutionSpatialSpectralRadiometricTemporal

Spatial ResolutionThe dimension of one side of a single pixelThe extent of the smallest object on the

ground that can be distinguished in the imagery

Determined by the Instantaneous Field of View of satellite instruments (IFOV)

Determined by altitude, camera (lens), and film characteristics for air photos.

Spatial Resolution

Raster grid sizefiner

Coarser

Available Resolution

Satellites: ~ .61 m to > 1 kmAir photos ~ <0.6 m to large.

Satellite data resolutionMODIS: 250 - 1000 mLandsat MSS: 80 mLandsat TM5, 7, 8: 30 mSPOT: 20 m ASTER: 15 mIRS Pan: 5 mWorldview 3 Pan: 31 cm (!!)

Quickbird (Digital Globe, Inc.)

~ 2.4 m spatial resolution in multispectral bands.

MODIS

500 m spatial resolution

So…what about in the movies?

Zoom and Enhance! (YouTube)

The Truth!

Group discussion: Can you extract (zoom and enhance!) information from an image when the information is in an area smaller than a pixel?

Spectral ResolutionHow finely an instrument “divides up” the

range of wavelengths in the electromagnetic spectrum

How many spectral “bands” an instrument records Landsat 8

bands (in red) overlaid on plant spectral reflectance.

https://landsat.usgs.gov/tools_spectralViewer.php

Case 1Measure the EMR across a wide rangeE.g., the visible portion of EMRAssign a single DN for sum of all visible light

energy hitting the sensorAnalogous to black and white (panchromatic)

filmCalled a panchromatic band

blue

green

red

0.4 0.70.60.5UV Near-infrared

Case 1

Case 2

Measure EMR across narrower ranges

E.g., Blue, green and red bands

Assign a DN for each of these wavelength ranges to create 3 bands

Case 2

blue

green

red

0.4 0.70.60.5UV Near-infrared

Coarser (lower) Spectral Resolution

Finer (higher) Spectral Resolution

RGB

Red Green Blue

400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000 2100 2200 2300 2400 25000.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000 2100 2200 2300 2400 2500

0.0

0.2

0.4

0.6

0.8

High Spectral Resolution

Low Spectral Resolution

Wavelength (nm)

Wavelength (nm)

Refl

ect

an

ceR

efl

ect

an

ce

Spectral Resolution

Remotely sensed imagery are RASTER data composed of grids of pixels organized in bands (layers)

Size of pixels is called spatial resolution of sensor

Number of bands is called spectral resolution of sensor

Digital numbers are associated with pixels and tell you relatively how much light came from that area on the ground to the satellite sensor.

Summary