16: Remote Sensing - Satellites

Question 1

SATELLITE advantages and disadvantages

Answer 1

Advantages:

• high resolution
• sensors collect multiple EM bands
• wide image coverage (swaths)
• temporal resolution
• rapid data acquisition
• once in orbit they can work for a long time

Disadvantages:

• cost
• development time
• satellite launch failure (rocket explosion)

Question 2

ORBITS TYPES (3)

Answer 2

Geostationary
Intermediate circular orbit
Polar/Near Polar

Question 3

KEY PLATFORMS

Answer 3

AVHRR - Advanced very high resolution radiometer
2399km swath, 1.1km resolution, 2x NOAA satellites, 20 year climate record

MODIS - Moderate resolution imaging spectrometer
2330km swath, 250, 500 and 1000m resolution, complete coverage 1-2 days

ASTER - Advanced spaceborne thermal emission and reflection radiometer
60km swath, 15-90m resolution, coverage 16 days, no blue band, so can’t create true colour image

Landsat 8 (LDCM)
185km swath
15, 30 and 100m res
16 days

Worldview-2
Panchromatic with 0.46m res
Multispectral with 8 bands and 1.2m res
26 hour coverage with 18km swath

Worldview-3
Panchromatic at 0.3m res
Multispectral at 1.2m res with 8 bands
SWIR at 3.7m res with 4 bands
CAVIS 30m res with 12 bands

Question 4

CLASSIFICATION METHODS

Answer 4

Supervised classification
led by analyst

Unsupervised classification
led by computer and then categories are defined by analyst at the end

Question 5

History of remote sensing

Answer 5

1858 - first aerial photos from hot air balloons, taken from several 100s of meters high

1903 - pigeon camera with timing mechanism

WW1 and WW2 photographs from air planes to record enemy positions

1962 airborne intelligence informed the Cuban Missile Crisis, U2 aircrafts very high up

1960s Spy satellites, Corona system (US), first satellite images for military purpose, film photographs, resolution 1.5x1.5m

1964 meteorological satellites, Nasa’s Nimbus-1, multispectral (water vapour, liquid water, cloud temperatures)

1979 satellites detected the ozone hole

1970s Landsat program, records repeated images of land around the globe

Question 6

Swath

Answer 6

Satellite’s field of view

• width of data collected
• ‘mowing the lawn’
• wide swath typically has lower resolution, can monitor earth in a short amount of time
• narrow swath possible to have very wide resolution, generally monitor a small area

Question 7

INSTRUMENTS

Answer 7

multispectral scanner (MSS): most commonly used scanning system for collecting data over different wavelength ranges

information from a narrow wavelength range is gathered and stored in a channel / band

Question 8

Multispectral scanner instrument types

Answer 8

Panchromatic: single channel/wide band

Multispectral: multi channel (<10) narrow bands

Superspectral: multi channel (>10) even narrower bands

Hyperspectral

• imaging spectrometer
• HySpex (visible and near infrared)
• 100+ bands

Question 9

DATA ACQUISITION

Answer 9

humans: using visual interpretation to classify features in an image

computers: digital classification? we have to be aware of how the computer sees the world… numbers
- spectral pattern recognition

overall aim: assign all image pixels to particular classes/themes

result: classified image comprised of pixel mosaic, each belonging to a particular class, essentially a thematic map of the original image which we can then use for analysis

Question 10

How do satellites acquire data?

Answer 10

Scanning - 2 main scanning modes/methods to acquire multispectral image data

Across track scanners (ATS)

• scans at tangent to flight path
• 1 pixel at a time
• “whisk broom” (has a mirror which is rotating)
• during forward movement the satellite scans across 1 line, needs some correction for the forward movement of the satellite

Along track scanners (AlongTS)

• scans parallel to flight path
• multiple pixels at a time in strips
• “push broom”

Question 11

Geostationary orbit

Answer 11

• high earth orbit (~36000km = exact rotation time of the earth over 24 hours)
• generally fixed on 1 region (follow a set point on planet)
• weather/communication satellites

Question 12

Intermediate circular orbit (ICO)

Answer 12

• medium earth orbit (~20000km)
• orbital periods of 2-24 hours (travel across the sky)
• GPS/GLONASS
• communication satellites

Question 13

Polar/Near Polar orbit

Answer 13

• low earth orbit (LEO) (400-2000km)
• passes above or nearly above both poles
• passes over many regions in fixed period
• landsat / cryosat
• sun-synchronous

Question 14

AVHRR

Answer 14

Advanced very high resolution radiometer

one of the first multispectral images that could map the earth’s surface completely, ground resolution is not very high (1.1km) with 2399km swath, can scan the earth’s surface in a short time

Question 15

Supervised classification

Answer 15

• analyst first decides which categories / classes to look for
• identifies in imagery homogenous representative samples of surface cover types of interest (training areas)
• computer use this information to process and classify the image into the areas of interest

Process:

1. form images of data
2. choose training pixels for each category
3. calculate statistical descriptors

Question 16

Unsupervised classification

Answer 16

• spectral classes grouped first (based on numeric info in data)
• clustering algorithms used to determine natural (statistical) groupings/structures in data)
• the analyst matches the groups to info classes

Process

1. separate data in groups with clustering
2. classify data into groups
3. assign name to each group

Question 17

Land surface classification workflow

Answer 17

1. Obtain multi-spectral digital imagery
2. Identify categories using field knowledge
3. Extract signature of pixels in training samples
4. Define surface classes in multi-spectral space
5. Assign each pixel to one of these classes