GE3502GE5502 Geographic and Land Information Systems

1
GE3502/GE5502Geographic and LandInformation
Systems
Lecture 25 Remote sensing in GIS for thematic
mapping
2
Thematic mapping

• Shows the distribution and extent of features or
  themes e.g. climatic zones, household income,
  population density, soil, vegetation, geology and
  land use.

3
GIS Data where does it all come from?

• So far during this course we have covered
• Digitising and scanning (lecture 11)
• Information already classified and transformed
• Limited by what data has been collected
• Often limited information on how data was
  collected
• Field work (lecture 21)
• Labour intensive
• Time consuming
• Usually cover a relatively small area

4
(No Transcript)
5
Remote Sensing

• Strictly speaking, remote sensing refers to the
  use of satellites that obtain data about large
  regions of the Earths surface
• A working definition is information that has
  been gathered about the Earth indirectly, or
  remotely
• We will use the term to include the use of
  satellites, aerial photography and other airborne
  information gathering devices such as
  side-looking airborne radar (SLAR), scanning
  radiometers, colour video and digital photography

6
Classification of geographic data

• Remotely sensed satellite images and digital
  photographs are referred to as primary data
  sources
• These are collected specifically for use in GIS
• Secondary data sources are those that were
  originally captured for another purpose and need
  to be converted into a form suitable for use in a
  GIS project
• e.g. scanned maps or photographs
• Topographic maps

7
Principles of Electromagnetic Radiation

• Visible and IR is reflected, shortwave radiation
• Thermal IR is emitted, longwave radiation

Atmospheric Transmission Curve
8
Seeing Heat
Coincident images of the Brazilian rain forest at
.657 µm (red) and 4.05 µm (thermal infra-red),
respectively. The visible light image shows
only smoke, while the thermal infra-red image
shows details of the fire beneath it.
9
Remote sensing and resolution

• Resolution is the key physical characteristic of
  remote sensing systems
• Four aspects to resolution
• Spatial
• Spectral
• Temporal
• Radiometric
• The sensor you decide to use for acquiring
  remotely sensed data will depend, in part, upon a
  combination of the above factors

10
Spatial resolution

• The minimum size of an object that can be
  resolved
• Typically the pixel size is used to define
  resolution

11
NOAA AVHRR 1km pixels
12
SPOT 10m pixels
13
Quickbird 0.6m pixels
14
Spectral resolution

• Refers to the parts of the electromagnetic
  spectrum that are measured
• Different objects emit and reflect different
  types and amounts of radiation

15
Spectral resolution of Landsat TM imagery

• Band 1 is for bathymetry Bands 2 and 3 are for
  vegetation discrimination bands 2 and 4 are for
  measuring plant vigour bands 5 and 7 are for
  measuring water stress in plants, and for
  discrimination of rock types. All are 30 m
  spatial resolution.
•  
• Band 6 is for thermal mapping (120 m). There is
  also a Band 8 on LANDSAT 7 which takes black and
  white (panchromatic) imagery with a pixel size of
  15 m.

Thematic Mapper Bands (?m) 1 0.45 -
0.52 (blue) 2 0.52 - 0.60 (green) 3 0.63 - 0.69
(red) 4 0.76 - 0.90 (NIR) 5 1.55 - 1.75 (MIR) 6
10.4 - 12.5 (TIR) 7 2.08 - 2.35 (MIR)
16
Temporal resolution

• Temporal resolution describes the frequency with
  which images are collected for the same area
• There are two basic types of commercial remote
  sensing satellites
• Earth orbiting
• Geostationary

17
Radiometric resolution

• This refers to the number of energy intensities
  that the sensor can resolve (i.e. smallest
  differences that can be discriminated).
• 2 bits 4 grey levels or colours.
• 8 bits 256 grey levels or colours (LANDSAT TM
  and MSS, SPOT).
• 10 bits 1024 (NOAA)
• This property has important implications in the
  display of remotely sensed data. Each value in 8
  bit data range can be displayed on a 256 colour
  or greyscale

18
Radiometric resolution
Level 2
Level 8
Level 64
19
Satellites

• Most commercial RS satellites carry two sensors
• Single band panchromatic sensor measuring
  radiation in the visible part of the spectrum
• Multi-spectral sensor (MSS) measuring green, red,
  blue invisible radiation
• Panchromatic imagery usually has a finer
  resolution

Quickbird satellite
20
Why passive sensors work
21
Recording EMR

• Low radiant energy levels (Absorbed) are given
  low BVs and high radiant energy levels are given
  high BVs (Reflected).
• The BV is usually recorded as a digital number
  (DN a positive integer) for display and
  analysis in a computer.

22
Scanned Data Format

• Brightness values are recorded in a grid, or
  raster, format. This is a matrix of cells or
  pixels organised into an array of rows (lines)
  and columns (samples).

23
Unprocessed TM image
24
Why use satellite imagery?

• Examine large areas of the Earth at once
• Measure things that are dangerous and difficult
  to examine from the ground
• Measure features that are continuous (change
  gradually across space)
• Examine how a place changes over time

25
SPOT image of Townsville
26
Things to consider

• Cost satellite imagery can be very expensive
• Can require lots of room to store and use on your
  computer
• Requires processing before you can use it in a
  GIS this can be very time consuming
• Spatial and temporal resolution will be limited
  to availability of satellites

27
Where to get satellite imagery from

• ACRES Australian Centre for Remote Sensing
• BOM - Australian Bureau of Meteorology
• NOAA National Oceanographic and Atmospheric
  Association
• NASA - National Aeronautics and Space
  Administration
• JCU Dept of Electrical and Computer Engineering
• TESAG Cartography (Adella Edwards)
• Note Always check a few different sources when
  chasing data it is often possible to enter data
  share agreements where agencies provide data at a
  greatly reduced cost.

28
Aerial Photography

• Aerial photography has long been a primary source
  of base map data for many common products
• Two major problems
• The ability to obtain required classifications
  from the photography requires the use of an
  interpretation key
• Rectification (scale, relief and tilt
  distortions) and lack of reference grid the
  photo must be georeferenced to a coordinate
  system, which requires the use of ground control
  points (GCPs)

29
Distortions on an aerial photograph
30
AP of Townsville at 125000
31
Why choose aerial photography

• All the reasons you would use satellite imagery,
  plus.
• Finer resolution than satellite imagery (usually)
• You can control when and where images are taken,
  i.e. you are not limited to the position of a
  satellite in orbit
• You can coordinate a ground survey to take place
  at the same time

32
Things to consider

• Can be expensive to get all the photos you need
• Can require lots of room to store and use on your
  computer
• Requires processing before you can use in GIS
  can be time consuming
• Includes distortions due to the camera lens, and
  the angle and height of the aircraft

33
Sources of aerial photography

• Geoscience Australia
• GBRMPA Great Barrier Reef Marine Park Authority
• AIMS Australian Institute of Marine Science
• TESAG Cartography (Adella Edwards)
• Private companies

34
Summary of RS platform properties
35
Points to remember

• Remote sensing data are not gathered directly
• Electromagnetic signals are received as
  surrogates of what is actually on the ground
• The raw data must be processed by experienced
  interpretation specialist before object
  categories can be properly identified

36
Problems stemming from RS data

• The quantisation of space into pixels imposes a
  level of abstraction and simplification on the
  Earths features.
• Objects that are much smaller than the pixels
  cannot be directly identified
• Satellite remotely sensed raw data provides
  little information until it has been analysed
• Two major types of data processing
• Image enhancement
• Image classification

37
Image enhancement

• The raw data is grouped into statistically or
  visually useful patterns
• Algorithms are used to improve the visual
  appearance of features, patterns, edges, or even
  clusters
• The image may be filtered to remove isolated
  values that appear too prominently

38
Image classification

• Used to group pixels into categories to which
  ids can be assigned
• Two approaches used
• Supervised
• Unsupervised
• Both approaches use a sort of region growing
  method, where individual pixels are selected and
  combined into categories

39
Image classification

• The scale of imagery determines information
  content that can be mapped from that imagery
• There are BENCHMARKS for assessing the relative
  information content of an image
•  
• J. ANDERSON'S 1976 LAND USE AND LANDCOVER
  CLASSIFICATION FOR USE WITH REMOTELY SENSED DATA
• http//landcover.usgs.gov/pdf/anderson.pdf

40
More problems with RS data

• Different RS data sources may prove to be
  incompatible
• Problems
• Categories may not relate well to each other
• Data may be collected at different scales
• The above are particularly a problem if RS data
  are being used to update historical maps
• Changing atmospheric conditions
• Cloud cover
• Haze
• Sensors age may result in sensor drift

41
Getting the data into a GIS

• Data needs to be in a digital format
• This is already the case with satellite imagery,
  however aerial photos will need to be scanned
• The data needs to be in the correct format for
  the particular GIS software you are using
• Most image processing software has tools to
  export and convert data between different formats
• If the image is rectified you will need to know
  the coordinate system, projection and datum
• This is essential for when you are comparing
  different data sets of the same area

42
Ocean Temperatures
43
2002
44
Great Barrier Reef
Hotter
100 km
Cooler
45
Coral Bleaching

• Increased water temperatures lead to mass
  mortality of corals
• Sea surface temperature mapping is being used to
  predict future bleaching events
• Determine if bleaching is linked with global
  warming

46
Summary

• Remote sensing is useful for gathering
  information over large areas
• Remotely sensed data needs to be used with
  caution and requires validation by field measures
• Remote sensing is still a rapidly maturing field
  of investigation with much scope for future
  research