Remote Sensing

1
Remote Sensing
2
Remote Sensing

• is the measurement or acquisition of information
  of some property of an object or phenomena by a
  recording device that is not in physical or
  intimate contact with the object or phenomena
  under study

3
Remote Sensing

• includes aircraft, spacecraft and satellite based
  systems
• products can be analog (e.g., photos) or digital
  images
• remotely sensed images need to be interpreted to
  yield thematic information (roads, crop lands,
  etc.)

4
Principles

• sensors measure the amount of energy reflected
  from the earths surface

5
The remote sensing process
6
Principles

• energy sources and radiation principles
• different sensors measure different parts of the
  electromagnetic spectrum

7
Electromagnetic spectrum
8
Visible light

• optical systems use photographic film or
  electro-optical scanners
• also work in near-infraredaerial photography
  based on cameras on aircraft
• satellite-based systems

9
Aerial photography

• important for updating large scale topographic
  maps (e.g., new roads, urban areas)
• stereo-effect pairs of images that are displaced
  produce 3-D effect
• allows for measuring elevation

10
Satellite based systems

• Landsat , SPOT, etc.
• also Russian, Indian, Japanese, European, and
  Canadian
• panchromatic versus multispectral
• Landsat 7-8 spectral bands some in visible
  spectrum
• new and planned systems have many more
  (hyper-spectral images)

11
Landsat TM image ofHongkong
(bands 7,4,3 - 60m resolution) shows vegetation
in green, urban areas in purple/ white, water in
blue/black
Source Eosat
12
Panchromatic image of Atlanta
from the Russian KVR 1000 camera (2-5 m
resolution)
Source Eosat
13
Washington D.C.
Merged Image of Washington D.C. combining Landsat
TM and KVR 1000 data (resampled to 5m resolution)
Source Eosat
14
Satellite based systems

• data recorded for pixels picture elements
• size on-ground of a pixel varies from 1m to 60m
  or more for commercial systems
• images are sent back from satellite as very large
  raster data sets

15
The remote sensing process
16
Digital image processing

• digital satellite data usually need considerable
  processing
• registration and atmospheric correction
• analysis - measurement- classification-
  estimation

17
Measurement

• temperature
• vegetation biomass- Normalized Difference
  Vegetation Index (NDVI)
• elevation
• crop condition
• urbanized area

18
Classification

• identify and map areas with similar
  characteristics
• assign meaningful categories such as land-use or
  land-cover classes to pixel values
• need training areas (ground-truth)
• statistical approaches

19
Classification

• reflectance varies with time of day
• often large uncertainty in classification-
  pixels may contain several classes
• despite good image processing systems requires
  lots of experience(part art, part science)

20
Estimation

• objective is to estimate total amounts of a
  quantity, or areas under cultivation for an
  administrative or management area
• examples crop areas, forest resources

21
Other systems

• meteorological satellites
• e.g., Advanced Very High Resolution Radiometer
• coarser resolution but higher frequency and
  larger areas covered
• designed for meteorology but used for many other
  purposes (e.g., NDVI)

22
Other systems

• radar remote sensing (e.g., microwave)
• advantages in areas where cloud cover is frequent
  (e.g., tropical areas close to the equator)
• difficult to interpret

23
Other systems

• aerial video
• visible light
• using off-the-shelf video cameras and
  post-processing systems
• cheap, rapid data collection for monitoring and
  data capture

24
Survey costs
25
Socioeconomic applications

• delineation of newly urbanized areas (e.g.,
  Quito, Manila)
• mapping of villages for population estimation
  (e.g., Sudan, W-Africa) with Landsat (rooftop
  surveys)
• Defense Meteorological Satellite Programs (DMSP)
  nighttime visible light emissions

26
DMSP data
(water areas masked)
27
Population Distribution 1980
Source U.S. Bureau of the Census
28
Census from heaven?

• DMSP data good for delineating urban/non-urban
• too little variation to link to population
  densities
• higher resolution data usually too expensive
  and not accurate enough for census-type
  activities
• but useful source of base maps for EA delineation

29
Remote sensing and GIS

• remotely sensed data is an important data source
  (currency, frequency)
• large scale e.g., cities revealed
• medium scale framework data, urban/non-urban,
  crop conditions, etc.
• small scale NDVI, global land cover data sets

30
Remote sensing and GIS

• requires considerable processing to achieve high
  accuracy products
• image rectification and registration with GIS
  data sets difficult with raster systems
  (resampling)
• image interpretation guided by GIS data