Remote Sensing

Objectives of Lecture

• continue treatment of images (finish coverage of last lecture)
• show some slides of space images (connect to Labs)
• pick out some simple unity (time, space, attributes) in a sea of technical detail
• leave a set of pointers on the web for further exploration
• pointers to the politics of remote sensing applications
• Course Management Issues: questions about exams and learning

Basics of Electromagnetic Radiation

visible light (.4 to .7 micrometers) is just a tiny part of a big spectrum (from huge soft radio waves up to a kilometer down to hard gamma rays 10 to the -10 power meters)

• Towards the "hard" end, ultraviolet light turns into X-rays and gamma rays.
• In the soft direction, infrared covers a spectrum much broader than the visible window, followed by microwaves and radio frequencies. All of these are "light", but they require different kinds of detectors (antenae).
• Atmosphere is differentially transparent (this is why there is a Greenhouse Effect from carbon dioxide and infrared (heat) radiation...)
• Objects absorb and reflect light differently, this is the reason why vision is useful to detect differences in the environment. Using a broader range of spectra allows more distinctions.

Some overall themes in remote sensing

Time

• The Potential:
  • Rapid repeat coverage
  • Response to environmental threats, disasters
• Limiting Factors:
  • Delays in obtaining images, processing, interpretation, etc. (still experimental)
  • Clouds, haze, changed image characteristics, and clouds (did I mention clouds?)

Space

Sensors divide images in pixels (small rectangles that are the units for which the intensity of light is measured)

Early resolution (80 m or worse from weather satellites) was very crude.

Resolution sounds like the game. Higher resolution can detect more and more "features", but are you actually trying to distinguish every car on the freeway? Resolution makes interpretation and classification HARDER, not simpler.

Attributes

In only a few applications does the raw measurement of reflected light contribute directly (some infrared sensors are used to measure temperature at night for insulation and energy studies...)

Images have to be interpreted, classified, processed in some way. The idea was that spectral signatures would be easy, but the environment is quite complex. Classification accuracies have been disappointing. Change detection may be confused with interpretation error, and so on.

Still, some uses have emerged, and mapping technology takes a few decades to evolve, despite this era that expects doublings of processors speeds in three years (or less)...

Remote Sensing developments

What has become routine (as the civilian sector):

• Landsat (TM for thematic mapper) 30 m resolution, 6 usable bands; Landsat 5, launched in 1984 (and designed for a three year lifetime), is still operating (think about still using a computer you bought in 1984, let alone one you designed in the 1970s to launch in the 1980s...). Landsat 6 didn't make it to orbit in 1993.
• SPOT (French) 20 m with 3 bands (green, red, near IR), 10 m panchromatic. (SPOT 1 launched in 1986, followed in 1990 and 1993). More advanced pushbroom technology (less image geometry issues), aimable sensor, commercial strategy (images taken on demand)
• Canadian RadarSat and various shuttle radar missions (active imaging, instead of passive reflected light)
• Indian Remote Satellite Program has orbited various sensors vaying from the resolution of the first generation of Landsat to capabilities approximating SPOT (23.5 m for a four band scanner launched in 1995, 5.8 m panchromatic)

The civilian sector does NOT have the best imagery sytems that money can buy. Clearly the military and intelligence comunities had better birds.

Soviets stayed with film return systems longer, US invested in real-time direct images downlinked.

Some recent programs:

An altimeter for ocean measurement (TOPEX-Poseidon)

(technically not an image, just an elevation measurement, but this is remote sensing...)

Landsat 7

launched in April 1999, still 30 m resolution (15 m panchromatic, finally), ETM+ has 8 bands, thermal band has 60 m resolution (hard to focus); get the specs from NASA. This is essentially the sensor designed for Landsat 6 (which didn't make it to orbit in 1993).

Note: Muehrcke talks about the attempt to commercialize Landsat. Landsat 7 will be operated by US Geological Survey and firmly back in the scientific (non-commercial) sector...

USGS EROS Data Center has:

The commercial sector is "about to launch" sensors in the 1 to 5 meter range. It remains to be seen if these ventures will make any money.

SPIN-2 (SOVINFORMSPUTNIK, a film-return platform w 2 m resolution) is a privatized version of Soviet spy technology.

More to add, sorry...

Version of 18 January 2000