A Brief Explanation of Landsat Technology
Just How Do They Do That?
The images depicted in our posters and maps are created using digital data derived from the Landsat 5 satellite, or the recently launched Landsat 7 satellite. Landsat satellites orbit 705 kilometres (about 435 miles) above the Earth, and image an area every other 16 days. The Landsat 5 satellite has an onboard sensor called the Thematic Mapper (TM). The TM sensor records the surface reflectance of electromagnetic (EM) radiation from the sun in seven discreet bands. EM radiation refers loosely to light waves and other energy such as x-rays or microwaves (see below). Essentially, the satellite 'sees' reflected sunlight in portions of the spectrum including visible light and three bands beyond visible light (within the infrared portion of the spectrum). There's even a band in the infrared for the heat emitted from the planet primarily used in specialized applications.
The figure below illustrates where in the EM spectrum the TM sensor can 'see'. The rectangles depict the bandwidth recorded within that region of the spectrum. The curves illustrated on the image are examples of different land cover types such as vegetation, rock, or open water. Examining the curves we see how a cover type may have a similar response in one region of the spectrum and quite different within another. This helps explain the reason for the variety of colours within an image.
Many TM images utilize a combination of bands 1, 2, and 3 represented as blue, green, and red to visualize the data. This combination is similar to what humans see. To create some of our maps we utilize red light (band 3), and two of the infrared bands, (bands 4 and 5) depicted as blue, green and red. Recall, unlike humans, the TM sensor can measure non-visible, EM energy, that is, infrared light. We then convert the numerical values provided by the sensor into colours that are visible to us. These images are called false colour because we are converting non-visible, spectral information into a visible format we can use. Also, the band 3 data (representing red light reflectance values) is depicted as blue in the image.
Check out the images of the same area, but created by combining different TM bands as blue, green, and red (bgr). The first figure shows bands 1, 2, and 3 as bgr, while the second figure (345) is an example of a false colour composite. The 345 combination looks a little sharper than the 123 combination because water vapour, and other gases in our atmosphere can often blur this satellite's view of the earth. Different land-cover types in the 345 combination also have much greater contrast due to greater differences in their reflective properties within the infrared portion of the spectrum. The colour representation of any object in an image is partly a result of how bright it reflects EM energy in a chosen band, and what colours we use to represent that band.
Launched in the spring of 1999, the Landsat 7 satellite contains an improved Thematic Mapper sensor called the Enhanced Thematic Mapper (ETM+). The sensor records data in the same seven bands as the TM sensor aboard Landsat 5, but has an additional sensor that records data in a wide bandwidth (encompassing bands 2,3, and 4) in a panchromatic mode (black and white). This panchromatic sensor also has twice the resolution of the multispectral bands (15 metres as opposed to 30 metres). Below we see an example comparing the multispectral band 5 data to the panchromatic data. Depicted is a small airport north of Winnipeg. The image to the left is the band 5 data and to the right the panchromatic data. The band 5 data appears more blurry and edges are not as well defined (note the runways and where the highway passes beneath a bridge near the airport. The streets in the adjacent town appear more notable in the band 5 image (due to greater spectral differences between the street and adjacent lawns and trees within the narrow band 5 bandwidth). The spectral differences between the streets and adjacent lawns in the pan image is less partially as a function of the wider panchromatic bandwidth.
The benefit of this new panchromatic band is more fully illustrated in the two images below. On the left we can see a false colour rendition of the same area utilizing bands 5,4, and 3 as rgb. Recall, this data has a 30 metre pixel spacing (each individual pixel covers 30 x 30 metres). To the right, the same data has been combined with the panchromatic data allowing us to 'add' greater spatial resolution to the image while attempting to retain the same look. What we end up with is a 'sharper' image that's still in colour. Check out the streets in the town now. We can see greater differences between street and lawn while also better defining the actual streets.....cool!
Aside from enhancing the images to make them more interpretable we reproject the recorded information into a known map projection. The Earth is an irregular shaped object, slightly flattened at the poles and bulging out at the equator requiring us to "georectify" the scene. This involves projecting portions of the off-round Earth onto 2d map projection. A common form of projection is the Universal Transverse Mercator or UTM. In Canada, the majority of our national topographic maps are in a UTM projection. Satellite Impressions' image maps of Canadian locations are reprojected to a UTM projection.
This little introduction into the technology of our data source is intended to wet your appetite. If you'd like to learn more about remote sensing and the Landsat satellite here are two great places to start. Check out: