TerraLook Images | Collection Formats | ASTER Sensor | LANDSAT Sensor | IKONOS Sensor
All the images included in TerraLook collections are geo-referenced jpegs—in other words, compressed images that include location information (latitude/longitude). The jpg format was chosen to decrease the size of the images so a sufficient number of them could be placed on a single CD or DVD. While this decreases data quality somewhat; in most cases the loss in quality is minimal (a quality factor of 85/100 is used, 100 being best).
All digital images, whether from personal digital cameras or from those in space, are composed of pixels (picture elements). The images in a TerraLook collection are "composite" color images derived from the green, red, and near-infrared bands. Because neither ASTER or MSS has a blue band (this was omitted because blue light tends to be scattered a lot by the atmosphere and so is rather "noisy"), it is necessary to create “simulated true color” images, which means that, in essence, an artificial blue band is synthesized from the other bands and then a more or less realistic image is generated. For consistency across all images the same approach was used for Landsat TM images and also for IKONOS—even though these sensors do, in fact, have blue bands.
The end result is that live vegetation is green, though dead vegetation (such as grasslands during the dry season) may be tinted purple. The way to think of this is that dead vegetation is naturally brown, which as a lot of red color to it. However, in these images the synthesizing algorithm tends to put a bit too much blue into pixels of dead vegetation, and the blue combined with the red can lead to the purple tinge.
Collections utilize a simple format so they can easily be used by a variety of tools. GIS-savvy users can use their own tools if they prefer. The directory structure is as follows:
This directory contains shapefiles that describe the footprints of the scenes in the collection. The first set includes images from all sensors in the collection, and the sensors can be distinguished via the “satellite” attribute keyword. We apologize for the perhaps confusing nature of the names, which refer to “aster” although information on all sensors is included. This reflects the original design which was to include only ASTER data in collections, an idiosyncrasy that will be fixed in a later release.
aster.shp: a set of boxes that outline the image boundaries
asterline.shp: a set of "X"s, one for each box, to help discriminate overlapping footprints
asterpoint.shp: a set of points that mark the centerpoint of each footprint
To make these data easier to use with some other GIS tools such as ESRI ArcReader or ArcExplorer, the footprint information is also provided on a “per-sensor” basis. This makes it somewhat easier for users to control the display of the individual image layers. The filenames are:
This directory contains jpg images plus their corresponding World Files to provide geocoding. Degree of compression varies but is generally small. While there is a small amount image degradation this decreases the size of the images by more than a factor of 10.
This directory, which is not populated in all collections, contains topographic data useful for 3-D visualization. The format is that used by the Global Land Cover Facility (GLCF), and each file corresponds to a particular Landsat WRS-2 path/row.
Much of the data in TerraLook are from an instrument called ASTER. Most people, however, are not familiar with this instrument, and this section provides some basic information. To summarize very briefly: ASTER is quite similar to Landsat, but provides much more detailed images. Landsat, however, has a wider swath width, and a very large historical archive.
ASTER is a large, space-based, digital camera that started operating in early 2000. It acquires about 600 high-resolution images a day, each one covering an area of 60 x 60 km, with a pixel size of 15 m for bands 1-3. Unlike Landsat ASTER does not use a predetermined path/row system. Rather, each image is individually aligned to a particular target. Thus, covering large areas with ASTER is not as convenient as with Landsat, and the ASTER images in a collection are generally not as neatly aligned as are the Landsat images.
ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer) is basically a large digital camera bolted to a satellite. The satellite, called Terra, is the size of a small bus, was launched in 1999, and has four other instruments. It circles Earth at a distance of 705 km, from pole to pole, about every 100 minutes, crossing the equator at about 10:30 am local time.
ASTER itself takes about 600 pictures ("scenes") a day, each covering an area of 60 x 60 km. Like most satellite sensors, ASTER is much more complex than a hand-held digital camera. First, and most importantly, a separate image is created for each color (or more precisely, each wavelength range, or "band"). Because ASTER has a total of 14 bands, it actually acquires 14 different images for each scene. This is useful because different materials can look very different in different bands--by acquiring images in each of the 14 bands a lot can be learned about the materials being imaged. When an image is "processed", each band can be treated separately, leading to some very powerful (and sometimes very complicated) analysis techniques. To keep things simple, and to save space, the images on this disk are "composite" images derived from bands 1, 2, and 3. A website is planned that will provide the full multi-band images along with some simple tools to help analyze them.
Another difference between ASTER and a typical digital camera is that ASTER has three lenses--(called telescopes because of their size and power)--rather than one. In fact, ASTER is really three separate instruments, each one specializing in a different part of the spectrum. This is because photons in one part of the spectrum behave very differently than in another, so different technologies are used for each part.
Each ASTER image has about 16 million pixels (4200 x 4200), and each pixel corresponds to about a 15 x 15 m patch on the ground, or about four times the resolution of a typical Landsat image, which has a pixel size of about 30m.
The ASTER instrument and its operation is a joint project between the US and Japan. Japan designed and built the instrument, the Level 1 processing system, and the operations system, and performs the day-to-day mission planning and the Level 1 data processing. The US designed, built, and operates the Terra spacecraft and the associated ground system.
All ASTER scenes (currently numbering roughly one million) are archived at a data center in South Dakota, USA (as well as at the equivalent data center in Tokyo). Access to the data in the US archive is by one or both of the following tools (the first provides both search and order capability, the second only search--but a much friendlier search-- and an easy path to ordering):
** EOS Data Gateway (EDG): http://edcimswww.cr.usgs.gov/pub/imswelcome/
** Global Visualization Viewer: http://glovis.usgs.gov/
For more information on ASTER, the various ASTER data products, how to submit data acquisition requests, the work of the ASTER Science Team, and much more, please visit the US ASTER website:
Launched: December 1999
Expected lifetime: until at least 2009
Number of bands: 14
Number of telescopes: 3 (VNIR, SWIR, TIR)
15 m (VNIR)
30 m (SWIR)
90 m (TIR)
Repeat frequency: 1-16 days
Much of the data in TerraLook are from Landsat, which is a series of spacecraft with a historical archive that goes back to 1973. The earlier Landsat spacecraft contained an instrument called the Multispectral Scanner (MSS), which had a pixel size of about 80m. Later Landsat spacecraft had an instrument called Thematic Mapper, with a pixel size of about 30m. All Landsat images, regardless of which sensor acquired them, cover an area of 180km x 180 km. Also, Landsat acquisitions follow a predetermined path/row system, which divides the Earth’s surface into a series of paths (columns of images from North to South) and rows (rows of images from East to West). This makes it very convenient for comparing images acquired at different times...though with one complication. The path/row system used for the earlier Landsats, called World Reference System 1 (WRS1) is somewhat different from that used for the later Landsats, which uses the WRS2 system. This change was necessitated by a change in the orbit for the later spacecraft.
Landsat data for TerraLook Collections are taken from three epochs. These data are part of the "Geocover" global orthorectified dataset, fully explained in Tucker, CJ, DM Grant, and JD Dykstra (2004). NASA’s Global Orthorectified Landsat Data Set. Photogramm Eng Rem Sen Vol. 70, No. 3, March 2004, pp. 313–322 which is available online at http://glcf.umiacs.umd.edu/pdf/PERSMarch_04_313-322.pdf.
The earliest epoch covers a range of dates from the 1970’s and utilizes MSS data. In TerraLook this epoch is referred to as the “1975 layer”. The later epochs are 1990 and 2000, and these layers contain images that are generally within a year or two of these respective dates. In all cases the images were selected from a very large archive, with the primary selection criterion being cloud cover. Thus, depending on what the images are being used for, some care may be required when making comparisons between epochs, as the images may have been taken at different seasons.
Additional information on the Landsat instruments and data can be found at websites such as
IKONOS is a very high resolution commercial sensor, not operated by NASA or USGS. Most commercial data come with copyright restrictions, however, occasionally data are bought that can be redistributed with TerraLook collections. For example, after the Indian Ocean tsunami disaster in 2004 a variety of IKONOS images were made available by USGS to support emergency operations. It seems likely that, in the future, IKONOS data will generally be available only for TerraLook Collections in special circumstances such as for disaster relief operations.
IKONOS data have a pixel size of 1 or 4 m, depending on how it is processed, and so have a resolution that is at least 16 times greater than that of ASTER. The swath width is much smaller, though often multiple images are combined into a single larger image, making this somewhat less significant.