Image data from the NOAA Polar Orbiter weather satellites were received at 26 ground stations in 19 countries worldwide from April 1992 until September 1994. This coordinated international effort involved six space agencies including NASA, NOAA and ESA, with over 30,000 individual digital images being collected and shipped to the US Geological Survey's EROS Data Center for production of global vegetation maps of the land surface. 20 globe maps, each representing a ten-day period, were then shipped on computer data cartridges from the USGS to Planetary Visions in the UK for further processing and production of the full-color Satellite Imagemap.
Overlapping images of Europe and North Africa from adjacent orbits show some of the problems associated with raw satellite data. Although the images show the same part of the Earth's surface in the overlap area, the land features appear distorted due to the extremely wide field-of-view of the satellite's camera system, the Advanced Very High Resolution Radiometer (AVHRR). The geometry of the images must therefore be corrected before they can be stitched together to form a composite image of a larger area.
Daily composites of AVHRR images are geometrically corrected to an equal-area map projection of the whole world. The brightness of the image is also radiometrically corrected for variations in illumination and view angle, atmospheric scattering and absorption. The daily composite shows that large areas of each image contain clouds. To obtain a meaningful map of the land surface, images must be gathered over a longer period of time.
Ten-day composites are created from a stack of daily composites, with each pixel chosen to give the most cloud-free view of the land surface. This process was performed automatically on the basis of the different reflectance profiles of vegetation and cloud. Significant areas still remain covered by cloud for the full ten-day period. Each ten-day composite image consists of 3.6 gigabytes of 32-bit computer data.
A Vegetation Index image is computed for each ten-day period, by comparing the reflectance of the channel 1 image (red) with that of channel 2 (near-infrared). This gives a measure of the vigor of plant life containing chlorophyll and can be used to track variations in vegetation cover from season-to-season. High vegetation index values are shown in green and low values in yellow, with white representing no data due to cloud, snow or ice cover.
A single ten-day composite view of the British Isles shows large areas obscured by clouds. By taking several ten-day composites during the northern summer, a super-composite image was produced with minimal cloud cover. Several more ten-day composites were used to give summer coverage in the southern hemisphere. Some small scattered clouds and cloud shadows survive the automatic compositing process, however, and these had to be masked out by hand - a very time consuming process.
True color is not recorded by the AVHRR instrument, which has only one channel in the visible part of the spectrum. However, the same channels that are used to produce the vegetation index image can be used to produce a simulated natural color image. By relating the channel 1 and channel 2 values to the red, green and blue values of known land cover types, a consistent color scheme was established and applied globally. For instance, vegetation appears bright in the infrared channel and dark in the visible channel, whereas desert areas appear bright in both. The color scheme was calibrated in key locations where ground-level or Space Shuttle photography was available.
The compositing process which selects cloud-free pixels over the land does not work over water, so water areas are lost early in the processing. To achieve natural color in oceans and lakes, other data sources are therefore required. While some lakes and coastal areas are covered by high-resolution photography from the Space Shuttle, global coverage of the world's oceans was obtained from two low-resolution datasets. Bathymetry data - digital terrain data of the sea bed - was used in shallow seas, which appear brighter due to light reflected from the sea bed or from coastal sediment suspended in the water. Chlorophyll concentration was used in deep ocean and coastal areas, showing where plankton in the surface waters absorb light. The chlorophyll concentration data was derived from a ten-year composite of readings from the Coastal Zone Color Scanner (CZCS) on NASA's Nimbus-7 satellite. The courses of major rivers were plotted from a digital map database and merged in with the Imagemap where appropriate.
Digital terrain data has been used to enhance the appearance of landscape features in the Satellite Imagemap. A digital terrain model is a grid of height measurements showing the elevation of the land surface above sea level, shown as a brightness image on the far left, where bright areas represent higher land. A computer program can be used to "illuminate" a digital terrain model from any direction, casting shadows to produce a shaded image. Illumination is from the northwest in the terrain-shaded image (near left).
Satellite image data with a 1 kilometer pixel size shows large-scale physical features such as mountain ranges, forests and river valleys by their different colors and textures (far left). By adding terrain-shading, the satellite image can be enhanced to show the shape of the landscape as well as its color (near left). Before adding terrain-shading, the satellite image data was geometrically transformed from the equal-area map projection into a geographical latitude-longitude coordinate system to match the digital terrain data grid.
The Satellite Imagemap was created by Planetary Visions using data provided by the US Geological Survey EROS Data Center, computer hardware provided by Acorn Computers and image processing software provided by Spacetech Ltd.
SAMPLE SATELLITE IMAGERY
To view representative blocks of satellite imagery at 1 km, 4 km and 8 km resolution, Review Samples on the sample satellite imagery web page showing detailed area of the USA / West Coast, Europe / Italy, South America / Chile, and the Persian Gulf / Middle East. In addition to the 1 kilometer satellite imagery, we also offer a range of very high resolution sources running from as detailed as 2 meter, thru 10, 15, 20 and 30 meter, 160 and 250 meter from various remote sensing sources. Check out the Hi-Res Satellite sample images on our Earth Observation Data Sources page.