Julie A. Robinson, Brett McRay, Edward L. Webb*, and S. Alan Spraggins
Earth Sciences & Image Analysis Laboratory (Lockheed Martin Space Operations), Johnson Space Center, Houston, TX
* School of Environment, Resources & Development, The Asian Institute of Technology, Thailand
Reference for Original Presentation: Robinson, J.A., B. McRay, E. L. Webb, S. A. Spraggins. Multi-sensor
approaches to urbanization: using astronaut photography of Earth to fill data gaps. American Geophysical
Union Annual Meeting, San Francisco, 6-10 December 2002. Eos Trans. AGU, 83(47), Fall Meet. Suppl., Abstract
Scientists studying the rapid growth of urban areas around the globe often must
combine a variety remote sensing sources to get data that meets their needs. Depending on research questions,
these needs may include observations at specific points in time or with sufficient spatial resolution. Though
not widely used for urban studies, astronaut photography can uniquely fill some data gaps. Early photographs
taken during the Gemini and Apollo programs in the 1960s represent the oldest remotely sensed records for a
number of world cities. The archive of astronaut photography of Earth is maintained in a single location,
offers valuable information on urban boundaries over the last 40 years, and can fill gaps in time series
studies. Although working with digitized photographs differs from satellite data, such images can be used as
3-band data, georeferenced, and used with image analysis techniques such as supervised classification and
texture analysis. The results of simple land cover classifications can approximate results that would be
obtained from Landsat. Using digital cameras from the International Space Station, astronauts are now
routinely acquiring photographs of urban areas with 6 m or better spatial resolution. These images serve as
valuable sources of information that can be analyzed directly or used to verify analyses of other sensor data.
Astronaut photographs of cities are available for public searching on the web at NASA Johnson Space Center's
"The Gateway to Astronaut Photography of Earth." The site includes tools
for searching the over 400,000 photographs taken to date as well as a special collection, "Cities from Space," of
outstanding city photographs.
A. Astronaut photography includes the oldest records of city "footprints" from an orbital vantage point
- Astronauts have been photographing Earth since the first orbital flights in the 1960s.
- Cities have been photographic targets from the beginning.
The population of the Cairo metropolitan area has
increased from less than 6 million in 1965 when the first picture was taken, to more than 10 million in 1998
(United Nations Population Division, World Urbanization Prospects, the 1999 revision). Population densities
within the city are some of the highest in the world and the urban area has doubled to more than 400 square
km during that period. Extraordinary rates of population growth are expected to continue, with a predicted
population of around 14 million by 2015. (Gem05-1-45778 and STS088-739-91)
Changes in the "footprint" of Mexico City between 1969
and 1996. These two images differed greatly in scale and look angle. It was still possible to georeference
the images to one another. The inset shows detail from the 1969 image that has been rotated to match the
orientation of the 1996 image. A sketch map illustrating the major canals and other distinctive features
visible in the images. Small arrows indicate Licenciado Benito Juarez International Airport used for scale.
(AS9-19-3011 and NM22-741-54B, Robinson et al. 2000).
Built-up areas identified in 1969 or 1974 photographs
(yellow) and 1996 photographs (red) overlaid on the registered and resampled image from NASA-Mir (1996):
(A) San Francisco Bay Area; (B) Mexico City; (C) Vancouver; (D) Dallas;
(E) Chicago; (F) Las Vegas. (Full version of image.)
B. Astronaut photography can be used as 3-band digital data for remote sensing image analysis techniques.
- Film can be digitized with high resolution to reconstruct red, green, and blue (or infrared, red, green)
bands, while digitally-acquired images are already 3-band multispectral images (Robinson et al. 2002).
- Depending on image parameters, basic land use classifications can perform as well as
classifications using Landsat.
- Texture analysis has also been demonstrated using astronaut photography digitized from film.
Orbital photography can be used for land use
classification. In this example of land use in coastal Thailand, the original orbital photography (
top) and Landsat TM (bottom) images were classified and overlaid to detect the agreement between
ground-referenced classifications. For the complete scenes, overall classification accuracy was 81.3%
for the orbital photographs and 83.3% for the Landsat TM. The major differences between the
classifications are at the edged of polygons, because the astronaut photograph has 10 m spatial resolution
compared to 30 m for Landsat TM. (STS059-100-58 and 59,
Landsat TM, Webb et al. 2000).
Texture analysis using astronaut photography. In this
example, a film photograph was digitized to generate a 3-band image. To identify differences in vegetation
patterns, we mean and standard deviations of texture in a 3 ´ 3 window, and then grouped areas of high
texture variability using an unsupervised classification. In this case, the marked areas correspond to
elephant damage to vegetation in Botswana. (STS008-33-993,
Robinson et al. 2001).
C. Spatial resolution of recent astronaut photography from the International Space Station exceeds 6 m / pixel and has focused on cities of the world. This imagery can be used as data in its own right, or as higher resolution ground "truth" imagery.
D. A 40-year searchable archive provides access to all 450,000 + photographs of Earth taken by astronauts and supplements time series studies of cities.
- Web-based collections of the best city images provide rapid access to city images, while detailed technical search tools allow identification of images that meet particular research needs. All tools and digital images currently are available online for free at http://eol.jsc.nasa.gov/sseop/.
- Because astronaut photographs vary in spatial resolution, we have developed tools for quickly estimating approximate spatial resolution based on characteristics recorded in the database such as seen in our Cities Collection at http://eol.jsc.nasa.gov/cities/.
Robinson, J. A., D. A. Liddle, C. A. Evans, and D. L. Amsbury. 2002. Astronaut-acquired orbital photographs as digital data for remote sensing: spatial resolution. International Journal of Remote Sensing, 23(20):4403-4438.
- Jensen, J. R. 1983. Urban/suburban land use analysis. In J. E. Estes (ed.) Manual of Remote Sensing, 2nd ed. Vol. II. Interpretation and Applications. Falls Church, Virginia, Am. Soc. Photogram., pp. 1571-1666
Robinson, J. A. and C. A. Evans. 2002. Space Station Allows Remote Sensing of Earth to within Six Meters. Eos, Transactions, American Geophysical Union 83(17):185, 188.
Robinson, J. A., K. P. Lulla, M. Kashiwagi, M. Suzuki, M. D. Nellis, C. E. Bussing, W. J. Lee Long, and L. J. McKenzie. 2001. Conservation applications of astronaut photographs of Earth: tidal flat loss (Japan), elephant impacts on vegetation (Botswana), and seagrass and mangrove monitoring (Australia). Conservation Biology 15:876-884.
Robinson, J. A., B. McRay, and K. P. Lulla. 2000. Twenty-eight years of urban growth in North America quantified by analysis of photographs from Apollo, Skylab and Shuttle-Mir. In Dynamic Earth Environments: Remote Sensing Observations from Shuttle-Mir Missions. John Wiley & Sons, New York, pp. 25-42. <http://eol.jsc.nasa.gov/newsletter/DynamicEarth/Chapter3/Cp3.htm>.
Webb, E. L., Ma. A. Evangelista, and J. A. Robinson. 2000. Digital land use classification using Space Shuttle-acquired orbital photographs: a quantitative comparison with Landsat TM imagery of a coastal environment, Chanthaburi, Thailand. Photogrammetric Engineering & Remote Sensing 66:1439-1449.
Welch, R. 1982. Spatial resolution requirements for urban studies. International Journal Remote Sensing, 3: 139-146.
Back to Online Publications