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.


Astronaut-acquired Orbital Photographs as Digital Data for Remote Sensing: Spatial Resolution

Click here for the PDF version


Menu Introduction Background Factors that determine the footprint Other characteristics that influence spatial resolution Estimating spatial resolution of astronaut photographs Summary and Conclusions 4. Other characteristics that influence spatial resolution

     Beyond basic geometry, many other factors internal to the astronaut photography system impact the observed ground resolved distance in a photograph. The lenses and cameras already discussed are imperfect and introduce radiometric degradations (Moik 1980). Vignetting (slight darkening around the edges of an image) occurs in astronaut photography due to light path properties of the lenses used. A lack of flatness of the film at the moment of exposure (because cameras used in orbit do not incorporate a vacuum platen) also introduces slight degradations. We list a number of additional sources of image degradation that would affect GRD of astronaut photographs.

4.1. Films and processing

4.1.1. Colour reversal films
Table 3      Most films used in NASA hand-held cameras in orbit have been E-6 process colour reversal films (Ektachromes) that have a nominal speed of 64 to 100 ASA, although many other films have been flown (table 3). Reversal films are used because damage from radiation exposure while outside the atmospheric protection of Earth is more noticeable in negative films than in positive films (Slater 1996). The choice of ASA has been to balance the coarser grains (lower film resolving power) of high-speed films with the fact that slower films require longer exposures and thus are affected more by vehicle motion (approximately 7.3 km s-1 relative to Earth's surface for the Space Shuttle). Extremely fast films (> 400 ASA) are also more susceptible to radiation damage in orbit (particularly during long-duration or high-altitude missions, Slater 1996) and have not traditionally been used for Earth photography. The manufacturer stock numbers identifying the film used are available for each image in the Astronaut Photography Database (Office of Earth Sciences 2000).

4.1.2. Colour infrared films
     Colour infrared film (CIR) was used during an Earth-orbiting Apollo mission (Colwell 1971), in the multispectral S-190A and the high-resolution S-190B camera systems on Skylab (NASA 1974, Wilmarth et al. 1977), and occasionally on Shuttle missions and Shuttle/Mir missions (table 3). The CIR film used is a 3-layer Aerochrome, having one layer that is sensitive to reflected solar infrared energy to approximately 900 nm (Eastman Kodak Company 1998); protective coatings on most spacecraft windows also limit IR transmittance between 800 and 1000 nm. This layer is extremely sensitive to the temperature of the film, which creates unpredictable degradation of the IR signature under some spaceflight conditions.

4.1.3. Film duplication
     The original film is archived permanently after producing about 20 duplicate printing masters (2nd generation). Duplicate printing masters are disseminated to regional archives and used to produce products (3rd - 4th generation) for the media, the public, and for scientific use. When prints, slides, transparencies, and digital products are produced for public distribution, they are often colour adjusted to correspond to a more realistic look. Digital products from 2nd generation copies can be requested by scientific users, and we recommend them for remote sensing applications. Care should be taken that the digital product acquired for remote sensing analysis has not been colour adjusted for presentation purposes. Based on our qualitative observations, there is little visible increase in GRD in the 3rd or 4th generation products. There is a significant degradation in fidelity of colour in 3rd and 4th generation duplicates, because an increase in contrast occurs with every copy.

4.2. Shutter speeds
     The impact of camera motion (both due to the photographer and to the motion of the spacecraft) on image quality is determined by shutter speed—1/250 to 1/500 second have been used, because slower speeds record obvious blurring caused by the rapid motion of the spacecraft relative to the surface of the Earth. Generally 1/250 was used for ISO 64 films (and slower), and after the switch to ISO 100 films, a 1/500 setting became standard. New Hasselblad cameras that have been flown beginning with STS-92 in October 2000 vary shutter speed using a focal plane shutter for exposure bracketing (rather than varying aperture) and 1/250, 1/500 and 1/1000 second are used.

     Across an altitude range of 120 - 300 nautical miles (222-555 km) and orbital inclinations of 28.5° and 57.0°, the median relative ground velocity of orbiting spacecraft 7.3 km s-1. At a nominal shutter speed of 1/500, the expected blur due to motion relative to the ground would be 14.6 m. When cameras are handheld (as opposed to mounted in a bracket), blur can be reduced by physical compensation for the motion (tracking) by the photographer; many photographs show a level of detail indicating that blur at this level did not occur (e.g. figure 3, D). Thus, motion relative to the ground is not an absolute barrier to ground resolution for handheld photographs.

4.3. Spacecraft windows
     Most of the photographs in the NASA Astronaut Photography Database were taken through window ports on the spacecraft used. The transmittance of the window port may be affected by material inhomogeniety of the glass, the number of layers of panes, coatings used, the quality of the surface polish, environmentally induced changes in window materials (pressure loads, thermal gradients), or deposited contamination. Such degradation of the window cannot be corrected, is different for each window, and changes over time. See Eppler et al. (1996) and Scott (2000) for discussion of the spectral transmittance of the fused quartz window that is part of the U.S. Laboratory Module (Destiny) of the International Space Station.

4.4 Digitised Images from Film
     When film is scanned digitally the amount of information retained depends on the spectral information extracted from the film at the spatial limits of the scanner. To date, standard digitising methodologies for astronaut photographs have not been established and film is digitised on a case-by-case basis using the equipment available.

4.4.1. Digitising and Spatial Resolution
     Light (1993, 1996) provided equations for determining the digitising spatial resolution needed to preserve spatial resolution of aerial photography based on the static resolving power (AWAR) for the system. For films where the manufacturer has provided data (Eastman Kodak 1998, K. Teitelbaum, pers. comm.), the resolving power of films used for astronaut photography ranges from 32 to 100 lp/mm at low contrast (object/background ratio 1.6/1, table 3). The AWAR for the static case of the Hasselblad camera has been measured at high and low contrast (using lenses shown in table 1), with a maximum of approximately 55 lp/mm (Fred Pearce, unpubl. data).

     Based on Light's (1993, 1996) method, the dimension of one spatial resolution element for a photograph with maximum static AWAR of 55 lp/mm would be 18 mm/lp. The acceptable range of spot size to preserve spatial information would then be

and 6 mm £ Scan Spot Size £ 9 mm. Similarly, for a more typical static AWAR of 30 lp/mm (33 mm/lp, low contrast, Fred Pearce, unpubl. data) 11 mm £ Scan Spot Size £ 17 mm. These Scan Spot sizes correspond to digitising resolutions ranging from 4233–2822 ppi (pixels/inch) for AWAR of 55 lp/mm and 2309–1494 ppi for AWAR of 30 lp/mm. Scan Spot sizes we calculated for astronaut photography are comparable to those calculated for the National Aerial Photography Program (9–13 mm, Light 1996)

     Widely available scanners that can digitise colour transparency film currently have a maximum spatial resolution of approximately 2400 ppi (10.6 mm/pixel). For example, we routinely use an Agfa Arcus II desktop scanner with 2400 ppi digitising spatial resolution (2400 ppi optical resolution in one direction, and 1200 ppi interpolated to 2400 ppi in the other direction) and 2400 ppi was used to calculate IFOV equivalents (tables 2 and 4). For some combinations of lens, camera, and contrast, 2400 ppi will capture nearly all of the spatial information contained in the film. However for film with higher resolving power than Ektachrome-64, for better lenses, and for higher contrast targets, digitising at 2400 ppi will not capture all of the spatial information in the film.

     Improvements in digitising technology will only produce real increases in IFOV to the limit of the AWAR of the photography system. The incremental increase in spatial information above 3000 ppi (8.5 mm/pixel) is not likely to outweigh the costs of storing large images (e.g. Luman et al. 1997). At 2400 ppi, a Hasselblad frame is approximately 5200 ´ 5200 or 27 million pixels (table 2) while the same image digitised at 4000 ppi would contain 75 million pixels.

     Initial studies using astronaut photographs digitised at 2400 ppi (10.6 mm/pixel, e.g. Webb et al. in press, Robinson et al. 2000c) indicate that some GRD is lost compared to photographic products. Nevertheless, the spatial resolution is still comparable with other widely used data sources (Webb et al. in press). Robinson et al. (2000c) found that digitising at 21 mm/pixel provided information equivalent to 10.6 mm/pixel for identifying general urban area boundaries for 6 cities, except for a single photo that required higher spatial resolution digitising (it had been taken with a shorter lens and thus had less spatial resolution). Part of the equivalence observed in the urban areas study may be attributable to the fact that the flatbed scanner used interpolates from 1200 to 2400 ppi in one direction. The appropriate digitising spatial resolution will in part depend on the scale of features of interest to the researchers, with a maximum upper limit set of a scan spot size of approximately 6 mm (4233 ppi).

4.4.2. Digitising and Spectral Resolution
     As for spatial resolution, radiometric or spectral resolution is a function of both the original image on film and digitising parameters. When colour film is digitised, there will be loss of spectral resolution. The three film emulsion layers (red, green, blue) have relatively distinct spectral responses, but are fused together so that digitisers detect one colour signal and must convert it back into three (red, green, blue) digital channels. Digital scanning is also subject to spectral calibration and reproduction errors. Studies using digital astronaut photographs to date have used 8 bit/channel. However, this is another parameter that can be controlled when the film is digitised. We do not further address spectral resolution of digitally scanned images in this paper.

4.5. External factors that influence GRD
     Although we do not discuss them in detail in this paper, factors external to the spacecraft and camera system (as listed by Moik [1980] for remote sensing in general) also impact GRD. These include atmospheric interference (due to scattering, attenuation, haze), variable surface illumination (differences in terrain slope and orientation), and change of terrain reflectance with viewing angle (bidirectional reflectance). For astronaut photographs, variable illumination is particularly important because orbits are not sun-synchronous. Photographs are illuminated by different sun angles and images of a given location will have colour intensities that vary widely. In addition, the viewing angle has an effect on the degree of object-to-background contrast and atmospheric interference.

<< Back            Next >>