On the use of space photography for identifying transportation routes; a summary of problems

It has been widely suggested that space photography may be used for updating maps of transportation networks. Proponents of the argument have suggested that color space photographs of the resolution obtained with Hasselblad 80 mm lenses (about 300 feet) contain enough useful information to update the extensions of major U . S . highways. The present study systematically documents for the Dallas-Fort Worth area ' the potential of such'space photography in detecting, and to a lesser degree identifying, the existing road networks.Color separation plates and an enlargement of the color photograph were produced and all visible ; roads traced onto transparencies for study. In addition, a four county : sample area was selected for more detailed analysis in comparing the actual network with those roads detected and other linear elements mistaken for roads. Attempts were made to discover to what degree road width, surface type, topography, land use, and linearity affect road visibility. ' . ' Major roads and roads under construction were the most visible while lower class roads and roads in urban areas had the poorest return. Road width and classification were found to be the major determinant in visibility, varying from 100'per cent visible for divided highways to 15 per cent visible of bladed earth roads. In the five county sample area, 4 per cent of the linear elements detected were not roads; but were rather-streams and rivers, field borders, pipelines, telephone cables, oil field borders, railroads, lakeshore, waterways, trails to'oil rigs, and unknown elements. The sun was at an angle of some 40° above the horizon at the time of the photography, and, as would be anticipated, ^ roads vary in visibility as a function of fore or back-lighting. -In summary, space photographs of this resolution proved to be difficult to use for accurate road delineation. Only super highways in rural areas with the greatest road-width v/ere completely identifiable, the width being about 1/3 that of the resolution cell. Lower class roads were confused with other linear elements or were simply not visible. ' We would anticipate that, to positively detect and identify all but the : smallest of roads, the resolution should be about 50 feet. INTRODUCTION The problem of using outdated road and transportation maps or finding accurate new ones has confronted almost everyone from time to time. This is true not only for super highways in the U . S . but also in identifying trails and dirt roads in remote areas that are incorrectly listed or omitted entirely due to outdated or inadequate information. Present methods rely on costly field studies and random if any air photo reconnaissance. . .,V; -Supported .by U:. S. G . S.: Contra ctUSG 8^4-08^00]J0848 -,^^^,fc^^£ !v^0-w5.̂ ;.̂ ,.î >;g ,̂;>^^%^*^»^ i":;"?.**.-.:-•'"-" • • ~-~ ' " ; • . ia.:v;.i^>_-->.'lt>ha.s.beenj.wi_dely.jsugg_este.^.jhati,space photo.graphs. may be used for .up-dating maps of . -••transportation networks. .Trie prppqnents"df"s.uch 'a View"afgue~i.bat ihe'exfremely^high^'sltitude of ̂ Af^-J^ ^'^ space photography would permit coverage of large areas with single photos, thus eliminating many -•'=•* • • : . : .,,.., inaccuracies in present procedures, while consistently providing current information on the area-. ' . ' " ' • New and existing roads could be identified and transportation maps accurately and easily up-dated. The necessary construction of new and expansion of old transportation networks into areas of recent and future urban development could be rationally planned. Existing natural and future man-made obstacles to proposed road construction could be readily foreseen and compensated for, thus eliminating much costly right-of-way purchase at a later date. Presumably, this same identification ability in up-dating maps would be even more valuable in road analysis of remote areas. In order to use space photography for such purposes it is necessary not only to detect the presence of linear elements (i.e. roads,, railroads, pipelines, etc.) but also consistently to iden. . tify and discriminate between them. It has been suggested that color space photographs of the resolution obtained with the 80 mm lens employed in the Apollo VI space photography, which lie in the range of 180 to 450 .feet (on axis and 20° off axis) with a contrast ratio of 1.6:1,' contain enough useful information to update transportation maps showing at least the major highways in the U. S. , / and that higher resolution photographs, obtained with 150 to 300 mm focal lengths would enable even minor roads to be detected. ' ' . • . . . • -•.«*r:.^-v,,-«-jî .-^^.-i^4-%;vU«rt!,^Ctao./v«>.: -.. . . :. -. : g,̂ '. . . . . . . . . . ,_ . . . , The present study, the first of a series, systematically documents for the Dallas-Fort Wortli ' •'-''--' area the potentials and problems of Apollo VI color space photography in the detection and identi. fication of existing road networks. This area was selected for examination because of the apparent number of visible roads and the variety of landscapes, soils and land uses available. Figure 1 is a 3 times reproduction of the space photograph showing the area covered, while Figure 2 gives both the Interstate and U. S. Highways, with major towns and cities. We have found in this study that, to be consistently detected a road normally should have a width (road plus shoulder width) of equal to or greater than one-fourth of the resolution cell at a contrast ratio of 1.6 to 1. For the Dallas area this means that only roads (actually roads plus shoulder width) no less than 70 feet widewould be fully, and unambiguously, detected. As the road width decreases other elements enter into a resolution cell resulting either in misidentification of roads with other linear elements, or a failure to detect roads. Consequently, for certain identification we estimate at this time that the resolution of a f i lm should be of the order of the narrowest road which one desires to identify. With resolutions of 50 to 100 feet one should be able to detect paved, two lane highways, and a fifteen foot resolution should enable detection even of narrow dirt roads in most cases, provided sufficient background contrast is maintained. Further studies are planned for the fall of 1969 using photographs with resolutions of 10 feet to 100 feet to document the validity of this hypothesis. • CAMERA AND FILM DATA ' " .The Apollo VI space photograph shown in Figure 1, is one of a stereo triplet obtained April 13, 1968, at 8:43 a .m. local sun time, from an altitude of 128 statute miles. The camera employed •was a 70 mm J. A. Maurer Model 200-G sequence camera. Each frame (41° field-of-view) was taken 8.64 seconds apart with 65 per cent overlap at 1/500 second and f 5.6. Spacecraft stabilization was such that the photograph has less than one-half degree tip in the direction of flight and one-third degree tilt across track. The lens used was a f 2.8 Kodak Ektar of 76 mm focal length without a filter. In order to reduce the high atmospheric luminance from short wavelength Rayleigh scattering a sharp cutting Wratten 2E haze filter was used which cut with 6.5 per cent transmission at 420 nanometers, 59 per cent at 430,nm, 80 per cent at 440 nm, and 90 per cent at 480 nm. Beyond 500 nm it is virtually flat in its response at 91 per cent transmission to 720 nm. The film used was Kodak type SO-121, a high resolution, high contrast, aerial color positive. . . •( Since sufficient quantities of flight emulsion were not available prior to shipment of the camera to the launch site no resolution testing was done using the flight film and flight lens. .However the manufacturer's lens test at f 2.8, 1/1000 sec with Plus-X (type 3401) film gave the. \ following average of radial and tangential resolutions with a high contrast target: 0", 71 line pairs/ ;. mm; 5°, 70; 10°; 65.S; 15°, 46; 20°. -35. 5; and 25° , 13. Areal weighted average resolution was 45 line pairs/mm. It has not proved possible for us to test this lens with SO-121 film but, from performance tests with other lenses and films we'believe this lens-film combination would be In the preparation of this section we have drawn heavily from an updated, unsigned manuscript "Camera System and Calibration," Apollo AS-502 (Apollo VI) by the Instrumentation and Electronic Systems Division, Manned Spacecraft Center, Houston, Texas. Quotations labelled 'camera system' are from this document. . • • likely to have a performance at a contrast ratio of 1.6:1 (a common ratio between roads and background in this area, as well as being a standard value) of the order of 50 line pairs at 0° (nadir) and 5°; 45 at 10°; 30 at 15°, 2Q at 20° , and 10 or less at 25° . These resolutions give equivalent ground resolutions on the ?0 mm film, with a mean scale of 1:2,800,000 of 180 feet at 0° and 5°, 200 at 10°, 300 at 15°, 450 at 20° , and 900 at 25°. A rough areal weighted average resolution would be 30 line pairs pe'r mm, equivalent to 300 feet with a 1.6:1 contrast ratio. The photographs were taken through a hatch window inclined at such an angle to the camera that the light rays reaching the camera ranged from "near normal at the comers in the direction of flight to approximately 10° at the center of the leading edge. The angles of incidence increase toward the trailing edge when they become greater than 50 degrees" (Camera Systems). The significance of this is that using a refractive index of 1.5442 for quartz, the maximum polarization of light in the plane of the surface occurs at 57°4 ' . In addition the Apollo window consists of a series of plates, thus multiplying polarization effects as the rays pass from surface to surface, :first through an optical grade quartz heat shield, then multilayer blue-red reflection coatings, and then through 2 panes of Corning #1723 aluminosillacate glass, themselves coated with reflection reducing layers. 'Thus the light rays reaching the film