NATMAP EARLY DAYS, MAP COMPILATION FROM AERIAL PHOTOGRAPHS 1948 ‑1970S

 

David R. Hocking

 

 

 

This paper describes the methods used to compile maps from aerial photographs, commencing 50 years ago, by the National Mapping Section / Office as it was variously known in those early days before becoming the Division of National Mapping (Natmap). Some of these products were: aerial photo indexes, mosaics, radial line plots, slotted template assemblies, 'shift & trace' or Zeiss Sketchmaster or Wild A6 plotted map compilation sheets. These map‑substitutes and planimetric small scale maps at 1:253 440 and 1:250 000 were urgently needed by users, such as Natmap surveyors and geo‑scientists who wanted information about the country they were working in and needed to record their findings as accurately as possible in relation to the terrain. With the first priority special mapping needs being satisfied, a more precise method of slotted template planimetric adjustment of very large blocks of aerial photos for horizontal position at 1:100 000 scale was introduced in the 1960s. Vertical control for the 20 metre contour interval specified for the 1:100 000 scale topographic map series was obtained using airborne radar /laser terrain profiling along the sidelaps of the aerial photo coverage. Stereoplotting of map detail and contouring from stereoscopic models formed from overlapping aerial photos using Kem PG2 and Wild B 8 instruments is described.

 

 

 

 

Aerial photographs, until recently, have been the main provider of topographic information for a national mapping program. Satellite imagery is now being used to revise major features of the small scale, 1:250 000 data base. The high resolution, 1 metre ground sample distance, satellite imagery expected to be available during 1998/99 will be used to upgrade the medium and large scale, 1:100 000 and 1: 250 000 data bases. Aerial photography will continue to be used for the large scale project mapping needed for engineering, mining, construction, and so on and, no doubt, both digitized aerial and satellite imagery will be used on some projects.

 

During the past 80 years the major events in the imagery available for mapping Australia at small and medium scales are:

 

The 1:50 000 scale Fairchild K17 camera aerial photos obtained from 25 000 feet by RAAF, 87 PR (Photo Reconnaissance) Squadron, 1947‑1953. These photos provided the basic data for the 1:253 440 and 1:250 000 R502 map series.

 

The 1:80 000 scale Wild RC9 or RC10 camera aerial photos taken from 25 000 feet and obtained by Natmap using contractors during the 1960s and 1970s. These photos provided the basic data for the 1: 100 000 series and the 1:250 000 National Topographic Map Series.

 

The eagerly awaited availability in 1998/99 of 1 metre high resolution images from the Space Imagery Eosat (SIE) 'Ikonos 1' satellite. This imagery will provide the basic data for up‑grading medium and large scale data bases. See www.spaceimage.com for more details.

 

SUMMARY OF AERIAL PHOTOGRAPHY DEVELOPMENT IN AUSTRALIA

 

1920s              After the First World War some general purpose cameras were used and photos taken with the camera held over the side of the aeroplane.

 

1924                RAAF, Royal Australian Air Force, first areas for mapping, using a P18 plate camera which had a 4"x 5" format.

 

1927                RAAF, F8 (Eagle 1), 100 exposures on film, 7"x 8¼" format with 8¼" lens - 19 000 square miles photographed in eastern Australia.

 

1930                RAAF, F8 or Williamson Eagle 111.

 

1936                RAAF, Williamson Eagle IV, 7"x 9" format. Commercial aerial survey companies started operating.

 

1939                RAAF photographed areas for NAS ‑ (Aerial Geographical and Geophysical) Survey of Northern Australia.

 

1940                RAAF used Eagle IV to photograph large areas of Australia.

 

1942‑45           US Air Force acquired small scale reconnaissance photography with simultaneous exposures of three cameras, one vertical. two oblique depressed 30 degrees, one port and one starboard normal to the flightline, to cover strategic areas of Australia.

 

1943                RAAF obtained the Fairchild K17 9 x 9 / 6 inch, wide angle (90 degrees) Metrogon lens cameras. (Fig. 1)

1944                Victoria and Tasmania set up state government aerial survey agencies to provide photogrammetric services. Aerial photography obtained by commercial companies.

 

1945‑53           Demand increases with RAA.F us the Fairchild K 17, 9 x 9 / 6 inch and, later   Williamson OSC (Ordnance Survey Camera).

 

Civilian agencies use the Williamson Eagle IX, 9 x 9 with a choice of 6, 8¼, 10, 12 or 14 inch focal length lens. In 1948, the RAAF 87 (PR) Squadron photographed

~ 650 000 square km in northern Australia. Until 1954 all aerial photography work for the Commonwealth was done by the RAAF, ‑ 3.25 million km2. (Fig. 2).

 

1954‑59           Commercial companies acquired aerial photography for the Commonwealth.

 

1960                Natmap purchased Wild RC9, 230 x 230 / 88 mm superwide angle SWA (120 degrees) camera. Initially, the RC9 camera was hired to the successful contractor, but contractors such as Adastra Airways, Civil Aerial Surveys, Kevron Aerial Surveys and Queensland Aerial Survey Co. soon purchased their own SWA cameras, either Swiss Wild or German Zeiss.

 

1976‑80           High altitude 12 500 m 'Lear Jet' Wild RC 10, 150 mm (1: 80 000) and 88 mm

(1:140 000) scale photography.

 

 

1970‑98           Commonwealth, State and Territory mapping and geo‑spatial information agencies increased their use of commercial aerial survey companies for aerial photography,

photogrammetric, survey and mapping services (see www.ozemail.com.au/~aasa for more details).

 

High resolution Satellite Imagery

 

The Earthwatch, 'Earlybird' 3 metre, high resolution imaging satellite was unfortunately lost in space (see www.digitalglobe.com). Space Imaging Eosat's, 'Ikonos' 1 metre, high resolution imaging satellite was due to be launched in 1998, but has been delayed until mid‑1999 (see www.spaceimage.com).

 

 

K17 AERIAL PHOTOGRAPHY

 

In 1947 a 'Four‑mile to the inch', 1: 253 440 scale map area of 1 degree latitude x 1½ degrees longitude was covered by 15 runs of 40, K17 aerial photos taken from 25 000 feet with 60% forward lap and 25% side lap, a total of about 600 photos for stereo cover or 300 for non‑stereo coverage. When Natmap field survey work started in 1948, aerial photos were scarce with a single set, the field set of matte photos, being shared between the Natmap and CSIR (Commonwealth Scientific and Industrial Research) field parties. Reliable maps did not exist and field workers had to lay out the photos in a rough shingle mosaic to get some idea of the country they were working in. Laying out hundreds of photos in the field was a challenging task with any sort of a breeze blowing!

 

The Tennant Creek Reconnaissance Map, produced in 1948 by W.A. (Alan) Thomson and K.O. (Ken) Johnson, was the earliest compilation by Natmap using the photogrammetric method of radial line plotting of map detail from aerial photos. (Salt 1933) (Fig. 3)

 

Photo Indexes were prepared by plotting aerial photo centres on the best available base map and joining these to form the runs of photos. Runs were labelled with the first and last photos numbered and every fifth photocentre along the run. Photo coverage was then checked for compliance with coverage and overlap specifications.

 

Photo Mosaics in this era were usually prepared by laying a framework of slotted template runs at photo scale (1:46 500) between survey control points. Ideally, control would be identified on Runs 1 and 15 and East and West Key Runs. (Fig. 4) The slotted template positions would be pricked through to the kraft paper base sheets, the photos laid over the framework positions with the other photos used to fill in the gaps by matching detail. (Fig. 5) This was known as a shingle mosaic with Run 1 overlapping Run 2 and so on. Run and photo numbers were added, major detail annotated and then the shingle mosaic was photographed in six sections, (Fig. 6) mosaiced at 7 miles to an inch, and enlarged to 4 miles to an inch. (Fig. 7).

 

These 1:253 440 (4 miles to the inch) scale photomaps were produced for use as a base for geological, soil, timber, aeronautical, geographical and other maps. The photomaps were prepared from unrectified aerial photographs controlled by slotted template plotting based generally on astronomical fixations. (NMO 1955 Map Catalogue).

 

 



 

 



SLOTTED TEMPLATE ASSEMBLY

 

A photograph is not a map. However, on a near vertical perspective photograph of terrain with gentle slopes, directions from the photocentre to points of detail on that photo can be considered true radial directions. This is the basis of radial line plotting and the Slotted Template Assembly (STA) method of extending control points for mapping from aerial photographs.

 

In 'Point Selection', photocentres and points of detail, called pass points, tie or wing points, were selected and transferred to the overlapping photos along the run and to the adjoining runs. Ideally, some survey control points were selected as pass points.

 

In 'Template Making', the photocentre and pass points were pricked through from the photo on to the template material, such as exposed x‑ray film. A hole was punched at the photocentre and slots cut, using a 'Cassella' slotted template cutter, through the pass points and transferred photocentres to represent the radial directions. The slotted templates were annotated with the run number and photo number then trimmed.

 

'Template Assembly' involved drawing a rectangular coordinate grid at the nominal photo scale, for example 1:46 500, on the assembly base board and plotting the survey control points. Studs were fixed at control points and templates laid between the control points, along the runs and adjoining runs with floating studs at the photo centres and in the slots. The slotted template assembly (STA) was laid flat without strain the positions of the photo centres and pass points were pricked through the holes in the centre of the studs to the base compilation sheets. The templates were taken up, STA positions circled, photo centres labelled and the index to adjoining sheets shown on the drafting film map compilation sheets together with the control points, grid ticks and map sheet corners. (Hocking 1967)

 

 

PLOTTING MAP DETAIL

 

Overlapping photos were viewed stereoscopically and map detail, interpreted, selected and marked up with coloured inks within the lines joining the pass points. Map details were transferred to corresponding positions on the compilation base sheet by 'the shift and trace method' which meant matching the points on the photograph to the STA points on the map compilation base sheet and tracing off the marked up detail. (Salt 1933) It is important to realize that there was more than a 5 to 1 (1:46 500 to 1:253 440) reduction between the compilation scale and the final map scale. Later, Zeiss sketchmasters were used to transfer annotated photo detail to map compilation sheets more quickly and more accurately. (Fig. 8)

 

Some indication of the elevation of the terrain was given by plotting spot heights derived from aneroid barometer readings in the field. The spot heights were obtained about every four miles (one inch on the map) along roads, tracks, at creek crossings, homesteads, aerodromes, high ground etc. (Hocking 1985)

 

The 1:253 440 Planimetric Series was drawn on the Transverse Mercator Projection in zones 5 degrees wide with origins at intersections of parallel of latitude 34ºS with the Central Meridian of each zone. These planimetric maps were produced for some special priority areas of Australia, showing all types of natural and cultural features, including railways, roads, tracks and towns. Relief was indicated by spot heights and hachures. (NMO, 1955 Map Catalogue)

 

TOPOGRAPHIC MAPPING

 

'Precise' Slotted Template Assembly (STA) for 1:100 000 Planimetry

A most significant improvement occurred in 1960 with the introduction of the super wide angle, Wild RC 9 camera taking 1:80 000 scale photos which meant that less than 200 photos (stereomodels) covered a 1:250 000 map area compared with the 600 needed with K17 photo coverage. During the 2 or 3 years transition from K17 to RC9 photos an obvious disadvantage was that photo interpretation of map detail was more difficult on the smaller scale 1:80 000, RC9 photography. However, most Natmap operators were well experienced and soon able to cope with the smaller photo‑scale.

 

Film diapositives (0. 10 mm. thick) were printed using a U4A projection printer fitted with a 7000 metre flying height correction plate for earth curvature, air refraction and RC9 lens distortion. Point marking was done using 8x or 10x magnification stereoscopes and either orange 'Letraset' or black 'Mecanorma' rub­down dot and circle point marks to avoid pricking or drilling holes in the film diapositive.

 

Spot photos of control points were available to allow positive identification of survey control on the diapositives. To obtain spot photos, the control points were marked on the ground, then using a Hasselblad camera near vertical photos were taken at 500, 1500 and 3000 feet above ground. Using a Bausch and Lomb Zoom 95 Differential Stereoscope the control was then transferred from the larger to the smaller scale photographs which provided an accurate position on the mapping photos of the control point. (Fig. 9) This technique was a vital requirement in order to maintain map accuracy.

 

Control for 1:100 000 mapping was based on the Australian Geodetic Datum 1966. However, in the period leading up to the adoption of AGD66 a provisional datum based on the '165' figure (semi‑major axis 6 378 165 m) compared with the adopted '160' m provided coordinates sufficiently accurate for 1: 100 000 scale mapping.

Control point density was ideally half a degree around the perimeter with one degree spacing of control within the block. However, the control point density usually averaged between a half and one degree network of geodetic control (Ford 1979) and airborne Tellurometer 'Aerodist' control. (McMaster 1980)

 

Using the 'South African' pattern slotted template cutter, normal templates were made from 0.35 mm thick white 'Flovic' blanks with a pre‑punched centre hole. (Fig. 10) Azimuth templates were made from 0.35 min thick clear 'Cobex' using the 'Skinner' constant radial offset slotted template cutter. The 'Skinner' cutter facilitated changing the template scale, usually from 1: 80 000 to 1: 100 000. (Fig. 11) Azimuth templates were used to reduce the bowing of a run of templates by bridging with the direction of the photo centre 2 photos distant marked on every second photo. This was done by using the second of three successive photos and accurately transferring the photo centres of the first and third photos to the middle photo. These transferred image points were joined with a fine line and image points along this line transferred to the first and third photos. Then on the first and third photos a line was drawn radially from the photo centre points through these marks. Every second template was an azimuth template and every other a scale template. (Gamble 1950) This technique is analogous to strengthening azimuth in a survey traverse by reading the angle at station 1 to 3 and at station 3 to 5.

 

Stereo templates were used occasionally to locate templates firmly at control, to get a strong fix on a control point; for example, where the control point was close to the base line and a strong radial intersection was not possible. The stereo templates were prepared from stereo instrument plots of a relatively oriented model of overlapping photos, thereby transforming the perspective projection to a parallel projection with the pass points and control point plotted at approximately STA scale. These points were pricked through to two sheets of 'Flovic' and, using the Zeiss Radial Secator RS1, slots were cut from diagonally opposite pass points used as radial centres and a (double) stereo template prepared.

 

Base sheets and overlay sheets were 0.1min double‑matte 'Ozatex' pre‑printed with standard marginal information for 1: 100 000 manuscript mapping. Grid intersections, 10 minute graticule intersections, control points and register holes were plotted using the 'Decograph' coordinate plotter. Contour and vegetation overlays were prepared as plain sheets of the same size and material as the base sheets. Registration holes, 4mm in diameter, were punched near the corners of all base sheets and the base sheets were joined together with 2 registration studs. The internal edges were covered with PVC adhesive tape to provide a smooth surface for the templates to slide on. Control studs were securely stuck in position on the base sheets. The templates were assembled systematically from west to east and north to south. Coast ties and island runs were added when W‑E runs covered the coast. (Fig. 12).

 

After the assembly was completed, laying flat without strain, the stud positions of photo centres and pass points were pricked through to the base sheets in the normal way. As the templates were lifted, all points were circled in pencil and identified as necessary.

 

A general comment on the use of this more precise method of slotted template adjustment is appropriate. Natmap recognized that most of the inland undeveloped areas were relatively flat with slopes less than 5 degrees and laid many very large blocks of slotted templates. To maintain the accuracy of position around the perimeter of the block, an overspill of templates was laid to the next line of control where possible. For example, Block 6 covered 19 x 1:250 000 areas or 110 x 1: 100 000 areas of the Northern Territory. (Fig. 13) Some statistics for Block 6 are:

 

 

 

Block 6

Overspill

Total

No of templates

3076

704

3780

No of studs

6250

1450

7700

                plus control studs

130

 

 

                plus register studs

220

 

 

 

 

 

 

Area

315 000 sq. km.

 

379 200 sq. km.

 

 

It is interesting to compare the Block 6 area with the combined area of the states of Victoria and Tasmania which is approximately 300 000 sq. km. (Some of these comments on Block 6 STA are based on a 1971 report by Mr. R.G. Foster, Supervising Draughtsman. Bob, who unfortunately died recently, was the acknowledged expert on precise slotted template adjustments in Natmap).

 

When mapping under pressure (Lambert 1963), and before massive number‑crunching computers were available for large analytical block adjustments, there was a lot in favour of STA block adjustments of 2000 to 3000 models, with direct transfer of the minor horizontal control positions to the base map compilation sheets. (Fig. 14)

 

 

VERTICAL CONTROL

 

To plot map detail using stereo‑plotting instruments, such as the Wild B 8 and Kern PG 2, in which a virtual model is formed by stereoscopic observation of a pair of overlapping aerial photo film diapositives, the model must be levelled and scaled. Three vertical control points, not in a straight line, are necessary to level the model, and two horizontal control points are needed to set the correct scale.

 

In large scale project mapping it is common to establish the individual vertical control points by direct ground survey methods. However, for the medium scale, 1:100 000 national mapping program with 20 metre contours, the vertical control was obtained by airborne terrain profiling along the sidelaps of the aerial photo coverage and this provided vertical control in every model. Selected key runs were flown over photo‑identified 3rd order level benchmarks and the east‑west profiles adjusted to the level network.

 

Approximately 200 000 km of radar terrain profiling was obtained under contract by Adastra Airways during 1967‑72 from 3000 m above ground to avoid turbulence. The radar sampled an area about 50 m in diameter which was considered satisfactory at that time in desert areas of low relief. A similar amount about 205 000 km of laser terrain profiling was flown by Adastra during 1971‑75, also from 3000 m. This sampled a much smaller area, about 1‑2 metres of the ground.

 

The reductions and adjustments were done by Natmap personnel and it was remarked that 'enough profiling tapes had been processed to encircled the earth 5 times'! (Wise 1979)

 

                     

 

 

 

 

STEREO‑PLOTT1ING MAP DETAIL

 

In the 1950‑60s the stereo‑plotting of map detail was carried out by photo‑interpretation of map detail from a stereo‑model of overlapping aerial photos and transferring selected detail to the map compilation sheet. Initially, Natmap used Wild A6 and Zeiss Stereotope instruments for stereo‑plotting map detail. However, the A6 was designed to plot from wide angle 150 min / 6 inch photography, not the superwide angle 88 mm / 3.5 inch focal length aerial photos, and so the A6 was used for training and plotting planimetry only. The Stereotopes were used mostly for plotting detail for base mapping for resources surveys in Papua and New Guinea. (Fig. 15) The bulk of stereo‑plotting of planimetry and contours from the super‑wide angle 230 x 230 / 88 mm RC9 and later RC10 photography for the 1: 100 000 mapping program was done using Kem PG2 and Wild B8 analogue stereo‑plotting instruments.

 

The aerial photo diapositives were oriented on the left and right carders, viewed stereoscopically and the stereo‑model formed (relative orientation). The stereo‑model was levelled to the vertical control and scaled to the horizontal control (exterior orientation). A pantograph was used for changing from model scale to the plotting scale of the map compilation sheet.

 

Selected planimetric detail was plotted by placing the index floating dot mark in the stereo‑model on to the detail, and plotting that detail by lowering the ball point coloured ink pen on to the compilation sheet. The colours used were: blue for watercourses, lakes and other water features, red for roads, tracks and so on, and black for railways, buildings etc. Contours at 20 metres plus the odd numbered 50 metres required for the 50 metre interval on the 1:250 000 series were plotted in brown on an overlay registered to the map compilation sheet. The index mark was set at the required contour level and kept in contact with the stereo‑model and that particular contour plotted on the overlay, the index mark then fixed at the next contour level, plotted and so on. Spot heights were used to assist with the interpretation of the terrain. (Hocking c.1967)

 

Vegetation was plotted in green on another overlay registered to the map compilation sheet. (Fig. 16, A to G) Separate overlays were used to avoid having too much map detail on one sheet. Sheets of 'Letraset' cultural symbols were prepared and used during the map compilation.

 

 

CONTRACTORS

 

Up to 20 private sector firms were involved at various times, under contract on such operations as aerial photography, point marking, template making and stereo plotting of map detail. These firms included: Adastra , Geosurveys, AAM Surveys, Watsons, Associated Surveys, Alpha Aerial Surveys, Aerial Surveys, Civil Aerial Surveys, Photomappers, Photec, Qasco, Aerometrex, P. Livings, Pike & Partners, Southern Aerial Surveys, GH&D, GeoSpectrum, and others.




The contract photogrammetry was considered to be a successful operation for both Natmap and the contractors. Putting work out to contract forced Natmap staff to prepare unambiguous specifications of the work to be done and this proved to be a most demanding yet useful exercise in sorting out various procedures for compiling maps. It is well known that it is difficult enough for separate sections working in the same building to follow the same procedures, let alone different firms scattered all round Australia.

 

The contractors' staff learnt to compile medium scale maps from aerial photographs and, in addition, contract mapping provided a base load of work for the firms involved. Further more, and most importantly, these mapping contracts increased the pool of skilled people available to do map compilation work in the event of a national emergency. It is worth noting that many of these firms continue to operate in the private sector.

 

 

ACKNOWLEDGEMENTS

 

1 wish to thank the people who helped me remember. However, memory fades and any errors are mine. Clive Freegard worked wonders with the old photos. Keith Barrie, Kevin Crane and Alan Thomson kindly read the paper and suggested improvements. My wife, Iris, an ex­Natmapper, sorted out the diagrams and so forth. Finally, sincere thanks are due to all the people who worked in the National Mapping Melbourne and Canberra offices during those years, helping to put Australia on the map.

 

 

REFERENCES

 

Ford, R.A. (1979) 'The Division of National Mapping's Part in the Geodetic Survey of Australia', The Australian Surveyor, 1979, vol 29, nos. 6,7,& 8.

Gamble, S.G. (1950) 'A Suggested Improvement in the Slotted Template Method of Controlling Vertical Air Photographs', The Canadian Surveyor, vol x, no. 3.

Hocking, D. R. (c. 1967) 'Notes on Automatic Photo­interpretation for Medium Scale Mapping' unpublished.

Hocking, D. R. (1985) 'Star Tracking for Mapping ‑ An Account of Astrofix Surveys by the Division of National Mapping during 1948 ‑ 1952' paper presented at the 27th Australian Survey Congress, Alice Springs.

Lambert B. P. (1963) 'Mapping Under Pressure' paper distributed at the United Nations Conference on Science and Technology, held at Geneva.

Lines J. D. (1992) Australia on Paper ‑ The Story of Australian Mapping, Fortune Publications, Box Hill.

McMaster C. G. (1980) 'Division of National Mapping Aerodist Program' Technical Report. No 27, Department of National Development and Energy, Canberra.

Map Catalogue (1955) First Edition, Compiled by the National Mapping Office, Department of the Interior, Canberra.

Salt, LS.A. Lieut. R.E. (1933) A Simple Method of Surveying from Air Photographs, Professional Papers of the Air Survey Committee No. 8, H.M. Stationery Office, London.

Vincent D. (1982) Mosquito Monograph, David Vincent, Adelaide.

Wise P. J. (1979) 'Laser Terrain Profiling' Division of National Mapping Technical Report No. 26, Department of National Development and Energy, Canberra.