The Laser Years :
Recollecting Airborne Terrain Profiling for Photogrammetric Vertical Control
in the Division of National Mapping, 1962-1982
Compiled by Paul Wise, 2017-19
About this paper
This paper provides the most detailed account of the Division of National Mapping's airborne terrain profiling program that operated from 1962-1982. This program generated profiles of the terrain from which the location and height of selected points were extracted. These selected points were then used to control the aerial photography from which Natmap produced contoured 1: 100,000 scale topographic compilations.
While published papers about the profiling program are available they only reflected points in time whereas the story was a continuum. The continuum comes from the, possibly fallible, human recollection and this should be borne in mind by readers.
This paper was written with the valuable assistance of Natmapper and laser profiling party leader Rod Menzies. Input from Dave Abreu, Harry Baker, Ed Burke, Simon Cowling, Laurie Edebohls, John Ely, Oz Ertok, Andrew Hatfield, Graeme Lawrence, Michael Lloyd, Mike Morgan, Carl McMaster, Bill Stuchbery, Rom Vassil, and the support of Con Veenstra, John Manning, Syd Kirkby and Andrew Turk, is acknowleged. I am extremely grateful to Laurie McLean whose paper Recollections of the Aerodist Years provided a basic template and the impetus for this work.
Chapter 1 - Photogrammetric based Mapping for Australia
The majority of vertical control or points of known height, that the Commonwealth of Australia's Division of National Mapping required to produce contoured 1: 100,000 scale topographic national mapping, was obtained by airborne terrain profiling. Airborne terrain profiles were acquired over more than 70 per cent of the Australian mainland. Other methods were used to obtain vertical control over another 7 per cent of Australia, with the remaining area already having sufficient vertical control for generating contours at 1: 100,000 scale.
Airborne terrain profiling for vertical map control was, at the time, the most efficient method. More importantly it was necessitated by historical decisions made in the context of how Australia’s medium scale, national topographic mapping programs would be achieved. In his 1971 paper, Nat Map’s then Assistant Director Joe Lines posed the question why terrain profile recording? His answer was that the reasons lie fundamentally in the basic compilation technique originally adopted after an extensive testing programme some time ago. It was thus the decision to adopt the Arundel Method of map compilation that in turn led to the need for vertical control. At the time airborne terrain profiling was the only means of the timely and economical acquisition of such a volume of control for continental Australia.
Following World War 1 the British War Office had recognised the significance of aerial photography and the need to apply it to best advantage. In the late 1920s a captain in the Royal Engineers, and a brilliant mathematician Martin Hotine, later Brigadier Martin Hotine CMG CBE RE (1898–1968), was attached to the Air Survey Committee of the Geographical Section of the General Staff (GSGS) at the War Office. Hotine was one of a group of established mathematicians working on some of the problems associated with the simple graphical methods of surveying topographic information from vertical air photography. Thus was born the system of block photographic coverage by straight parallel strips, in some literature also called a run, in which each photograph in a strip overlapped its predecessor by at least 60%. Each subsequent parallel strip then also overlapped its predecessor by up to 25%. This forward and side overlap allowed the relief and aircraft distortions of a block of aerial photography to be rectified to established ground control by graphical means. The system of block, near vertical, aerial photographic acquisition and adjustment became known as the Arundel Method after the area in England where the idea was developed.
After receiving training from the Royal Air Force in 1926, Royal Australian Air Force (RAAF) Flight Lieutenant G Packer returned to Australia with these latest concepts in aerial photography acquisition and application. Packer subsequently became the RAAF’s chief adviser on aerial survey matters including the navigation and flying requirements fundamental to utilising the Arundel Method. Subsequently in 1931 the Australian Survey Corps produced the Albury 1 inch to 1 mile map sheet which was compiled with the significant aid of RAAF acquired aerial photography. When produced in 1936 however, it was the Sale 1 inch to 1 mile map sheet that was the first to be fully compiled from the Arundel Method of aerial photography acquisition and adjustment. Consequently, when the mid 1955 Recommended Plan of Mapping Operations was approved for a 1: 250,000 scale national topographic map series of Australia, the Arundel Method would be the basis for the series compilation.
The overwhelming problem that could not be solved for what became known as the R502 map series was how to supply all the vertical control that would be needed to enable the plotting of contours. At least one point of known height would have been required at a grid spacing of less than 10 kilometres or somewhere around 100,000 height points strategically located in relation to each aerial photograph. As there was no practical way at the time of acquiring these data, the R502 map series was produced with most map sheets devoid of contours and only having scattered spot heights. The second national Australian map coverage the National Topographic Map Series (NTMS) was to be contoured. As will be seen, by then enough work had been done to determine that airborne terrain profiling was the solution to obtaining the vertical control needed for the contour generation for the NTMS series.
Initially, vertical control for contouring the NTMS came from contract airborne profiles acquired using the Canadian Applied Research Limited (CARL) Mark V, Airborne Profile Recorder (APR). The APR system owned and operated by Adastra Airways of Sydney was contracted by Nat Map between 1962 and 1973 to acquire the necessary terrain profiles. Later, the Australian developed, Department of Supply, Weapons Research Establishment, Laser and Optics Section, Laser Terrain Profiler (LTP) Mark 1, was operated by National Mapping personnel from 1970 to 1979, to acquire airborne terrain profiles to complete the program of vertical control acquisition for the NTMS. The APR system used microwaves or radar as the basis of its height measuring system, whereas the LTP used a continuous wave laser. Thus, radar profiling as opposed to laser profiling can be found in the literature when referring to the use of APR and LTP. For clarity, WRE also developed a Mark 11, Laser Terrain Profiler for the Royal Australian Survey Corps. Soon after, Fairey Australasia Pty Ltd became the licensed manufacturer for future such equipment under the commercial name of WREMAPS. Nat Map’s laser profiler was then identified as WREMAPS1 and the Survey Corps’ as WREMAPS2. Nevertheless, for all its operational life Nat Map’s system was simply known as the laser terrain profiler (LTP) and that name will be used throughout this paper. Please refer to Figure 1 below.
Figure 1 : Weapons Research Establishment designed decal rebranding their Laser Terrain Profiler, Mark 1, as WREMAPS1.
Like National Mapping’s Aerodist horizontal mapping control program, the airborne terrain profiling program was then thought to be a once only element in the national effort to have a topographic mapping coverage to meet Australia's then present and future requirements. Thus, Nat Map’s airborne terrain profiling program forms a major part of Australia's national mapping history. The first contract radar profile was flown in late 1962 in the Mount Coolon–Clermont area of Queensland and the last, in 1973 in South Australia - New South Wales. Nat Map flew its first laser profile in 1970 outside Broken Hill, New South Wales and its last south of Giles, Western Australia in 1979.
Airborne terrain profiling field parties usually comprised only a handful of people; a core group including pilot, to fly and operate the airborne equipment plus ground support when and where considered necessary. Apart from a suitable airstrip with aircraft fuel, only good flying conditions were required for airborne profiling operations. However, to extract the heights for photogrammetric model vertical control from both the contract radar and divisional laser terrain profiles an office component existed within Nat Map which also provided Nat Map field party support, planning and preparation, and contract management. Staff from this group were often rotated with the Nat Map laser airborne terrain profiling field staff.
Up to the end of 1976, Nat Map’s airborne terrain profiling operations were based at its Melbourne office in the Rialto Building at 497 Collins Street, Melbourne. From 1977 until 1982 when the last heights were extracted from the laser terrain profiles, airborne terrain profiling operations were based at its premises in Ellery House at 240 Thomas Street, Dandenong. As part of Nat Map’s 1977 change of office location, the radar terrain profiling material was sent to Commonwealth Archives as by this time all necessary height data had been extracted.
For the most part airborne terrain profiling operations were a function of the Topographic Surveys Branch of the Topographic Office. John Dunstan (Joe) Lines (Nat Map 1948-1976) was Assistant Director, Topographic Office to 1976, after which Sydney Lorrimar (Syd) Kirkby (Nat Map 1959-1984) took over the role. Orest Jacovlavich (Bob) Bobroff (Nat Map 1958-1982) initially headed the then Control Surveys Branch as supervising surveyor and later the Topographic Surveys Branch. From 1970, Rom Vassil (Nat Map 1965–1984) was the senior surveyor directly responsible for airborne terrain profiling operations. During 1967 and 1968 Rom had been engaged in vertical control field survey work with the Johnson Ground Elevation Meter. Rom replaced Edmond Francis Norman (Ted) Seton (Nat Map 1957-1970) on Ted’s decision to retire and return to his native Queensland.
Whilst having moved on in National Mapping by the time the Laser Terrain Profiler became operational the role of Leonard George (Len) Turner (1932-2002) should not be overlooked. As Supervising Surveyor (Topographic, 1966-1970), Len had carriage of horizontal control surveys and for vertical control surveys as well as other topographic mapping functions. Under Len, not only was the Aerodist system introduced (McLean, 2015), but contract airborne terrain profiling was evaluated and commenced. Further, Len instigated and managed Nat Map’s alliance with the Department of Supply’s, Weapons Research Establishment in developing the WREMAPS1 Laser Terrain Profiler. Two other Natmappers, Adrian John Wright (1944-2008) and Roderick William (Rod) Menzies (Nat Map 1976-1987) made significant contributions to Nat Map’s airborne laser terrain profiling operations; Adrian from 1971 to 1982 and Rod from 1976 to 1981. Electronics’ technicians John Ely (Nat Map 1966-1997), Ozcan Ertok (Nat Map 1971-1991), and Mick Skinner (Nat Map 1966-1973) were largely responsible for the operation and maintenance of Nat Map’s laser profiling system throughout its life. Their indepth knowledge of the various systems allowed minor failures to be rectified quickly and avoid significant downtime. The success of Nat Map’s airborne laser terrain profiling program is primarily due to their expertise.
After the 1977 internal reorganisation of the Melbourne based Topographic Office, Andrew Glen Turk (Nat Map 1969-1978) replaced Rom Vassil and with the later departure of both Bobroff and Turk, John Manning (Nat Map 1966-2004) as supervising surveyor and Paul Wise (Nat Map 1969-1987, airborne LTP 1972-1982) as senior surveyor oversaw the final years of the airborne laser terrain profiling program in National Mapping. A small lunchtime ceremony on 9 September 1982 marked the occasion of the completion of Nat Map’s airborne LTP program. Subsequently, with the move to digital mapping and stereodigitising data capture required to meet 1: 50,000 scale map accuracy standards, a second generation airborne terrain profiling system was then developed and this system is discussed later in this paper.
Photogrammetric mapping can be thought of as exploiting photographs, film or digital, as the primary data source for the provision of map detail. Although both aerial photographs and maps show an overview of the earth's surface, an aerial photograph is not a map (Crum, 1995). An aerial photograph is a perspective projection of the scene photographed. The position of detail in an aerial photograph is erroneous due to radial distortions caused by the optics of the camera, the relative instability of the aerial camera platform, and height differences in the terrain. For practical purposes, maps are directionally and geometrically accurate, given they show an orthogonal projection of a three dimensional world onto a two dimensional sheet of paper.
Through the principles of photogrammetry, the errors in near vertical aerial photography can be removed. The removal of these radial errors then permits the detail extracted from aerial photography to be accurately plotted, within the limits of the map scale, to make a map. Accurate extraction of the map information from aerial photography required that the aerial photography be controlled both horizontally and vertically. This horizontal and vertical control was provided by having points, in the terrain as well as identified in the aerial photography, with coordinates and/or height from purposely obtained survey data. For the most part, the acquisition of the aerial photography and its control was undertaken by separate programs dictated by the environment, technology and resources available at the time.
From time to time, various organisations including National Mapping, produced Orthophotomaps. Orthophotomaps were a mosaic of orthophotographs which were aerial photographs from which all the terrain, earth and camera related distortions had been removed and therefore all the features in the photograph were now truly positioned and to scale. For various reasons Orthophotomaps failed to meet the expectations of the producer and the user but can still be found in map libraries. The traditional paper line map has still to be replaced completely, despite technological advances.
After the Second World War, Australia embarked on its first national topographic map coverage at a scale of 1: 250,000. This coverage was known as the R502 series. While the R502 series derived its information entirely from aerial photography it was controlled horizontally by a mix of traditional survey, where available, and astronomic observations for position or astrofixes elsewhere. Only a few sheets were contoured (23%), as already mentioned, with spot heights derived from land and airborne barometric surveys shown on the remainder of the map sheets. Terrain was depicted by the method of hachuring or shading. The R502 mapping program had been underway for some ten years when in 1965 the Australian government decided to now accelerate the mapping program. Compilation of each map sheet (30 minutes of latitude by 30 minutes of longitude) would be at 1: 100,000 scale, with publication at both 1: 250,000 and 1: 100,000 scales. Importantly for many users both scales would now be contoured; at 20 metres vertical interval on the 1: 100,000 scale map sheets and 50 metres vertical interval on the 1: 250,000 scale maps. The production of contours for the whole of Australia within the planned timeframe however, would mean that traditional heighting techniques for providing the required vertical control would be inadequate.
From the outset, resources allocated to this new mapping program, the National Topographic Map Series or NTMS, were never at the level required for the envisaged map publication schedule to be met. A more considered program was thus adopted being that all map sheets would still be compiled at 1: 100,000 scale but in the sparsely settled inland areas publication would only be at 1: 250,000 scale. Elsewhere, publication scale remained at 1: 100,000 scale. Figure 2 below shows a copy of the map with the thick red line the boundary between the two publication programs. (Only 1: 250,000 scale map publication would occur inside the red line on the diagram.) This map printed in red as shown in Figure 2, with its thick red line of demarcation became commonly known as the Red Line diagram. Notwithstanding the less than initially anticipated resource allocation, the NTMS program was hardly an inexpensive undertaking. The overall cost of the 1: 100,000 scale and 1: 250,000 scale maps produced under the program has been stated as A$600 million (O’Donnell, 2006).
The National Topographic Map Series at 1: 100,000 scale comprised some 3,062 map sheets which were completed around 1988. Some 1,460 map sheets covering the more remote inland of Australia were only produced to the compilation stage (O’Donnell, 2006). All 1: 100,000 scale compilations were used to derive new 1: 250,000 scale maps, gradually replacing the earlier R502 map series of the same scale. The 1: 250,000 scale NTMS program of 544 map sheets was completed in 1991.
Figure 2 : Map index, commonly known as the Red Line diagram, showing the areas of responsibility for survey, compilation and map production as agreed by the National Mapping Council, in 1974.
This map further indicates that only 1: 250,000 scale publication would occur inside the thick red line on the diagram.
Aerial photography was again the primary data source for the NTMS program. The then technologically advanced Wild RC9 aerial survey camera replaced the K17 camera, or its equivalent. In the late 1950s, Wild Switzerland introduced the RC9 aerial survey camera with its superwide angle (nominally 88 millimetre focal length) lens and 230mm film format. When flown at 25,000 feet above sea level (ASL), the RC9 camera produced photographs with a nominal scale of 1: 80,000. The number of 1: 80,000 scale photogrammetric models covering a 1: 250,000 scale map area was approximately 70% less than the number of 1: 50,000 scale models from the K17 camera. (A photogrammetric or stereoscopic model generally consists of two aerial photographs that overlap by 60 per cent and provides a three dimensional view of the terrain). The reduced number of photogrammetric models to be controlled and used to plot the topography yielded flow on economies for other components of the NTMS program. Thus, the estimated overall cost effectiveness of 1: 80,000 scale photography was greater than that indicated by the reduction in the number of stereoscopic models alone (Lines, 1992). Given the magnitude of the NTMS program, such economies could not be ignored. Figure 3 below is a diagrammatic representation of aerial photography acquisition with 60% forward overlap and 25% side overlap. In Figure 4 below is an example of a photogrammetric or stereoscopic model consisting of two aerial photographs that overlap by 60 per cent. The coloured circles in the four corners show the areas where horizontal and vertical model control was optimally required.
Figure 3 : Diagram showing aerial photography acquisition with 60% forward overlap and 25% side overlap.
The coloured circles in the four corners show the areas for optimal horizontal and vertical control.
A program thus commenced to acquire systematic, 1: 80,000 scale, monochrome, near vertical photographs of 1: 250,000 scale map sheet areas. Each such sheet area was covered by 8 or 9 parallel flight strips or runs from 25,000 feet (ASL) with the photography having a minimum of 80% forward overlap and 25% side overlap. As this aerial photography was to be acquired by private contractors, its provision was controlled by the National Mapping Council’s, Standard Specifications for Vertical Aerial Photography. Given the cost of logistics versus film and paper, the 80% forward overlap specification allowed a safety margin for meeting the minimum 60% forward overlap photogrammetric requirement. Between 1960 and 1975, all but the logistically difficult areas of Australia were captured on film as shown in Figure 5 below.
Figure 5 : Map showing areas not covered by standard mapping photography after 1975.
Although this aerial photography program reduced the number of photogrammetric models overall to be controlled and used to plot the topography and contours, each such photogrammetric model still required a minimum of four control points, located as indicated in Figure 4 above. These points, essentially three points plus one check point, with known position and height in the terrain would be sufficient to scale and orient each photogrammetric model. No technology then could effectively produce both position and height for the vast numbers of points in the terrain that would be required, so the processes for providing such control were separated. Horizontal and vertical control were acquired by separate programs and bought together at the photogrammetric plotting stage.
Figure 6 below is a diagrammatic representation of seven photogrammetric models in an aerial photography flight strip showing the indicative locations of both horizontal and vertical control points within the strip. As can be seen, separate to the control points are the respective photo centres, also called pass points, which appear on consecutive images within a single flight strip, and the pass, tie or wing points located near the model’s corners so as to be common to the applicable adjacent flight strip. The photo centre pass points along with the corner pass, tie or wing points enabled the individual photogrammetric models to be related to one another both along the flight strip and across flight strips. As can be seen more clearly in Figure 6, all the pass, tie or wing points are separate from the required horizontal and vertical (not shown) control points.
Figure 6 : Diagrammatic representation of 7 photogrammetric models in an aerial photography flight strip.
In the early 1960s, test work was carried out using a block of RC9 superwide angle aerial photography at a scale of about 1:40,000 in the vicinity of Canberra. This photogrammetric block lent itself to being controlled with varying densities of ground control, with superfluous control not used in each test to be used as an accuracy check on their positions as derived from each test. Extrapolating from this work, it was concluded that second order control spaced in the corners of geographic squares with sides of one degree of latitude and longitude would adequately provide the horizontal component for 1: 100,000 photogrammetric control. Each of these geographic squares would contain 4 x 1: 100,000 scale map sheets. Any additional control from the geodetic loops would be a bonus (Lines, 1992). Even so, on the vast, relatively flat regions of Australia establishing and surveying this required second order control was a daunting task. Aerodist or Airborne Distance measuring was the solution selected by National Mapping and became the technology behind the intensification of horizontal control for photogrammetric mapping.
Aerodist in National Mapping has been extensively covered by McLean (2015), so only a summary is provided here. In essence, the Aerodist airborne line measuring program and the intense sub program of Ground Marking for Aerodist resulted in a grid of Aerodist control points, or Aerodist stations, having coordinates to second order standard. Please refer to Figures 7 and 8 below. The horizontal adjustment of the Aerodist measured lines from each of the Aerodist stations to those Aerodist stations nearby and to the applicable existing first order geodetic control was done in blocks as shown in Figure 9 below. These Aerodist block adjustments (BA’s) were separate entities from the soon to be discussed photogrammetric blocks.
Figure 7 : The Australian First Order Geodetic loops (green triangles) and lower order networks (shaded grey) with mainland Aerodist station grid and offshore Aerodist stations for second order photogrammetric control.
Figure 8 : The Nat Map Aerodist lines measured (purple lines) to enable the positions of the established Aerodist stations to be calculated.
Figure 9 : Map showing the Aerodist block adjustments and their identifier.
Demand, either real or perceived, was the driver behind the production and publication of the new NTMS. This explains why both Aerodist and airborne terrain profiling operations appear to randomly take place around Australia from year to year in no particular sequence. As explained above, the adoption of the Arundel Method by Nat Map meant that from its earliest years Nat Map had used graphical radial triangulation or commonly slotted templates to mechanically intensify the mainly perimeter horizontal control to the photogrammetric model level, within a photogrammetric block, as shown in Figure 10 below. Thus, a skill base had been developed within the organisation, allowing a mass production approach. Several assemblies or graphical photogrammetric block adjustments could be undertaken simultaneously using mainly radial templates which, with the inclusion of nadir templates, stereo templates and azimuth templates as necessary, met 1: 100,000 scale mapping accuracy requirements. Figure 12 : National Mapping’s Photogrammetric Blocks showing the blocks adjusted using the Slotted Template process (light hatches) and the blocks adjusted analytically (dark hatches). More detail about aerial photography acquisition, photogrammetric blocks, slotted template assemblies and stereoplotting may be found at Annexures A and B.
Figure 10 : Example of graphical radial triangulation or slotted template assembly.
The fixed perimeter control of the block is mechanically converted via the individual templates, to absolutely locate each photocenter and pass point at the final map scale.
A photogrammetric block comprised a number of adjacent 1: 250,000 scale map sheets. There was no specific size, rather the block’s extent was determined by the location of suitable horizontal perimeter control. Such control, Geodetic and/or Aerodist, had to be dense enough to prevent any outwards deformation of the block during adjustment. The photogrammetric blocks processed by National Mapping are shown in Figure 12 below. Later, when control intensification was undertaken using computer aided techniques the same use of perimeter control also enabled any large residuals to be highlighted during the analytical block adjustment. Figures 13 and 14 below show slotted template assemblies laid out.
Figure 11 : The Australian First Order Geodetic (green triangles) and lower order networks (shaded grey) with the Nat Map Aerodist Horizontal Control network (brown points and purple lines) overlaid with the Photogrammetric Block boundaries and their numeric block identifier.
Figure 12 : National Mapping’s Photogrammetric Blocks showing the blocks adjusted using the Slotted Template process (light hatches) and the blocks adjusted analytically (dark hatches).
Figure 13 : Circa early 1970s photograph of a slotted template assembly with (L-R) Bob Foster, Ian Pasco, Len Bentley and Brian Martinesz, in the Rialto building in Melbourne.
Figure 14 : 1979 photograph of National Mapping’s last slotted template assembly with (left) Bob Foster and (right) Dave Hocking at Nat Map’s building in Dandenong.
Vertical control was directly acquired in relation to the photogrammetric models comprising the block. With the aerial photography overlapping along and across each parallel flight strip, vertical control was optimised by having the heights of points in the four corners of the photogrammetric model as shown in Figure 4 above. A single height point could then be used as part of the control for up to four photogrammetric models (two along the flight strip and two on the adjacent flight strip). Vertical control for analytical block adjustments was acquired differently as is discussed below. Photogrammetric blocks thus became useful units for both contract and inhouse airborne terrain profiling operations, with the advantage that the data was collected at about the same time and with likely similar atmospheric conditions.
Some 52 photogrammetric block adjustments were undertaken by Nat Map. Analytical or numerical photogrammetric techniques were used to simultaneously intensify both the horizontal and vertical control in 17 of those blocks. Please refer to Figure 12 above.
National Mapping’s numerical photogrammetric MODel BLOCK (MODBLOCK) adjustment program along with FORMIT and MODSTRIP were developed by Dr CWB King from April 1972. Photogrammetric model joins within each strip of adjoining models were tested by the strip formation program FORMIT. Joins of strips of photogrammetric models to ground control, where applicable, were tested by the strip adjustment program MODSTRIP. The final join of strips to each other and to ground control was performed by the block adjustment program MODBLOCK.
King had originally developed MODBLOCK whilst Head of Photogrammetry and Mapping with the Iranian Oil Operating Companies in Teheran, Iran in the mid 1960s. MODBLOCK written in the FORTRAN computer language was able to compute a block adjustment of any size within the working space of eight method of least squares adjustment equations (King, 1967). The only additional storage requirement, over and above that of a single strip, was that for the solutions of each strip, but these were column vectors and not arrays. If there was any shortage of computer storage space, these data could be conveniently kept on a tape drive. At that time, King’s MODBLOCK ran on an IBM 7040, and utilised five tape drives for storage. While initially MODBLOCK ran on an IBM 1620, small storage, variable word length, decimal computer that was particularly suited to dealing with technical problems, MODBLOCK was subsequently expanded and improved to run on a UNIVAC 1108, and in Natmap ran on a Control Data Corporation (CDC) CYBER 76. With the introduction of mini and micro mainframe computers for Natmap’s digital mapping program, MODBLOCK was eventually able to be run inhouse.
While other numerical photogrammetric block adjustment programs existed their cost and adaptation were prohibitive to National Mapping. It was also considered that adjustment programs using the solution of polynomials were not sufficiently rigorous for Nat Map’s purposes. Such polynomial based programs assumed that the distortion in a strip was regular, which was not so. This disadvantage was of little import in a single strip where there was insufficient information to discover the true distortions. In a block adjustment, the many common points in the sidelaps of parallel strips gave a great deal of information as to the progress of distortion, which information was not fully utilised in a polynomial adjustment imposing regularity
The answer was a rigorous adjustment where all photogrammetric models, each with seven degrees of freedom, were adjusted simultaneously; or all photographs, each with six degrees of freedom, were adjusted simultaneously as in a bundle adjustment from comparator observations. This procedure involved the solution of many thousands of equations and would exceed the capacity of even large computers.
MODBLOCK was based on the seven unknowns of each independently observed model being solved simultaneously for all models of a block in a full least squares solution. The constraints were that all transformed values of observations of ground control points should as nearly as possible fit their true ground values, and that all transformed values of observed points in a model must as nearly as possible fit their transformed values in every adjacent model (if any) in which they also fall. This meant that each photogrammetric model was made to fit all its neighbours at once, irrespective of the way strips had been flown. This approach was clearly better than the polynomial method of forming up strips as a first step, without considering overlapping strips, and then holding the model connections fixed in the subsequent adjustment.
The number of equations to be formed and solved was still significant, but the process was broken down into stages that did not affect the rigour of the solution.
As the mapping photography to be used in the block adjustments was all flown east-west to cover a map sheet at 1: 250,000 scale and additionally had three north-south tie strips flown across each map area, the simultaneous adjustment of this pattern of photography permitted the amount of height control to be reduced. This fact impacted the operational acquisition of terrain profiling.
A feature that made the MODBLOCK program easy to use was that sorting routines had been incorporated. The user had only to select the two principal points and the two projection centres as the first four points of every model. Then any control, check or connection point between models of a strip, a tie point between strips or a single point was automatically discriminated. The final transformed coordinates were set out model by model for ease of plotting, with all common points correctly meaned over all the models in which they appeared. Within each model, points were sorted in order of ascending point name. Subsidiary programs plotted the vectors of residuals at control points and/or check points for visual interpretation of the validity of the adjustment and also plotted all transformed coordinates on base sheets for map compilation.
As mentioned above, the use of computer based adjustments meant that the laser profiling scheme could be reduced and was thus adjusted to a more evenly spaced grid. National Mapping’s approach was to establish horizontal and vertical control at the corners of units of about 120 models, which usually constituted areas about 100 square kilometres. Several of these units were enclosed within perimeter control for the block. Perimeter control was usually the 1° or 30¢ network of trig points with vertical control provided at a spacing of 4 to 6 models (about 60 kilometres) along all flight strips.
In areas which were to be adjusted by numerical photogrammetric techniques the terrain profiling pattern was therefore planned so as to acquire profiles along the block perimeter and then approximately equidistant, not exceeding 60 kilometre intervals, between those limits. This resulted in an almost square grid of profiles over the block. As there was more flexibility in where the profiles could be obtained, where possible all profiles were planned to overfly as much existing vertical ground control as possible. More detail may be found at Annexure C.
For the most part however, Nat Map’s airborne laser terrain profiling operations were undertaken as described in Part 2 below.
The difference in height between two points was traditionally found by angle and distance measurement from the first point to the second point and simple trigonometry. The Indian Pattern Clinometer or the Tangent Clinometer as shown in Figure 15, measured the angle of elevation or depression and also gave its tangent. As the height difference between the point of observation and that being observed was distance to the observed point by the tangent of the angle read by the clinometer, having the tangent easily readable as well as the angle meant no tables needed be carried. The two vanes folded down for portability.
Figure 15 : Indian Pattern Clinometer or the Tangent Clinometer showing the eye vane right and object vane left, engraved with the angle and its tangent.
This method developed into a technique called Simultaneous Reciprocal Vertical Angles. The technique relied on the vertical angles being read simultaneously at each point to the other thus removing any atmospheric effects. This technique was largely used on geodetic surveys to carry height from point to point using theodolites to accurately obtain the angles at both end points to the other.
Nat Map specified that Simultaneous Reciprocal Vertical Angles be observed when the air was most evenly heated. This was not before about 2 hours after the sun had crossed the meridian in the area, and not later than about 4 to 4½ hours after the sun had crossed the meridian. In practice, between 2 and 3PM was a good time in the days observing routine. As shown in Figure 16 below, the vertical angles observed during a geodetic survey were nearly always angles of depression.
Figure 16 : Showing the level planes at right angles to the plumb lines at two points on the earth's surface.
Note that the line of sight is below this plane at both points and thus the vertical angles observed were nearly always angles of depression.
For field computations the following formula was adopted to compute the height difference between stations. It was based on two assumptions; one, that for the small vertical angles normally involved, the mathematical Sine and Tan functions of the angles are the same, and two, the slope distance and horizontal distance between the two points were almost the same, so no inaccuracy is introduced by using the Tan function and the slope distance between stations as measured by electronic instruments. The formula was, for two stations A and B and taking the signs of the observed vertical angles into account :
Difference in Height = Slope Distance A-B x Tan½ (Angle at A - Angle at B)
If there was a considerable difference in height between the two stations the problem was resolved by initially using the above formula to calculate the approximate height of the forward station B. Using the now known heights of both stations to convert the slope distance to the horizontal distance, and then recomputing the heights with this accurate horizontal distance and the Sine function instead of the Tan function of the vertical angles. It was important that distances were not reduced to sea level for these height computations.
In determining heights from Simultaneous Reciprocal Vertical Angles one final correction was required. This correction was required because the angles were observed by two separate instruments to two different targets all set at different heights in relation to the height of the respective station marks as shown in Figure 17 below.
Instrument and Target Correction = (Instrument height at A - Instrument height at B + Target height at A - Target height at B) / 2
Figure 17 : Showing the relative heights of instruments and targets at two separate stations.
As instrumentation developed, spirit levels evolved where a bubble was used to indicate a horizontal plane about a point. By reading where this plane intersected a graduated staff held vertically over another point, the height difference between the two points could be determined. The evolution of this technique was called spirit levelling or just levelling. The introduction of automatic levels greatly improved the speed and accuracy of a levelling survey. The use of automatic levels reduced the time taken and human inaccuracy in the centring of the level’s bubble i.e. setting the horizontal plane. The automatic level had a prism which swung freely and thus under the force of gravity set the horizontal plane. A damper was incorporated to steady the line of sight and the whole unit of prism and damper was called the optical stabiliser. Once the instrument was set approximately horizontal using its small circular (pill) bubble, the optical stabiliser could then itself swing automatically into a position that set the true horizontal line of sight. A diagram showing this optical/mechanical solution is at Figure 18 below. Even with this development acquiring the vertical control needed for photogrammetric mapping was far too slow. Nevertheless, like the loops of the geodetic survey which provided a national set of fixed, homogenous positions on a single national datum, loops of third order levelling provided a similar set of fixed, homogenous height points on the Australian Height Datum (AHD). Again, like their horizontal counterpart, these loops of heights were too sparse to be used as photogrammetric control but were the basis for controlling the vertical control or height intensification programs as will be discussed.
Figure 18 : Diagram showing an example of the optical mechanism automatically correcting the line of sight when the level’s telescope is tilted.
Rather than the instrument man and staff men walking between their respective observing and setup locations motorised vehicles were sometimes used. The most sophisticated version used specially modified vehicles to mount and transport the levelling equipment and operators. These vehicles allowed for the rapid setup and observation without leaving the vehicles. In 1983 the United States reported on their testing of this equipment in their paper A Test of the Swedish Motorized Levelling System presented at the 10th United Nations Regional Cartographic Conference Asia and the Far East.
Vertical Control from Barometric Heighting
Barometers had been seen as an alternative to levelling. Provided the atmospherics were favourable, barometers could be carried from point to point very easily and from their readings height differences obtained. After World War Two, surplus aircraft altimeters increased the availability of pressure sensitive instruments and the aircraft altimeter had the advantage of having an increased vertical range as compared to the barometer. The word barometer will be used hereonin to refer to both types of instrument. Three barometers were usually mounted in a box for ease of transportation and so as to be able to identify a failure in any instrument. With one set of three barometers at a base station, usually of already known height, a roving set or sets of barometers would be transported to points where heights were required, as shown in Figure 19 below. Later comparison of the readings would allow heights of the points visited to be calculated.
Barometric heighting was used during the astrofix acquisition program for the R502 mapping program. Barometric observations were taken by astrofix parties at creek crossings, homesteads and airstrips to provide some indication of the extent of the relief in an area. With the availability of light aircraft and helicopters the barometric heighting technique evolved further.
Figure 19 : (Left courtesy Peter Hocking) A set of three aircraft altimeters being used for barometric heighting during astrofix operations.
In the 1950s, the Department of the Interior was requested to assist the Bureau of Mineral Resources (BMR) with survey control for the network of gravity measurements being planned across Australia. For gravity purposes, BMR were to establish, by helicopter borne barometric observation, stations on an approximate 7 mile by 7 mile grid. As height at each such station was necessary for the reduction of the gravity data, the grids of spot heights that emerged from this work became of value to topographic mapping and charting. Several hundred thousand square kilometres of sedimentary basins in Western Australia, Northern Territory and western Queensland were covered in this manner by helicopter borne survey parties (Lines, 1992).
In the reduction of the gravity observations, heights with errors in excess of 1 metre above mean sea level became significant. Height control was therefore provided by levelling traverses along roads and tracks in the area of the survey. This levelling was carried out by the Department of the Interior. Not only did this work assist the gravity surveys but also contributed to loops of levelling used for later establishing the Australian Height Datum.
The only photogrammetric blocks recorded as being controlled by barometric heighting in National Mapping, were Blocks 5A and 8 as shown in Figure 17 below. This work was done in 1968 by a field party led by Ted Seton which used a helicopter to obtain observations at the required points. For this work, and all future work needing accurate atmospheric pressure readings, Nat Map used the Aneroid barometer by Mechanism Limited of Croydon, UK, as shown in Figure 16 above.
Figure 20 : Photogrammetric Blocks which used barometric heighting techniques to provide vertical control for mapping.
The so called Mechanism Barometer had a digital readout of the pressure direct to 0.1 millibar. Reading errors were thus largely eliminated, but to avoid instrument failure or mishandling, readings from two instruments were taken at all points during a barometric survey. With these instruments, National Mapping’s approach was generally to read all base and roving barometers at the base station before departure, read all barometers at hourly intervals while the roving barometers were at their distant location(s), and then read all base and remote barometers on return of the roving barometers to the base station. The base station was generally located near to a bench mark or other point of already known height. If hourly readings were not possible at the remote location(s) then a graph of the readings at the base station was drawn to permit interpolated base station readings to be obtained for the same instant of time as the remote readings. It was normal practise to draw a graph of both base and remote station readings, as shown in Figure 21 below. If then the graphed curves were not generally parallel, unsuitable weather conditions, or errors in reading the barometers were indicated and unreliable heights could be expected. Each time the atmospheric pressure was observed ambient air temperature was also recorded.
Figure 21 : Sample of graphs for base and remote barometer readings.
Computation of final heights used formulae documented by Lt Col CA Biddle, RE, Senior Lecturer in Surveying, University College, London (Biddle, circa 1960s). Simplistically :
(height at point b – height at point a) = KTv (log(pa/pb))
where pa and pb were the pressures (millibars), at each station respectively.
At the time logarithmic (log) tables were widely used and since such tables were required for one term in the above formula, it was very convenient to complete the computation by logs. Hence :
log (hb – ha) = log (KTv) + log (log pa – log pb)
To simplify the calculation further a table of log (KTv) values was prepared for a range of values of temperature in degrees Celsius where :
log KTv = log (221.266 *(°C + 273)) for heights in feet, or
log KTv = log (67.442 *(°C + 273)) for heights in metres.
When calculating machines became more readily available, the formulae used were :
Difference in height (feet) = Difference in pressure x (Temp °C +273) x 221.266, or
Difference in height (metres) = Difference in pressure x (Temp °C +273) x 67.442.
Again to simplify the calculation further tables of (Temp °C +273) x 221.266 and (Temp °C +273) x 67.442 for a range of values of temperature in degrees Celsius, where prepared.
Toward the end of the 1950s aerial photography and photogrammetric techniques were commanding an increase in accuracy, specifically that of vertical control. Technological evolution in engineering, electronics, and materials saw the implementation of new ground and airborne data gathering.
Vertical Control by Specialised Vehicle
Deumlich (1982), outlined the then Soviet WA-1 or Automatic Altimeter, which not only computed the elevation differences when driving over the terrain but also recorded the profile of the path covered and permitted the reading of elevations and elevation differences on counters. Please refer to Figure 22 below. This circa late 1950s device used a pendulum for outputting impulses proportional to the slope angle and using a measuring wheel to determine distance travelled. The profile was recorded on a roll of photographic paper. Up to 90 kilometres could be profiled in a seven hour day and with the later WA-56 instrument an accuracy of ±4 to ±10 centimetres could be achieved with double running and slopes between ±15 degrees.
Figure 19 shows the pendulum (1) which ended in a thin wire touching a globoidal drum (2) uniformly rotated by a motor (3). The groove (4) represented a contact line. A relay (5) was operated by every contact with the pendulum wire. The distance measurement was made by determining the revolutions of the wheel (9). The result was also conducted by the reductor (8) to the integration sockets (6) and (7), their rotation being controlled by the redactor; (10) represented the armature of the socket. Elevation was measured by counter (11), the travelled road distance by counter (12). The result of the levelling was recorded on a tape of photographic paper (15). The optical system of projection was indicated in (14). The light cylinder (13) effected the transformation of the angle of rotation into a linear displacement. It consisted of an opaque cylinder, the surface of which was pierced by screw-lines at two diametrically opposite places. As a result, an automatically recorded profile was obtained.
Figure 22 : Schematic of components of WA-1M Automatic Altimeter.
In April 1961, the United States Geological Survey took delivery of two Johnson Ground Elevation Meters (JGEM, sometimes just GEM)(Speert, 1962). During his 1963 visit to the United States, Bruce Lambert Nat Map’s Director, saw a demonstration of the Johnson Ground Elevation Meter and along with the Royal Australian Survey Corps, two were ordered for Australia. Nat Map’s JGEM was delivered in August 1964, and accepted after tests at Balcombe in Victoria, in conjunction with the RA Survey’s vehicle. Figure 23 shows a photograph of Nat Map’s JGEM.
Figure 23 : National Mapping’s Johnson Ground Elevation Meter.
The Sperry Sun Well Surveying Company, not Sperry Rand as in some literature, manufactured JGEM was a highly modified General Motors Corporation truck having 4 wheel drive and simultaneous front and rear steering. Using a very sensitive pendulum and a fifth bogey wheel the difference in height was determined using the principle of the sine of the inclination of the vehicle multiplied by the distance run. The JGEM’s specifications were developed by the United States Department of the Interior, Geological Survey, Topographic Division. These specifications came from many years of use of the 1952 trailer mounted version of the equipment. Please refer to Figure 24 below.
Figure 24 : Trailer version of the Johnson Ground Elevation Meter (best photo quality available).
As the fifth wheel required a relatively smooth surface to maintain accuracy, the quality of Australia’s roads did not allow maximum accuracy to be achieved. Nevertheless, National Mapping did obtain results with an error of less than 10 feet (3 metres) in 50 miles (80 kilometres) at 15 miles an hour (25kph) on reasonably good road surfaces, averaging 100 miles (160km) per day. Overall, between 1964 and 1970, around 5% of all National Mapping vertical control was obtained using the Johnson Ground Elevation Meter. In flat country, where the sealed road system was reasonably extensive, the JGEM demonstrated that a network of heights could be obtained which would allow contouring for 1: 100,000 scale mapping from 1: 80,000 scale aerial photography. Contours in photogrammetric blocks 3 and 4 in Queensland, block 18 on the Victoria/New South Wales border and part of block 32 in New South Wales were plotted from heights obtained by JGEM traverses. For their part, RA Survey used their JGEM to obtain height information within some fifteen 1: 250,000 scale map sheets. The Photogrammetric Blocks which used heights obtained by the Johnson Ground Elevation Meter to provide vertical control for mapping are shown in Figure 25 below.
Figure 25 : Photogrammetric Blocks which used heights obtained by Nat Map’s Johnson Ground Elevation Meter to provide vertical control for mapping are shown in blue; green areas are where RA Survey used their JGEM.
As well as being used to obtain vertical mapping control points, Nat Map’s JGEM was applied to two other tasks where reasonably accurate elevation data were required. One such task was to obtain longitudinal height profiles of a number of airstrips, mainly in outback Queensland and New South Wales. A consistent height datum for each airstrip profile was obtained by a connection to the nearest Bench Mark. These profiles were used in the calibration of the radar based Airborne Profile Recorder system. The other task was to provide check heights to ensure that the contractor’s APR heights were to an acceptable accuracy.
While height acquisition by vehicle was faster than traditional levelling, its areas of operation in Australia were extremely limited and for the majority of Australia’s then outback, entirely unsuitable.
Vertical Control by Radio Altimeter
To assist with the compilation of aeronautical navigation charts in remote parts of Canada, its National Research Council investigated a means of obtaining airborne terrain profiles. Experiments with airborne altimetric measuring instruments thus began around 1946. The APN-1 radio altimeter, as shown in Figure 26 below, had been designed to be installed in aircraft to provide direct measurement of altitude relative to the terrain during flight. A transmit and receive antenna were mounted on each side of the fuselage with the frequency difference between the transmitted and received signal being converted to the aircraft to ground distance or height above terrain. This value was conveyed to the pilot via a cockpit mounted meter. Height above terrain however, was limited to 4000 feet with the APN-1 and its testing proved that specifically designed equipment was needed to acquire height data from an airborne platform over extensive areas.
Figure 26 : Configuration of an APN-1 Radio Altimeter showing cockpit instrumentation and transmit and receive antennae.
An airborne pulsed radar instrument was designed, which by 1947 was giving terrain profiles to an accuracy of 200 feet. This accuracy was adequate for the Canadian’s purpose as the required contour interval was 500 feet.
In his 1948 report A Pressure-height Corrector for Radar Altimetry, McCaffrey of the National Research Council, Canada, stated that…in order to compensate for deviation in height - it is impossible to maintain constant height over long periods - pressure altimeters have been photographed at the rate of 1000 to 2000 times per hour during flight. Data so obtained has been used to correct the radar graph to the nominal operating height with an accuracy of perhaps ten to fifteen feet.
McCaffrey’s report went on to describe a new datum stabiliser using a Hollsman type altimeter. This was successful but was later superseded by the Hypsometer using the boiling point of toluene to detect pressure changes.
The Hypsometer thus became an essential element of the terrain profiler system, essentially monitoring the airborne datum during terrain profiling operations. The airborne datum was the aircraft altitude, but as aircraft altitude is measured by the change in atmospheric pressure, the airborne datum was actually a pressure surface. Maintaining a constant flight level required maintaining a constant pressure or isobaric surface. As precisely following this surface was impractical, the approach was the measurement of any departure from a set altitude and later make the necessary corrections. At profiling altitude, the atmospheric pressure at that height became the reference sample and throughout profiling operations any deviation in height and therefore pressure was measured and recorded. The deviations could then be used to adjust the radar terrain profile. As the deviations were relatively small, generally less than ±100 feet, their magnitude could be measured to a high accuracy.
This accuracy was achieved by using the boiling point of toluene to detect pressure changes. Inside the Hypsometer, a vessel of liquid toluene was kept boiling by resistors. A sensing element mounted above the boiling toluene formed one side of a bridge circuit; the other side of the bridge was continually balanced by a servo operated potentiometer in the Hypsometer chassis. An outside pressure change meant that to keep the bridge balanced extra electrical current had to be supplied to the circuit. This additional electrical current was electronically converted to a change in height and applied to the measured terrain clearance to give the terrain profile. The terrain profile then being the record as referred to the selected initial flying height or isobaric surface. The change in height as measured by the Hypsometer was not directly recorded but could be deduced from the difference between terrain clearance profile and the terrain profile at any given point.
The Photographic Survey Corporation (PSC) Limited of Toronto, developed similar equipment for commercial use. Speert (1950) stated that in 1950 the United States Geological Survey (USGS) undertook what is believed to be the first attempt in the United States to obtain ground elevations for topographic mapping by airborne electronic methods. The work was done under contract, by Photographic Survey Corporation, Limited, of Toronto, Canada, and its affiliated company, Kenting Aviation, Limited, using the Airborne Profile Recorder (APR). Approximately 78,000 square miles in central Alaska were covered on this project. Of 29 control elevations examined in the reconnaissance portion of the project, only one feature exceeded the allowable spread of 30 feet. Four runs over this feature showed height values with a total spread of 44 feet, but three of the values fell within a spread of 24 feet. One reading was obviously bad. Except for this feature, the maximum spread was 26 feet, and the average spread was 14 feet. Unfortunately, there were no opportunities to check the absolute elevations of any of the control features in this area.
By 1955, the Airborne Profile Recorder (APR Mark IV) was refined enough to give accuracies of ±15 feet and around 1959 the Mark V APR was introduced, which was relatively smaller and lighter with an accuracy of ±10 feet. The Canadian Applied Research Limited (CARL) Mark V, radar profiler was able to operate from 1,500 to 45,000 feet and measured the aircraft to ground distance. The 3 centimetre band radar beam was radiated at 2000 Hertz and projected in about a 1° cone from an antenna in the centre of a 44 inch (1.1 metre) parabolic reflector. This reflector was mounted to point directly below the aircraft such that an area with a radius of about 55 meters on the ground would be illuminated from a normal operating altitude of about 10,000 feet above sea level.
The earliest radar profiler trials used an F24 camera adapted to 35mm strip film for the positioning camera because no commercial camera was available. This led to the later development of the CARL 35mm positioning camera, and associated controls as used on the CARL Mark V APR. Camera magazines held 400 feet of 35mm film and lenses of 90mm or 28mm focal length could be used. Controlling the positioning camera’s rate of exposure was an intervalometer. The axis of this camera, marked by the cross in the image, was directed down the axis of the radar reflector. For accuracy, it was critical that the vertical axes of the radar antenna and positioning camera were collinear.
A few lines on page 25 of the Ottawa Citizen newspaper of Wednesday 19 November, 1958, stated that the Canadian designed and built Airborne Profile Recorder (radar surveying unit) has been selected by the USAF for its massive, program of bringing the world's geography up to date…The Recorder is made by Canadian Applied Research Limited, a company owned by A.V. Roe (Canada) Limited. Annexure D provides an overview of PSC’s development. Starting with PSC’s foundation to its ownership by AV Roe (Canada) Limited, a wholly owned subsidiary of Hawker Siddeley, and later integration with de Havilland Canada’s Special Products division to form SPAR (Special Products and Applied Research) Aerospace Limited, to today being a part of MacDonald Dettwiler as MD Robotics, a subsidiary of its MDA Space Missions division. The overview concludes with MDA’s association with National Mapping’s, Australian Landsat Station (ALS) and later Australian Centre for Remote Sensing (ACRES).
The USAF adoption of the APR appears to have resulted in its satisfactory performance in tests conducted in the United States as reported in the 1961 paper, Investigation of the Air-Borne Profile Recorder, by Chester Slama of the US Army Map Service (Slama, 1961).
Vertical Control by an Airborne Profile Recorder
The Airborne Profile Recorder (APR) Mark V, was made by the then Canadian Applied Research Limited (CARL), a company owned by AV Roe (Canada) Limited, which in turn was a wholly owned subsidiary of Hawker Siddeley.
The CARL Mark V, APR radar transmitter/receiver, 35mm positioning camera, Hypsometer and associated graphical chart recorder, power supply and controllers were combined into three major components for aircraft installation. Figure 27 below shows this instrumentation prior to installation in an aircraft.
Figure 27 : Canadian Applied Research Limited (CARL), Mark V, Airborne Profile Recorder instrumentation.
The radar reflector is left with 35mm camera in foreground and operator’s console right.
The radar’s parabolic reflector was mounted to point directly below the aircraft. The 35mm positioning camera was rigidly mounted to the reflector assembly so that the optical centre of the camera was in alignment with the beam centre of the antenna. Electronics controlling the radar’s transmitter-receiver was mounted close to the antenna. This equipment weighted some 50lbs.
The size of the antenna and reflector in this system provided a 1° cone of the transmitted signals at the point of emission. At an operating altitude of 3000 metres these signals sampled a circular area of the terrain some 55 metres in diameter. To reduce the sampled area a much larger transmitter reflector would have been required and this was not considered practicable.
The operator’s console, as shown in Figure 28 below, was mounted in the aircraft cabin. It contained the graphical chart recorder and its controls at the top, with power panel below. Camera intervalometer and frame counter below that and near the base an inverter and the Hypsometer. A total weight of about 100lbs.
Figure 28 : Photograph (left) of the APR operator’s console as mounted in the aircraft cabin where below the graphical chart recorder, inverters and the Hypsometer can be seen.
The enlargement (right) shows the graphical chart recorder and paper chart.
The inverter converted the aircraft power supply from 28V DC, to 115V AC at 400Hz for the powering the APR system controlled by the power panel. The Hypsometer was connected to a pitot tube which conveyed outside pressure changes via a static line. A graduated dial (±150 feet range) on the hypsometer panel gave the operator an indication of height above or below the isobaric surface, and a repeat instrument was mounted to the right of the pilot to enable him to control aircraft height. Throughout the profiling operation the aim was to keep the aircraft as close to the selected isobaric surface as possible by keeping the Hypsometer gauge reading as close to zero as possible and certainly not outside the ±150 feet range. Many miles of profiling were hand flown by the pilot until automatic pilots (autopilot) became accurate and cost effective. While the autopilot then reduced the size of the aircraft’s departures from the isobaric surface, these instruments did not negate them totally.
The graphical chart recorder provided the permanent record of terrain profiling operations. The dual (red and blue) pen recorder was mounted on a hinged assembly which opened like a door to permit chart changes and inspections. Chart rolls were about 122 feet long and exactly 12 inches wide. Margins for the chart drive sprockets used 1", leaving 11" to record the outputs. A blank chart looked like a long roll of graph paper having a lightly printed grid. The grid itself was not really used other than the first printed line above the sprocket holes which became the chart datum. This line rather than the very first printed line was adopted as it was unlikely to be affected by any tearing, uneven rolling or mishandling of the chart. The vertical scale was set at 1.1 inches to 100 feet thus permitting a 1,000 feet rise or fall to be recorded. The horizontal scale was variable, since aircraft ground speed varied, but was around 1,250 feet per inch. The chart moved at 10 inches per minute (ipm) but could be varied from 5-20ipm in 5ipm increments. A servo motor actuated the red and blue pens across (left/right direction during recording) the chart to depict the terrain profile (red) and terrain clearance (blue) traces. The difference between the two traces gave the Hypsometer reading, which could be checked visually against the Hypsometer gauge. On the right hand edge of the chart a third pen recorded the positioning camera exposures. As the ink traces on the chart were the only record, maintenance of the operation and pen sensitivity was critical. Figure 29 below, shows an example of an APR chart and associated 35mm film.
Figure 29 : Example of an APR chart and associated 35mm film.
The terrain profile is the red trace and the location in the terrain is established by the 35mm positioning photography which is related to the chart by the blip made at the instant of each exposure.
The autostep could be operated either manually or electrically. The coarse range reading was shown on a dial marked from 1 to 20.5 (x 1,000 feet). A manual coarse range control from 0, 5, 10, 15, 20 (x 1,000 feet) enabled the operator to adjust the coarse range setting. The sum of the coarse range and step range dials were recorded on the chart by the operator and which, when added to the height indicated by the pens, gave the total height above the terrain.
The radar system was pulsed at a rate of about 2,000 pulses per second, and the echo delay time went to a quartz crystal controlled oscillator for maintaining precision of measurement. The terrain profile drawn on the chart was thus the result of a sampling of statistically derived groups of measurements. This approach smoothed out excessive noise in the profile and provided a better and more usable trace.
While not part of the APR equipment itself, a necessary requirement for profiling was a drift sight or drift meter. Such a sight was an optical instrument which allowed the terrain below the aircraft to be viewed from inside the cabin; it can be thought of as a periscope in reverse. It was used in aircraft navigation for determination of drift and to ensure the aircraft flew along a precise section of the terrain. Its design allowed it to be rotated through 360° and the rotation could be measured on a scale; 0° / 180° on the scale was generally aligned parallel with the aircraft’s centreline. Drift was determined by following an object in the terrain and to make the object appear to move parallel to the sight’s reticule lines by carefully rotating the sight. The amount of rotation, estimated to ¼°, from 0° was the aircraft’s drift. The drift sight can be seen in Figure 30, just to the left of the operator’s console, near the window.
Figure 30 : Photograph of the APR operator’s console as mounted in the aircraft cabin of an ADASTRA Hudson.
Just to the left of the console, near the window, is the drift sight (Photograph courtesy Owen Smith).
Maintaining a constant flight level during terrain profiling operations required maintaining a surface of constant pressure. Such a surface is known as an isobaric surface where the atmospheric pressure is equal at every point.
An isobaric surface is generally curved, and although a small part can be considered a plane it is never horizontal. There are also diurnal variations in its altitude. The altitude of the isobaric surface above sea level depends upon the density, and hence the temperature, of the intervening air column. In regions where the air in that column is cold and dense, the altitude of that isobaric surface will be lower than over a region where the air is warmer and less dense.
TJG Henry of the Meteorological Division, Department of Transport of Canada in 1947 determined that the slope of an isobaric surface could be calculated and this calculation became known as the Henry correction. Annexure E contains more detail and shows that the Henry correction is proportional to the sine of the drift angle times distance flown.
Figure 31 : Diagrammatic view of the Henry Correction for the southern hemisphere.
In summary, Henry found that in the southern hemisphere (for the northern hemisphere the results are reversed) that an aircraft flying from a low pressure to high pressure region will drift to the right or starboard, incurring starboard drift. In these same conditions, if the aircraft maintains a constant pressure altitude, that is it flies an isobaric surface, it will fly a path above the theoretical surface. Thus, with starboard drift the height correction needed, actual to theoretical, is positive. Conversely with port drift, the sign of the Henry correction is negative in the southern hemisphere. Figure 31 above diagrammatically shows a line A to B being flown and the atmospheric pressures at the time of flight. In these conditions, if at position A, the reference pressure of 700 millibar was captured and that 700 millibar pressure surface was precisely maintained by the aircraft to point B, then the isobaric surface has risen by 1,200 feet (10,600 minus 9,400 feet) above the theoretical surface. In theory, the Henry correction for this flight would be found to be around +1,200 feet. From Figure 31 it can also be seen that the reverse is true if the line B to A was flown. If the 700 millibar pressure surface had been captured at point B then at point A the Henry correction would theoretically be -1,200 feet.
It is very important to note that in this example the value for the Henry correction is simply applied in relation to the theoretical surface. Assume now that points A and B in Figure 27 are at sea level. The terrain clearance, aircraft to ground distance, would be 9,400 feet at A and 10,600 feet at B. If the reference altitude of 9,400 feet at A was taken to have been maintained to point B, then the height of point B would be calculated from reference altitude minus terrain clearance measured at point B (9,400 minus 10,600 feet) giving -1,200 feet as the height of point B. Clearly to get the correct height at point B the value for the terrain clearance at point B has to be reduced by the value found for the Henry correction. That is, reference altitude minus terrain clearance measured at point B reduced by the value of the Henry correction at point B (9,400 minus (10,600 minus 1,200) feet) now giving the correct value of zero feet. This example demonstrates why in some literature the sign of the Henry correction is reversed when that correction is made to the terrain clearance distance.
Thus, on any airborne terrain profiling operation, the slope of the isobaric surface was the Henry correction divided by distance flown and this value could be found by monitoring aircraft drift and calculating the distance flown, as described in Annexure E. With the slope of the isobaric surface calculated, the required level surface could be redefined at any point during profiling. In the southern hemisphere, to derive the correct sign (±) for the Henry correction the sign (±) for the determined amount of port drift is negative and conversely the sign (±) for the determined amount of starboard drift is positive.
Henry’s work on the theory of pressure pattern determination from drift measurements, was an advancement of that by Dr John C Bellamy (United States Air Force) in 1943. Bellamy developed a procedure for determining a close approximation to the beam wind component, that is the force of the wind acting on the aircraft and affecting its course. Using the geostrophic assumption and pressure measurements made at an aircraft’s altitude with radio and pressure altimeters, the drift angle calculated by this method of navigation was known as Bellamy Drift.
Mathematically modelling the atmosphere via Henry’s method was useful but all the real world variabilities could not be included is such a simple correction. Airborne terrain profiling operations were such that lots of small factors could accumulate and effect the final accuracy of the derived terrain heights. An operational procedure was thus developed which would account for any residual errors and refine the final terrain heights.
The procedure was to overfly a datum surface after takeoff and at profiling height, and while still at profiling height, overfly a datum surface before landing. While the datum surface was usually the same place, operationally it could be two different locations. Without these datum runs an operational flight could not yield reliable data. The size of the radar beam as it was reflected from the terrain demanded that such a datum surface be of significant size and its surface relatively plane. Further, the height of this surface needed to be found using a suitably accurate traditional survey method. Airfields, lakes or a calm sea thus became the most suitable sites for use as datum surfaces. In Australia, airfields became the datum surface of choice and the specific one that was being used as the operational base for the airborne operations generally became the datum surface.
To provide the necessary datum information any airfield used as a datum surface generally had the centreline(s) of its strip(s) levelled and connected to the Australian Height Datum (AHD). As the area around airstrip intersection(s) and/or the windsock could be easily identified on the positioning photography, heights of features in these areas were acquired as well.
After takeoff, the profiling aircraft would climb to operating altitude. Here the reference sample of air would be captured by the Hypsometer. With all equipment operational and initial drift found, the centreline of the airfield would be overflown. The airfield’s profile was recorded on the chart and the profile’s position by the 35mm camera. The aircraft would then navigate to the profiling area. During this time there was no need to run the profiler or camera but maintaining height with reference to the Hypsometer and the recording of air speed and drift were vital for later Henry correction computation. After completing the required terrain profiles and recording the navigation data on the return to base, the centreline of the airfield would again be profiled with the profiler and camera running. The aircraft would then land.
While the chart recorder and camera only needed to record the necessary profiles, it was critical that the Hypsometer be operational from the after takeoff datum overfly until after the prelanding datum overfly. During this same period of the flight, the pilot had to maintain the aircraft in close proximity to the isobaric surface as shown on his Hypsometer repeater. Any failure/problem with any one of these aspects could render the whole flight meaningless.
To validate a profiling flight, the Henry correction would be calculated for each section of the flight where the drift was constant. The mathematical sum of the Henry correction for each section of the flight, gave the total Henry correction for that flight. As the Henry correction at the after takeoff overfly was zero, the elevation of the airfield plus the height of the aircraft above the airfield, as given by the profile on the chart, gave the height of the isobaric surface on AHD. At the prelanding overfly, the elevation of the airfield plus the height of the aircraft above the airfield, was similarly calculated. However, this height had then to be corrected by total Henry Correction for the flight. In theory, this final value should have given the same height for the isobaric surface as at the start, but invariably there was a residual. As this residual accumulated as the flight progressed it could be proportioned per unit of distance flown.
Operationally therefore, the terrain heights extracted from the terrain profiles were not only corrected for the slope of the isobaric surface, but also for a residual which accounted for the myriad of other small variables which otherwise impacted final accuracy.