The National Levelling Survey
and the Australian Height Datum
Prepared by Paul Wise June 2014 and updated October 2015
The Geodetic Survey of Australia was completed in 1965 and provided a homogeneous network of horizontal positions. However, height in a national context had also to be determined and this was achieved through a separate process. As height was generally referenced to mean sea level which varied across the continent, the necessary national mean sea level datum had to be established. The National Levelling Survey generated level traverses to carry height values, relative to a locally adopted datum, in closed loops which were later adjusted individually and nationally. Subsequently, after the national establishment of a Mean Sea Level datum, all level traverse loops were then corrected to be relative to this datum which was called the Australian Height Datum (AHD).
Typical scene of level and staff in use.
Method and equipment
The most accurate way to obtain a height difference between two points was to use a surveyor’s level (technically a spirit level) and staff. Essentially, the spirit level instrument consisted of a telescope and attached spirit level tube. The telescope’s line of sight was set in a horizontal plane when the spirit level bubble was centred in the spirit vial. The levelling staff which was then graduated in imperial measurements (rather than in metric) was observed at two successive points and the difference in the observed staff readings gave the difference in height between the two staff stand points. By knowing the true height value of the first point and mathematically applying the height difference from the levelling gave the true height of the second point.
When the point of unknown height was some distance from the point of known height, a series of such setups and observations were required to carry the height from the initial point of known height to the destination point of unknown height. This series of setups and observations formed a level traverse (refer to the diagrammatic level traverse above). For efficient operation on such traverses, two staves were used; one being read (the backsight staff) while the second (the foresight staff) was moved forward to a suitable spot along the traverse route, to be read next (as shown in Instrument setup 1 above). Once the foresight staff was read, the level instrument itself was moved forward past the foresight staff to now look back and read what had been the foresight staff. Thus the foresight staff became the backsight staff. While this had been happening what had been the backsight staff had moved forward past the level instrument to become the new foresight staff (as shown in Instrument setup 2). This leap-frogging process of the foresight staff becoming the backsight staff each time the level instrument itself was moved forward, continued as long as necessary until the intervening distance between the two points was covered, and the foresight staff was at the destination point (as shown in Instrument setup 3).
The introduction of automatic levels greatly improved the speed and accuracy of the levelling survey. The use of automatic levels reduced the time taken and human inaccuracy in the centring of the level’s bubble. 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.
View of the staff through a modern level diaphragm showing crosshairs and stadia. The horizontal crosshair indicates a reading of 1.422 m.
The top stadia indicates a reading of 1.500 m and the bottom stadia 1.344 m giving an intercept of 0.156 m, meaning the staff is at a distance of 15.6 m (0.156 m x 100) from the level instrument.
The telescope of a level had in the diaphragm a central horizontal and vertical wire plus two horizontal stadia wires set equidistant, above and below the central horizontal wire. The central horizontal wire indicated the horizontal plane. The two stadia wires were set so that the distance between these two wires (known as the intercept) when read off the staff was one hundredth of the distance between the level and the staff when the line of sight was horizontal. The values read from the staff for all three wires were recorded, so that not only could the height difference be measured but also the distance covered by the level traverse could be determined. The traverse distance was required to ensure that the final levelling result met specification. The intercept readings were also used to monitor the level to staff distance so that these distances were kept approximately equal and were never excessive.
The type of levelling staff employed was also a consideration as some early models were found to be inaccurate. This was especially true of the telescopic type of staff. A folding wooden staff with warranted graduations was found to be the most satisfactory.
Levelling traverses for the national survey generally ran along roads or tracks. At set intervals of around 4 to 8 kilometres a bench mark was placed as the permanent mark and height reference for any future levelling work.
In practice, a National Mapping levelling field party consisted of an instrument man and two staff men, with each staff man having a vehicle. For each reading of the staff the staff man got out of the vehicle and raised the staff as required. After the backsight staff was read, that man drove to the instrument man’s position and picked him up and drove him to the next instrument set-up and then drove on to the next foresight staff position. To protect the level instrument, but to also save time dismantling it, the instrument man usually travelled between points sitting on the vehicle’s tail-gate cradling the instrument.
Experience had shown that 300 feet (100m) was the limit for good observation of the staff, and that keeping the backsight and foresight distances about equal reduced systematic errors. As a guide to optimum instrument and staff positions, and an indication to the party that the 300 feet length of sight was not being exceeded, a nylon rope 300 feet long was towed behind each vehicle. The true length of sight was observed by stadia and recorded in the level field book as explained above.
The Natmap procedure outlined above was for remote areas and was modified when operating in settled areas. Having two vehicles always nearby with food, water, shelter and communications provided security. In addition, there was an efficiency aspect in that the overnight camp could be set up at the end of a day’s work wherever that happened to be, and then work could be readily continued the next day.
Nat Map's then Supervising Surveyor Geodetic Branch, Klaus Leppert, outlined a number of the technical issues associated with the third order levelling in a 1967 paper Problems encountered in the Use of Third Order Levelling for the National Levelling Grid, which is available via this link.
Survey Heights Prior to the Australian Height Datum
Prior to any form of systematic levelling, height determination had been carried by using simple trigonometry (where height difference is equal to the distance between two points multiplied by the tangent of the vertical angle between the points). For small areas this method was acceptable and national height datum differences were irrelevant.
A highly refined version of this approach Simultaneous Reciprocal Vertical Angles was used during the geodetic survey to carry approximate height. As the name suggests, the vertical angle between any two points was read at both points to the other simultaneously. Such an approach meant that the atmosphere affected the theoretical light ray from each point equally. As height values were thus carried throughout the national geodetic survey from east to west across Australia, national height datum differences over vast distances could no longer be ignored.
Australia’s first two height datums were based on the tide gauges at Williamstown, Victoria (1859) and Fort Denison, Sydney, New South Wales (1866). However as Lines (1992) stated:…The physical datum [for New South Wales] was a brass plug set in the masonry of the Lands Department building in Bridge Street, Sydney. This was related to the tide gauge at Fort Denison, a small once-fortified island situated 6-7 kilometres inland from the entrance to Port Jackson, but considered sufficiently close to the entrance to represent the level of the open sea. For this purpose, the Fort Denison gauge was much better situated than those of Melbourne, Port Adelaide and Brisbane which were too remote from the open sea and in too confined waters.
Aeronautical charting increased the demand for more and accurate heights over large areas. To meet this demand Barometric and Altimetric instruments were used with varying success (Lines, 1992).
Original Fort Dennison tide gauge on the left of picture. (Reprinted with permission after Janssen et. al. (2013)).
Fuller (circa 1980s) recorded that the predecessor of the Australian Survey Office (ASO) in the Department of Interior was:…requested in 1950 to assist the Bureau of Mineral Resources (BMR) with survey control for the network of gravity measurements being planned across Australia, and later for surveys relating to seismic and geological work.
For gravity purposes, the observation stations were established 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 Interior as documented by Fuller above. Not only did this work assist the gravity surveys but later provided data on which the specifications for the national survey were based.
The contribution of the Department of Interior levelling to the national survey should be commended. Despite necessary minimal checks on the levelling and perhaps rough, but practical at the time, bench mark establishment, it adequately served the purpose for which it was intended; the provision of heights for the Bureau of Mineral Resources Gravity Section. Some years later when compared with the results of more modern levelling only a very small percentage of the Department of Interior levelling was found to be unacceptable for national levelling survey purposes.
Typical Department of Interior bench mark collar.
Department of Interior bench mark on the roadside.
In 1961 the Commonwealth Government made special funding available to support the search for oil in Australia and approval was given for some of this funding to be allocated to surveys which would link the existing levelling within and between the various sedimentary basins.
Map with the mainland sedimentary basins of Australia coloured for differentiation.
National Levelling Survey progress maps at 1955, 1960 and 1965.
By the early 1960s, Australia had some 4,800 kilometres of level traverses which had been completed by the respective state lands departments of Western Australia, Victoria and New South Wales. This work was in support of the States’ programs and was completed to varying standards. Another 16,000 kilometres of level traverses, mainly by the Commonwealth’s Department of Interior, supported gravity surveys of the sedimentary basins.
At this time, Bruce Lambert Director of National Mapping decided that if these surveys were to be of any practical value within a usable time scale then use would have to be made of contract surveyors using readily available equipment and working to third order standards of accuracy. On the recommendations of the Division of National Mapping some of the special Commonwealth funding went to advancing the national levelling program. Refer Lambert's 1989 personal audio narratives accessible via this link.
Between 1961 and 1966, GRL (Rim) Rimington Assistant Director of the Division of National Mapping, in cooperation with the officers of the various state lands departments, planned and organised these control levelling surveys and arrangements were made with the State Surveyors General for their staff to supervise the contract operations. More detail regarding the contracting arrangements can be found in Klaus Leppert's 1967 paper The Levelling Survey of Australia, a copy of which is available here. The National Mapping specifications for contract third order levelling may be viewed via this link.
By 1965 the total amount of control levelling had reached 129,000 kilometres. In this same year the Minister for National Development committed to undertake a topographic mapping program to cover the whole of Australia at a scale of 1:100,000 and with a 20 metre contour interval. The opportunity was then taken to extend the existing third order levelling surveys into a national network. At the end of 1970, 161,000 kilometres of levelling had been completed. A total of 232 contracts for bench marking and levelling had been arranged at an overall cost of almost $2 million.
The work selected for inclusion in the final simultaneous mathematical adjustment of the third order levelling network totalled 97,320 kilometres of levelling. Of this total, the Department of Interior contributed 15 per cent, the various States 22 per cent and National Mapping 3 per cent. The private sector played the major role, in that private surveyors under contracts arranged by National Mapping and overseen by the States, carried out 60 per cent of this levelling. Some of these levelling contracts involved some fairly arduous work. For example, in mid-1970 South Australian-based contract surveyor John Gibson and his assistants completed around 500 kilometres of two-way levelling from Pedirka in South Australia to Birdsville in Queensland. The work included a 370 kilometre stage across some 1,100 sand hills in the Simpson Desert. This major feat is commemorated on a plaque outside the Birdsville hotel along with monuments to other historic crossings of that desert. For more information on John Gibson’s life and career please use this link and for more details of John Gibson’s 1970 Simpson Desert leveling survey please use this link .
Early 1970s : John Gibson at the commemorative plaque at Birdsville (courtesy Wayne Gibson).
2012 : Inscription on the commemorative plaque at Birdsville (courtesy Laurie McLean).
Why third order levelling for the primary survey
The geodetic survey of Australia was to first order standard. Why then was the levelling survey of Australia to only third order standard?
Cost-effectiveness was the short answer to achieve the best result within the time and funds available. It was recognised that third order levelling provided heights that would be acceptable to the national topographic mapping program, general engineering purposes, and for the co-ordination of the levelling surveys undertaken in support of the gravity observations. It was also considered likely that the adjustment of the levelling network as a whole would produce a homogeneous survey approaching in accuracy the results obtained in the past from levelling networks surveyed to traditional patterns. If not, then subject to availability of funding, it could be strengthened at leisure with higher order surveys if these could be justified.
Another compounding factor with a then unknown degree of impact was crustal movement of the Earth’s surface. A high order levelling survey by its more precise nature would take significantly more time and incur considerably more cost to complete. However, by or even before completion its results could well have been impacted by vertical movements in the Earth’s crust that may have occurred during the many years it would take to carry out. Thus such a high order levelling survey was never a practical or economic consideration.
The practicality of the decision to work to third order standards was illustrated when four levelling traverses converged at the bench mark adjacent to the old government stock route well, Connor Well, about 90 kilometres north from Alice Springs along the Stuart Highway in central Australia. Based on the then adopted value for mean sea level at Port Hedland, Darwin, Bundaberg and Thevenard (near Ceduna) the height of the bench mark from the four separate traverses ranged within 1.3m.
The National Levelling Survey at 1970 prior to adjustment. The blue pin shows the location of Connor Well.
Mean Sea Level for Australia
While the level traverses were being completed, data for the determination of a mean sea level datum for national purposes was being collected. These data came from a practical and judicious spread of recording tide gauges around the Australian coast.
The number and placement of the tide gauges was based on the characteristics of the tide which was already known to change considerably around the Australian coastline. The daily tide range varied in places from 1 metre to about 10 metres; some places had two high and low tides each day, and some only one. Tides were affected by several components of astronomical and fluid motions and could not be predicted with any accuracy, except with the availability of data coming from an analysis of previous records.
In an ideal world, tide gauge data for a full lunar cycle of 18.6 years would have been collected but of necessity only one calendar year of observations was attempted. The actual full year of observations took place in 1967, although 24 stations had contributed data since early 1966, and most continued to contribute data until the end of 1968. By this time, the information was deemed sufficient for the determination of the Australian Height Datum.
Ultimately surveys and calibrations of the 25 mainland gauges were supplemented by later surveys of gauges installed at Melville Bay, Darwin, Centre Island (Gulf of Carpentaria), Port Kembla and Wyndham.
Copies of the recorder charts from all tide gauges in the program were sent by the tide gauge operators to the Horace Lamb Centre for Oceanographical Research at the Flinders University of South Australia. The Centre extracted the hourly heights of sea level and undertook all computations associated with determining mean sea level and tidal constants.
For the national adjustment the determined mean sea level height at each tide gauge was held at zero metres. In actuality, it was the height of the nearby reference bench marks that were held fixed in the adjustment. A simple example is to assume that the height of the permanent bench mark is 1 metre above the tide gauge when the tide is at Mean Sea Level height. In the adjustment, the height of the permanent bench mark would be held fixed at 1 metre which was the same as if the tide gauge height itself was fixed at 0 metres. The logic behind this approach was that tide gauges were highly vulnerable. If a tide gauge was damaged and then replaced, the only height affected was that between the tide gauge and the permanent bench mark. The whole adjustment did not have to be recalculated.
At each Australian tide gauge the difference in height between the tide gauge’s measuring system and 3 permanent tide gauge bench marks nearby is obtained by very precise levelling. Once this difference is obtained and mean sea level is established the heights of the permanent tide gauge bench marks are fixed. The bench mark height then becomes the permanent reference datum.
The Australian Height Datum
The Australian Height Datum (AHD) 1971 is defined as the datum surface derived from a simultaneous adjustment of the two way levelling network holding 30 tide gauges fixed at their mean sea level values. The adjusted height of the ground mark at the Johnston Geodetic Station is 56.300 metres. Observations for the determination of mean sea level were carried out from 1 January 1966 to 31 December 1968 at 29 tide gauges and from 1 January 1957 to 31 December 1960 at Karumba tide gauge.
By holding the height of mean sea level at zero metres at the 30 tide gauge stations connected to the levelling network, the land based levelling was warped to fit these zero values. Even so, the estimated standard error of adjusted heights in the centre of Australia was only 0.35 metres, sufficient for all practical purposes. Subsequent to the national adjustment, the National Mapping Council at its 29th meeting in Canberra in 1971 formally adopted this work as the datum to which all vertical control for mapping is referred.
When the adjustment was completed, there were 42,000 permanent bench marks with values in terms of the AHD spread across the continent. In this instance, Tasmania was not included in the AHD, as precision levelling over the intervening distance was not practicable. Tasmania was the subject of a later independent adjustment tied to four tide gauges in that State.
As described below, Natmap's levelling field parties also undertook a program of levelling to accessible first order trigonometrical stations. Not only was the height of the ground marks determined, but also the height of the cairn, vanes and top of the centre pole. During the first order geodetic survey, height had been carried between these horizontal control stations by observing reciprocal vertical angles throughout. Now with the completion of the levelling adjustment, and the additional levelling to the first order stations, trigonometric heights of all horizontal control stations were adjusted to the AHD.
With the completion of the adjustment of the levelling survey, and its subsequent transformation to the Australian Height Datum, another great task comparable in magnitude to the national geodetic survey had been accomplished. This national levelling survey is also an undertaking without a peer. In assessing its significance consideration must be given to the elapsed time from its start to its finish, the high degree of co-operation between the various government organistations and significant involvement of the Australian survey industry as well as the wide variations in the often harsh operational conditions across the continent.
Natmap staff known to have been involved in this program
David Cook, Harry Granger and Peter O’Donnell were party leaders on this program and had in their parties Robert Bryant, Don Gray, Bill Jeffery, Fred Reardon, Andy Rodgers, Peter Walkley, Mike Whelan and John Woodger.
Their work included :
It should not be forgotten that Nat Map Senior Surveyor Harry Granger passed away whilst on levelling field duties. The members of the Natmap Levelling Section in the mid-60s are shown in this photograph.
While researching for this article a number of papers were found that highlighted present day deficiencies in the AHD. As the original National Levelling Survey and Australian Height Datum article was primarily an overview of how the AHD came into being, such more recent papers were not then discussed. However, attention has since been drawn to Filmer and Featherstone's 2009 paper, Detecting Spirit-levelling Errors in the AHD: recent findings and issues for any new Australian height datum. After subsequent consideration of Filmer and Featherstone's paper, some further expansion of the AHD article was warranted for reasons of completeness and balance.
As explained in the primary article above, the establishment of the AHD in1971 was to provide a homogeneous framework of heights suitable for mapping and most other projects of the time, requiring a standard height reference. Since the adoption of the AHD, however, advances in computational programing and power plus the acquisition of independent GPS data and analysis of the 1975 rapid two-way first-order levelling along the northeastern coast of Australia, has allowed greater analysis of the AHD and indicated some inadequacies in its basis.
These deficiencies are associated with the quality of the original levelling observations and their reduction. Not that this work was sloppy just that in hindsight if atmospheric conditions had been consistently observed and then applied, loop misclosures may have been reduced and further possible errors detected. However, the holding of the height of mean sea level at zero metres at the 30 tide gauge stations connected to the levelling network introduced an artificial north-south slope of around 1.5 metres in the datum.
The amount of this north-south slope may seem relatively small (less than 1 millimetre per 2 kilometres or about 0.1 seconds of arc), but Filmer and Featherstone point out that it may now be significant to some Earth science related studies such as high-precision geodesy. In addition, digital elevation models (DEMs) based on AHD are used in Australia for resource and environmental management, river geomorphology and hydrological modelling, height-change analysis associated with seismicity and the computation of gravimetric terrain corrections. Independently derived DEMs from space-borne sensors indicate that several of the DEMs available for Australia are now suspect, most likely caused by the AHD being their foundation.
In its publication ICSM Guidelines for Digital Elevation Data the Intergovernmental Committee on Surveying and Mapping stated that...The Australian Height Datum (AHD) is the official vertical datum of Australia and should be used for all land surveys. These guidelines further outline the complex issues surrounding establishing any new vertical datum. Such a new vertical datum not only has to accurately reflect the terrain but also integrate with GPS elevations and bathymetric depths.
It would seem that despite the AHD having being established and used for over forty years and now showing its age, a suitable replacement vertical datum is still somewhere in the future.
Anonymous (undated), John Gibson’s Simpson Desert walk, text of newspaper article that appeared in the Adelaide press in the 1970s and kindly provided by Mrs Lorna Gibson in May 2014.
Balodis, Miervaldis (2006), Gravity Mapping in Australia, 400 Years of Mapping Australia Conference, Darwin August 2006, Paper 10 on CD-ROM.
Bureau of Mineral Resources (1969), The Sedimentary Basins of Australia and Papua New Guinea and the Stratigraphic Occurrence of Hydrocarbons, Bureau of Mineral Resources Record 1969/131.
Dooley JC, McCarthy E, Keating WD, Maddern CA, Williams LW (1961), Pendulum Measurements of Gravity in Australia 1950-51, Bulletin No. 46, Department of National Development, Bureau of Mineral Resources, Geology and Geophysics.
Filmer, M. S. and Featherstone, W. E. (2009), Detecting Spirit-levelling Errors in the AHD: recent findings and issues for any new Australian height datum, Australian Journal of Earth Sciences, Vol. 56, No. 44, pp. 559-569 accessed at : http://dx.doi.org/10.1080/08120090902806305
Fuller, Collin (circa 1980s), A Case Study of a Central Service Agency – the Australian Survey Office, drafted for DAS Annual Report and personal copy supplied by the author.
Janssen V, Commins R and McElroy S (2013), GNSS at NSW tide gauges... of CORS!, Position, 65, pp. 38-42.
Gibson, John (1997), The Simpson Desert AHD Levelling Survey-1970, in Tieline, newsletter of the Institution of Surveyors, Australia, South Australia Division Incorporated, Volume 10, No 8, December 1997. Copy kindly provided by Mrs Lorna Gibson in May 2014.
Gibson, Lorna (2014), Personal communication with Laurie McLean in May 2014.
Gibson, Wayne (2014), Personal communication with Laurie McLean during May-June 2014.
Granger, Harry William (1972), The Australian height datum, Australian Surveyor, Vol. 24, Issue. 4, pp. 228-237.
Intergovernmental Committee on Surveying and Mapping (2014), ICSM Guidelines for Digital Elevation Data accessed at : http://www.icsm.gov.au/elevation/ICSM-GuidelinesDigitalElevationDataV1.pdf
Lambert, Bruce Philip (1989), Personal narratives, National Library of Australia, ORAL TRC 2520, accessed at : http://nla.gov.au/nla.oh-vn547247
Lambert, Bruce P and Leppert, Klaus (1976), A study of the Australian National Levelling survey, Australian Surveyor, June 1976, Vol. 28, No. 2, pp. 89-95.
Leppert, Klaus (1967), Problems encountered in the Use of Third Order Levelling for the National Levelling Grid, Report on Colloquium 'Control for Mapping by Geodetic and Photogrammetric Methods' (Ed. P. V. Angus-Leppan) held at the University of New South Wales, 22-24 May, 1967, accessed at : http://www.gmat.unsw.edu.au/about/school_pubs/pdfmono/mono00.pdf
Leppert, Klaus (1968), The Levelling Survey of Australia, Proceedings, Fifth United Nations Regional Cartographic Conference for Asia and the Far East (E/CONF 52/5), Canberra, March 1967, pp. 114-115, accessed at : http://unstats.un.org/unsd/geoinfo/RCC/docs/rccap5/5th_UNRCCAP_econf.52_5.pdf
Leppert, Klaus (1973), Geodesy in Australia, 1956-72, Technical papers presented at 16th Australian Survey Congress, Canberra, 1973, pp.A1-A6.
Lines, John D (1992), Australia on Paper – The Story of Australian Mapping, Fortune Publications, Box Hill.
Lonsdale, GF (1962), Great Artesian Basin – Reconnaissance Gravity Survey using Helicopters, Queensland 1961, Bureau of Mineral Resources Record 1962/014.
Murray, Alice S and Tracey, Ray M (2001), Best Practice in Gravity Surveying, accessed at : http://www.ga.gov.au/image_cache/GA2236.pdf
O’Donnell, Peter W (2014) Personal Communication, April – May 2014.
Roelse Adrian, Granger Harry W and Graham John W (1975), The Adjustment of the Australian Levelling Survey 1970-71, Technical Report 12, Department of National Development, Division of National Mapping 1975 – 2nd ed.