Is It Time for a New National Height Datum?

Errors in AHD71

Even when the Australian Height Datum of 1971 was first established, it was known to contain distortions (Roelse et al., 1975, Featherstone 2006). The determination of AHD71 included many inadequate methods, which have caused numerous issues in the datum. The errors can be attributed to a combination of the quality of the, mainly third-order, spirit levelling observations used; the neglect of observed gravity corrections to the spirit-levelling observations; and, most importantly, fixing AHD heights to zero at mean sea level (MSL)  (Featherstone, 2006, p. 1). Whilst these errors have always been present, they have remained largely hidden.

In the 1971 adjustment of the basic levelling, the AHD height was held fixed to zero for mean sea level (MSL) at 30 tide gauges on the mainland, likewise for two tide gauges on Tasmania in 1983. There are several objections to this approach: vertical datums in many overseas countries are established from only one tide gauge; most of the MSL observations used in the AHD were observed over roughly a three-year period that does not properly sample the longest 18.6-year luni-solar tide; and the extra constraints due to unmodelled sea surface topography applied strain to the network adjustment, but this was countered by the (then) desire to have zero height at MSL. This fixing has caused the AHD to become distorted by approximately a metre, mainly in a north-south direction but other distortions exist (Featherstone, 2008, p. 123).

Many of AHD’s problems have remained largely hidden to land surveyors observing relative height differences over small areas with conventional surveying instruments. It is through the introduction of Global Navigation Satellite Systems (GNSS) and gravimetric quasi/geoid models that errors in the Australian Height Datum have come to prominence (Filmer, 2010, p ii.). It was predicted by Featherstone & Kuhn in 2006 that as geoid determination theories and source data continue to improve, notably with the GRACE (Gravity Recovery and Climate Experiment) and GOCE (Gravity field and steady-state Ocean Circulation Explorer) dedicated satellite gravimetry missions, the problems in the AHD will become even more apparent.

North-South Slope in AHD71

There are many sources of error in the AHD71, however perhaps the most notable error is the north-south slope which is evident throughout the datum. There is now compelling evidence for a north-south slope of approximately 1.5m in the AHD  (Featherstone, 2006). The Australian Levelling Survey (ALS) was used to determine AHD71, and in ALS the the tide gauges were held fixed to local Mean Sea Level, the zero-reference for the datum. This process was a contentious decision (Morgan, 1992). It was justified at the time on the basis of allowing the different States and Territories to use local MSL as their zero reference for AHD. The tide gauges were only measured for short periods (2-3 years), and this was not ideal procedure. In order to correctly determine MSL at a coastal tide gauge, regular and uninterrupted measurements are required of the full tidal signature  (Featherstone & Kuhn, 2006). The full tidal signature is dependent on the luni-solar tidal cycle, which is an 18.6 year cycle. By definition, the mean sea level is the average height of the surface of the sea at a tide gauge station for all stages of the tide over this 18.6-year period, usually determined from hourly height readings measured from a fixed pre-determined reference level (Blume, 1975, p. 17). As the tide gauges were only measured over a 2-3 period, they are aliased by the spatially varying tidal effects – that is, the tidal range and tidal frequency content is different at each tide-gauge, thus a limited observation period cannot determine the true MSL at each. This is estimated to cause errors of up to 10cm across the AHD. Equipment errors in the Tide Gauges also is estimated to have caused 10-15cm offsets from true MSL at each tide-gauge (Featherstone & Kuhn, 2006, p. 34).

Dynamic ocean effects were overlooked in the determination of MSL in AHD71, which is a major contributor to the suspected north-south slope in AHD71. Sea Surface Topography is the height of the ocean surface relative to a level of no motion defined by the geoid, and provides information on tides, circulation and the distribution of heat and mass in the Earth’s global ocean (Leben & Hausman, 2013). The effects of Sea Surface Topography on Australia’s coastline effectively causes the cooler, more dense water at the south coastline of Australia to be up to 0.5m lower than the MSL of Australia, and the warmer, less dense water in the north coastline to be up to 0.5m above the MSL. This causes a 1m difference between the adopted mean sea levels at these locations and an equipotential surface, and as a result the datum is suspected to have a north-south slope. The 1.5m north-south slope in AHD71 is still seemingly small for most applications of height; it is significant to some Earth-science related studies, notably high-precision geodesy (Filmer & Featherstone, 2009).

Due to these dynamic ocean effects (e.g. winds, currents, atmospheric pressure, temperature and salinity variations), the short duration of tide gauge observations (spanning, in some cases, a period of only 2-3 years) and the omission of observed gravity, MSL is not coincident with the geoid at these tide gauge locations. This introduced considerable distortions between these two surfaces of up to ~1.5 m into AHD71 across Australia, causing it to be essentially a third-order datum.

Errors in the Australian National Levelling Network

In 1971, the Division of National Mapping, on behalf of the National Mapping Council of Australia, carried out a simultaneous adjustment of 97,320km of the two-way levelling in Australia, holding mean sea level fixed at zero at thirty tide gauges around the mainland coast. The resulting datum surface, with minor modifications in two metropolitan areas, was named the Australian Height Datum of 1971 (AHD71) (Granger, 1972). Figure 1 below shows a map of the Australian National Levelling Network.

The Australian National Levelling Network was initiated in 1945. Work continued for many years after this, and by the end of 1960 a total of about 21,000 kilometres had been levelled. Most of this work consisted of one-way levelling. Whilst this was good progress, there was a massive amount of levelling still required, and so it was deemed that, in order to ensure these surveys were to be of any practical value within a usable time scale, the future surveys would have to be done by contract surveyors using readily available equipment and working to third order standards of accuracy (Morgan, 1992).


Figure 1 - Basic spirit-levelling traverses of the Australian Levelling Survey used to established the AHD in 1971 (Filmer & Featherstone, 2009, from Roelse et al, 1971))


 Although the use of several state authorities in this program ultimately sped up this section of levelling, the different state authorities each used different levelling practices and procedures, and as a result the levelling was of various standards  (Roelse et al., 1971). After much study of the ANLN, it was established that AHD is a homogenous third order network, which is not accurate enough for many applications of height today, or in the future (Morgan, 1992, p. 62).

AHD71 requires replacement or validation of many sections in the levelling network in order to remove suspect data. The levelling data is badly in need of a general adjustment so that the full effects of new and re-levelled sections can be distributed across the datum. Any such re-adjustment must be undertaken with more complete models, especially those that relate to gravity and the form of the geoid (ibid, p. 62).

Figure 2 below shows the quality of the spirit-levelling traverses of the Australian National Levelling Network (ANLN). In this figure, the yellow lines represent sections of first order levelling, the light green is second order levelling, thin purple is third, dark green is fourth order, red is one-way third order and blue is two-way levelling (Filmer & Featherstone, 2009).


Figure 2 - Quality of the spirit-levelling traverses of the ANLN (Filmer & Featherstone, 2009)


 It is evident that a major source of errors in AHD71 is found in those sections which were observed with one-way third order levelling techniques, which are shown as red in figure 2. It can be seen that unfortunately this type of observation is common in the data set.


If the solution to the issues of height datums in Australia is to improve on the current height datum, and not update it, then the Australian National Levelling Network will be a major focal point for the improvements (Filmer, 2010).  The Australian National Levelling Network has numerous problems that will need some attention if it is to be used in the determination of a new national height datum, yet it remains an invaluable resource for any potential new national height datum for Australia. The ANLN could be used as a basis for levelling, with additional levelling required to correct any errors present in the network; however this process presents a risk of degrading any new national height datum by including levelling data of suspect quality. This risk needs to be balanced by the risk of removing too much of the levelling network such that the redundancy of the network is compromised, and benchmark coverage is significantly reduced in some regions (Featherstone & Filmer, 2012).

Thomas Pollard z3323750 | School of Civil and Environmental Engineering | University of New South Wales