I recently read an article by David Doyle entitled “Why Doesn’t my Centimeter Match Your Centimeter?”, which painted an interesting portrait of the widespread use of GNSS-based data collection systems. Nowadays, almost everyone can claim to have surveys with centimetre or sub-metre accuracy anywhere in the world, which before was achievable only by a few rare specialists… In many cases, the positions do not always fall at the right place when they are integrated into geographical information systems (GIS). Unfortunately, it is often at this stage that we call on a specialist! The key is in the reference system, also known as datum.
Even if it can become relatively complex when it comes to details, the concept of a reference system when using the GNSS system is essentially identical to conventional surveying techniques. A surveyor always used a reference station for his survey using his theodolite. He stringently maintained his control point network on the territory being surveyed. All his surveys were in reference to his control network (points that are part of a series of controlled polygons). If certain polygons were not well linked to the rest of the network, integration with centimetre precision into the computer-assisted design program (or his geographical information system (GIS)) would be problematic, or even impossible, for the lots linked to the erroneous polygons. And this is exactly what happens with the GNSS system; if the reference is overlooked, GIS integration becomes a real problem!
Without realizing it, GNSS users also use a reference network. If they work in RTK (differential positioning), they use either a VRS (Virtual Reference System) or a station network that sends corrections. In general, these two systems are based on a national reference system (geocentric or not). Users can also use a portable base that they install themselves as needed. When they enter a coordinate into this portable station, they become responsible for the reference system used. During post-processing, the problem remains the same. A position must be entered for the base station and the reference must be known (i.e., the datum).
If they use satellite-based augmentation systems (SBAS), such as the private Omnistar, Veripos or StarFire systems, these systems are generally referenced by a geocentric system (which nowadays includes a temporal drift) defined by international bodies (IGN in France, which maintains, along with several research centres, the International Terrestrial Reference Frame (ITRF)). Today this method of positioning is known as PPP or Precise Point Positioning.
These different GNSS positioning methods therefore do not use the same reference! Each may provide highly accurate coordinates. However, these coordinates are only accurate with regard to their reference. Although this concept may seem very simple, in reality, the increasingly common use of these systems by non-specialists often produces strange results.
Below are a few examples of uses that may lead to confusion with reference systems, in terms of “compared with what?”:
A farmer who carries out micro-topography to analyze his fields is pleasantly surprised by the level of altimetric precision he can achieve (within a few cm), using a “single frequency” receiver. However, when he tries to juxtapose two fields, he may discover an altimetric deviation of up to 2 m. The reason for this is quite simple: he systematically used local base stations with average coordinates taken in the field as the base coordinates. Normally, he uses one base per field. By not taking into account the consistency of the coordinates for the base stations, without realizing it, this farmer was creating independent references for each of his fields. It is obvious that if the analysis had been done individually on each field, he would not have seen any problem. It is when the two juxtaposed fields are integrated that problems arise.
In Quebec, a surveyor uses a Department of Energy and Natural Resources station as a source for RTK differential corrections. This system is referenced to the NAD83 SCRS coordinates system. When he tries to integrate his survey points using the conventional method, which are based on the original NAD83 coordinates system, he will notice inconsistencies of a few to several centimetres. These two coordinate systems (NAD83 ≠ NAD83SCRS) may have inconsistencies of several centimetres from each other.
In Canada, an agronomist surveys the position of trees in the city using a portable GNSS system with an SBAS corrections system. Once the survey has been completed and he integrates the position of the trees into the city’s GIS, these all seem to be off by +/- 1.5 m. The SBAS reference systems are all geocentric (within +/- several cm) compared with the NAD83 SCRS system, which is positioned +/- 1.5 m from the Earth’s centre!
These three examples illustrate the type of error commonly found in the “reference system” category. The issue here is not GNSS system performance, since we are assuming that the positions obtained are accurate versus the chosen reference. Neither does the issue concern the variations between different reference systems, or the methods or precision of transformations between one reference system and another (this topic is very broad).
Lastly, we should find the answer to the question “But compared with what?” in what we commonly call “metadata”. Regarding GNSS position, this is information related to creating positions: the reference system used, statistics regarding the position estimate, date, different dilution of precision (DOP) values, type of GNSS signal used, etc.
For your next delivery, simply mention “Precise, but compared with what?” or deliver your positions along with full metadata!
Source: effigis