🌍Explaining datum shifts

Receiving onocoy corrections and your device gives you a different position than you are used to? Find out why.

  • My GNSS receiver provides a shifted position, why is that?

The issue lies within different coordinate frames. Let us dig deeper to see why.

Believe it or not the planet Earth crust is not solid, it moves and it has plasticity characteristics (it elongates and contracts) due to the forces and strains being applied to the different tectonic plates that compose our planet, that is called crustal dynamics.

For practical purposes we will focus on a single example, the EuroAsiatic tectonic plate which the European continent is part of, but the conclusions are applicable to all the other tectonic plates of the planet.

The Euroasiatic tectonic plate in Southern Europe displaces at a rate of 2.6 centimeters every year towards the North East direction. This means that a point in space over Europe changes its location and coordinates by 2.6 centimeters every year when observed from the reference frame (also known as datum) that GPS constellations of satellites use. The reference frame for GPS constellations is known as World Geodetic System of year 1984 (WGS84) G2296 (realization performed in the GPS era week number 2296) or also known as International Terrestrial Reference Frame in its latest realization of year 2020 (ITRF20).

As you can imagine, it is not ideal to measure coordinates in a reference frame that is not tied or locked to a tectonic plate (also known as being no-net-rotation reference frame) because that means that the coordinates of a given point in space located over a tectonic plate change over time. In fact, there are tectonic plates that move at an annual rate of up to 10 centimeters!

  • How can I fix that issue with my offsetted GNSS computed coordinates?

As you can imagine, to create local maps (within a given tectonic plate) and other coordinate based documents it is very convenient to have a reference frame that is tied to the tectonic plate which means that the reference frame would displace along the displacement of the tectonic plate so that the coordinates within that tectonic plate / reference frame would not change any more.

Continuing with our previous example about the European Continent and the Euroasiatic tectonic plate, this is exactly what the politicians and technicians of the European Union did in year 1989. A law was passed so that all the cartography created in the future had to be mandatorily made in the reference frame / datum tied to the Euroasiatic plate, hence the European Terrestrial Reference System of 1989 (ETRS89) was born.

So going back to your problem of offsetted coordinates, if you are within the European Union borders you would want to use, instead of the coordinates in GPS’s native datum (WGS84 G2296 / ITRF20) you have to tell your receiver to transform your coordinates to the more desirable reference frame / datum ETRS89.

Note that in this example we talked about the European case but if you are in the US you have to refer to the North American Datum of 1983 (NAD83), in Australia to the Geocentric Datum of Australia 2020 (GDA2020) and so on.

The specifics on how to configure a GNSS regarding the datum receiver vary from manufacturer to manufacturer and onocoy can not provide technical support to third party manufacturers. Therefore you will have to gather the specifics either from the user manual of the receiver or alternatively from the technical support of that receiver manufacturer.

  • So that means that if I’m in Europe and I express my coordinates in ETRS89 datum they will not move anymore right?

Not quite, while the displacement of the coordinates will not be quite as significant as 2.5 centimeters per year, you are still likely to have small variations of 1 or 2 millimeters per year, you may be wondering, why is this? Didn’t we say that now the reference frame is tied to the tectonic plate? Yes, that is the principle but unfortunately tectonic plates are not solid pieces of crust, they do suffer local deformations due to internal strains causing that different parts of the plate move in slightly different directions and speeds.

In any case, in a 10 year period, the coordinates in Europe would have shifted 25 centimeters when expressed in datum ITRF20 as opposed to 1 centimeter in datum ETRS89.

  • So, in the European case, if I see an offset of around 1 meter between the ITRF20 (default output of the GNSS receivers when using onocoy base stations) and the ETRS89 coordinates (local cartography for instance or non onocoy base stations), that is due to the difference in datum, right?

Right, in Europe the official networks of GNSS base stations are typically set up so that they already report their own coordinates in ETRS89 datum, that means that the RTK rover end user doesn’t even have to tell his/her receiver to shift datums because the base is already configured to provide positions in that ETRS89 destination datum meaning that the rover already outputs positions in the official European datum.

Due to the global nature of onocoy service, the base stations are reporting, through their respective NTRIP streams in RTCMv3 format, in a global datum (ITRF20). This means that you have to either configure the rover receiver to output coordinates in the right datum (ETRS89 in Europe) or provide your own means to convert those ITRF20 coordinates.

  • If the current difference between ITRF20 and ETRS89 datums is around one meter and is increasing by 2.6 centimeters every year, that means that at some point the difference between those two datums was zero, right?

Yes, you are correct, at the moment of ETRS creation, in 1989, that difference was arbitrarily set to zero and has been steadily increasing until today at a rate of 2.6 centimeters per year.

  • For those acquainted in reference frames, aren't you mixing concepts, ETRS is a system whereas ITRF is a realization, why are you doing this?

The reason the author of the present document has chosen to go this way is because GNSS receivers known to the author always output PVT solutions in the latest GPS realization or the latest PPP precise products, currently WGS84 G2296 or ITRF2020 in current epoch. On the other hand, while the European Terrestrial Reference System standardizes the reference frame in Europe to the point that states are legally obligated to provide their cartography and reference stations in its realization from year 2000 (ETRF00), the reference epoch is not standardized. Some countries use 2000.0, some countries use 2017.0, … therefore it is messy to refer to ETRS89 realizations because what is applicable for some countries / institutions is not applicable to others. From Dr. Zuheir Altamimi’s ETRS89 web page at: http://etrs89.ensg.ign.fr/: “(...) The same resolution also recognizes the diverse requirements regarding national implementations of ETRS89, and respects the different countries’ decisions on adopting their preferred ETRS89 realizations including the recommended ETRF2000.”

  • I’m a software developer and I would like to implement my own methods to convert coordinates. Can you point me in the right direction?

Our advice is to use the PROJ libraries in whatever programming language you are going to implement the source code (for instance pyproj for Python).

If you want to dive further into the transformation matter, Dr. Zuheir Altamimi does publish PDF papers every time there is a new ITRF/ETRF realization with the 14 parameters required to perform the transformations. Find a good starting point resource here: http://etrs89.ensg.ign.fr/pub/EUREF-TN-1.pdf

Regarding base station velocities in Europe you can use the closest velocity estimation available at the EUREF Permanent Network website (mind the different datums present in that web page).

Last updated