Skip Navigation | ANU Home | Search ANU
The Australian National University
Research School of Earth Sciences
Earth Physics

Atmospheric Pressure Loading - the effect on geodetic analyses


Atmospheric pressure loading causes an elastic deformation of the solid Earth. The magnitude of the effect can be up to 15-20 mm and it can vary by as much as 10 mm within a 24 hour period. As a result, this physical phenomenon affects the coordinates of geodetic sites (VLBI, GPS, SLR, DORIS etc) and should be accounted for in all high precision geodetic analyses. This web page will provide examples of time series of calculated atmospheric pressure loading deformations as well as GPS vertical time series with and without the loading accounted for.


Atmospheric pressure loading model

The atmospheric pressure loading is calculated by Tonie van dam using NCEP pressure data. The surface pressure is convolved using Green's functions to generate the deformation at each node of a 2.5 x 2.5 deg grid (in local N, E, U at each node). Oceanic regions are treated as an Inverted Barometer. The loading in coastal regions is calculated by subdividing the 2.5 deg grid down to a 2.5 km grid, testing whether each cell is predominently water or land and then convolving using the appropriate Green's function.

The 10-year mean pressure value is subtracted from each 6 hourly value at each node. The variations from the mean pressure are used to calculate the displacements; thus, the displacements are with respect to a 10 year mean pressure value. There is the potential here that a scale factor could be introduced into the geodetic analysis if the mean pressure value yields a systemeatic shift in the ITRF coordinates of a site. We are still investigating this ....

The diurnal and semi-diurmal "atmospheric tides" are first removed from the NCEP pressure data following the procedures of Ponte and Ray (2002). Therefore, the modelled atmospheric pressure loading is only the non-tidal component. The periodic deformation associated with the S1 and S2 tides is added into the analysis of the GPS data (within the GAMIT software module called "MODEL", specifically in subroutine etide.f), resulting in a complete atmospheric pressure loading model. This is exactly the same approach as described in Petrov and Boy (2004).

The atmospheric pressure loading was incorporated into the GAMIT software version 10.2, and binary grid files are available from 1990 to 2004 in the CM, CE and CF reference frames. At present, these grid files are NOT non-tidal files - that is, they do contain deformation associated with partially sampled S1 and S2 tides. There will be an incremental update of the GAMIT software in the next month or so (ie hopefully by the end of July) that will include software and grid files to permit the S1 and S2 tides to be applied fully in conjunction with the application of non-tidal loading, thereby completing the whole model.

ATM Loading time series

Here are some example time series of the vertical displacement caused by atmospheric pressure loading during 2001.
Click on each image to see a larger version.
European sites Australian sites Antarctic sites

The effect on geodetic analyses

Time series of atm loading, observed ground pressure, GPS analysis without and with atm loading applied at the GPS site at Landing Bluff, Antarctica, Kootwijk in Europe and Kerguelen Island in the Southern Ocean.
LDBF time series   KOSG time series   KERG time series


How 7-parameter transformations can affect station height and geocenter motion estimates from space-geodetic observations
Tregoning, P. and T. van Dam
J. Geophys. Res., 110, B03408, dio:10.1029/2004JB003334, 2005.

Atmospheric pressure loading corrections applied to GPS data at the observation level
Tregoning, P. and T. van Dam
Geophys. Res. Lett., 32, L22310, doi:10.1029/2005GL024104, 2005.

GAMIT scripts/code

The atmospheric loading model has been written into the GAMIT GPS software and was released in version 10.2. The grids that Tonie van Dam has provided to date do not have the partial S1 and S2 deformation removed, nor is there code to apply the periodic deformation fully.

This code has now been written and is undergoing final testing by Paul Tregoning at The Australian National University. I am hoping to be able to release this as an incremental update within a couple of months, along with a whole suite of new, non-tidal grid files computed by Tonie.

Return to Top of Page

Return to RSES