EIGEN-GRGS.RL04.MEAN-FIELD is complete to degree and order 300. It is based on GOCE-DIR5 for the part between degree 91 and 300. The time-variable gravity (TVG) coefficients between degrees 1 and 90 are obtained from a regression on the GRGS-RL04-v1 monthly time series of solutions (2002/09 – 2016/06). For degrees 1 and 2 this TVG part is extended to 1993/01-2019/02, included, through the use of a GRGS SLR-only solution based on the data of 5 SLR satellites (Lageos, Lageos-2, Starlette, Stella, Ajisai).

For more details on the RL04 Release, see RL04 description.

Outside of the measurements period (1993/01-2019/02 for degrees 1 and 2, 2002/09-2016/06 for degrees 3 to 90), the gravity field is extrapolated with a mean-slope assumption, i.e. we use the average slope of the signal over the full time period.

The modeling of the TVG part includes:

  • Two annual and two semi-annual coefficients for each year.
  • One bias and one drift for each year. (Note that the reference date of the bias is the beginning of the period of validity of the coefficient, i.e. the beginning of the year in most cases, see format). The bias and drift are in general coherent so that the result is a piece wise linear function, except in the case of earthquakes. Three major earthquakes have been introduced in the modelling: Sumatra on 2004/12/26, Concepcion on 2010/02/27 and Sendai on 2011/03/11.


Since there has been a change in the IERS conventions on February 1st 2018 on the definition of the mean pole model (Section 7.1.4 has been rewritten to include the secular pole and remove the former quadratic mean pole), the EIGEN-GRGS.RL04.MEAN-FIELD is provided in both the old IERS2010 convention (quadratic mean pole) and in the new one (linear mean pole). Only the C(2,1) and S(2,1) coefficients differ between the two versions.


Nota bene: For the computation of the periodic coefficients, see the note at the bottom of the « Introduction tab » page. A software kit in Fortran90 for reading and interpolating the mean model can be found here.


On April 9, 2019, the strategy for extrapolating in the past (before the GRACE era, i.e. before 2002) the coefficients of degree > 2 was changed from a “mean-slope assumption” to a “zero-slope assumption” because some tests on TOPEX altimeter cross-overs showed that the “zero-slope assumption” was leading to smaller Xover residuals. For degrees 1 and 2, the extrapolation is unchanged because the presence of SLR data allows an accurate determination of the their time evolution from 1993.

If you retrieved EIGEN-GRGS.RL04.MEAN-MODEL before April 9, 2019, we recommend that you reload the modified model (available by clicking on the download links above).