1. Introduction
The Space Research Centre in Warsaw is participating in the ESA project “Geodetic SAR for Height System Unification and Sea Level Research”. To observe the absolute sea level and enable the unification of the height systems, the physical heights of the tide gauge stations, referring to a common equipotential surface, e.g., the geoid, are needed [
1].
The geoid is of fundamental importance for geodesy, oceanography, and physics of the solid Earth and was introduced by C.F. Gauss as a refined model of the figure of the Earth [
2]. In geodesy, the geoid/quasigeoid is a reference surface for vertical datum [
3], orthometric heights [
4], a transformation of ellipsoidal heights into orthometric heights [
5], local and regional vertical datum unification [
6], and satellite orbits prediction [
7].
The first gravimetric geoid for Scandinavia and the part of Poland was computed in 1949 [
8]. Due to the small amount of gravimetric data that was achievable at the time, the accuracy of this geoid model was very low.
Due to political changes in Eastern Europe in the 1980s, gravimetric data were released and used to compute the first fairly accurate quasigeoid in Poland. The quasigeoid for the Polish part of the Baltic region was computed in 2006 by W. Jarmołowski [
9]. The last Polish quasigeoid for the Baltic Sea was computed in 2019 [
10]. Several geoid/quasigeoid computations for the Baltic countries and Baltic Sea were performed in the last 20 years, e.g., [
11,
12,
13] and others.
These consecutive geoid models have been progressively improved over the years by implementing a better theory, conducting gravity surveys to fill large data void areas, compiling more precise Digital Terrain Models (DTM) [
14], and embedding more accurate global gravity models derived from dedicated GOCE (gravity field and steady-state ocean circulation explorer) and GRACE (gravity recovery and climate experiment) satellite gravity missions [
15].
The paper aims to calculate an accurate quasigeoid for the Baltic area. The purpose of this work is to calculate the quasigeoid model in the Baltic sea area to explore the possibilities of SAR technology for Baltic height system unification and Baltic sea level research.
2. Materials and Methods
In the classical approximation, the geoidal undulation can be estimated by Stokes’ formula
where
R is the Earth’s radius, Δ
g is the gravity anomaly,
γ; is normal gravity, and
S(
ψ) is Stokes’ function, provided the topography has in some way been removed or shifted inside the geoid.
Gravity is usually measured on the Earth’s surface. To calculate the geoid using Stokes’ formula, the gravity (or gravity anomaly) must refer to the geoid and not to the Earth’s surface. The gravity must be reduced to the geoid. This may be done using one of several available methods. In [
16], reduction methods such as the Bouguer, isostatic, Rudzki, and Helmert methods are mentioned. Additionally, the gravity may be reduced by the residual terrain model (RTM) method [
17].
Thus, the gravity anomaly in Equation (1) has to be reduced to the geoid by applying free-air reduction, assuming there are no topographic masses
These masses are then placed on the surface of the geoid in the form of a layer with a density . This method of placing topographic masses on a geoid is called the Helmert condensation method.
The displacement of the topographic masses changes the gravitational field of the Earth, including the potential of the geoid (
Figure 1). The level surface that possesses the geoid potential after displacement is called the cogeoid: therefore, the equipotential surface computed from Stoke’s integral is the cogeoid instead of the geoid. The vertical distance between the geoid and the cogeoid
Nind can be computed according to [
18] from
which converts the cogeoid into a geoid.
To compute the cogeoid, additional reduction of the gravity value from the geoid to cogeoid is necessary (secondary indirect effect) and here, the free-air reduction is sufficient
This term is usually much smaller than 1 mGal and is often neglected.
To obtain the quasigeoid the geoid undulation,
N minus height anomaly
ζ is needed. This separation is given in [
16] as
where Δ
gB is the Bouguer anomaly, and
γm is the mean normal gravity (9.81 ms
−2).
To get the true value of the geoidal undulation, the integration should be extended to the whole sphere. For local or regional geoid computation, usually, a global geopotential model (GM) is combined with discrete local gravity data as well as height data. The geopotential model provides the low-frequency or long-wavelength part of the geoid. Local gravity anomaly data and height data provide the medium- and high-frequency components of the geoid spectrum, respectively [
19]. The formula for the practical computation of the local geoid by the remove–restore technique is
where
NGM is the geoidal undulation implied by the geopotential model,
is the contribution of the terrain-corrected mean free-air gravity anomalies reduced to the reference field, and
Nind is the indirect effect of the terrain reduction.
The contribution of the
GM coefficients can be computed by the spherical harmonic expansion formulas [
2] given below
where
G is the gravitational constant,
M is the mass of the Earth,
r is a geocentric radius,
a is semimajor of the reference ellipsoid,
Cnm and
Snm are the fully normalized harmonic coefficients of the anomalous potential, and
Pnm is the fully normalized associated Legendre functions,
γ0 is the normal gravity on the ellipsoid, and
N0 results from the difference in the mass of the Earth used in the IERS Convention and GRS80 ellipsoid.
The geoid or quasigeoid is estimated using Stokes’/Molodensky’s formulae with gravity anomalies, Δ
g, as input data. Before applying Stokes’ formula, the gravity anomaly must be reduced by the removing–restoring technique (R – R) [
20].
where the first term on the right-hand side of Equation (8) is the mean free-air gravity anomaly corrected for atmospheric attraction, the second term c is the classical terrain correction, given in linear approximation by [
21]. The third term in Equation (8) is the indirect effect on gravity, which, being very small (usually much smaller than 1 mGal), was neglected, and the fourth term is the reference anomaly computed by Equation (7)
In principle, the residual part of the geoid undulation
can be obtained by evaluation of the Stokes integral by discrete summation [
16]. The summation has to be repeated for every point.
For the computation of large-scale regional geoids, the summation has to be repeated for every point. For the computation of large-scale regional geoids, such as the current Baltic geoid with a 640 × 380 grid, the computational task is too big to be handled efficiently, even on medium-sized or large computers, not to mention microcomputers.
Therefore, the FFT-based numerical methods, which allow for a fast evaluation of discrete convolutions using all the data on the grid, are the only methods that allow for a fast evaluation of discrete convolutions using all the data on the grid and are the only realistic approach to the problem. Currently, several FFT-based techniques are available for the evaluation of the discrete Stokes integral. Two of those, namely the 2D spherical FFT and the multi-band spherical FFT [
22] methods, are in use.
Zero Degree
As long as IAG does not release a new geodetic reference system, GRS80 should be used for the regional geoid/quasigeoid computation, i.e., for the computation of gravity anomalies, disturbing potential, ellipsoidal coordinates, geoid heights, height anomalies, etc. A zero-degree correction has then to be added to reach the latest GM value and the conventional reference value
W0. The zero-degree term for the geoid separation is given by [
2]
where the parameters
GMo and
Uo correspond to the normal gravity field on the surface of the normal ellipsoid. For the GRS80 ellipsoid, we have
GMo = 398,600.5 × 10
9 m
3s
−2 and
Uo= 62,636,860.85 m
2s
−2 [
2]. The Earth’s parameters
GM = 398,600.4415 × 10
9 m
3s
−2 (IERS Conv. 2010) and
Wo = 62,636,858.18 m
2s
−2 [
1] were used in quasigeoid/geoid computation from geopotential. The mean Earth radius R and the mean normal gravity
γ on the reference ellipsoid are taken to be equal to 6,371,008,771 m and 9.798326 m s
−2, respectively (GRS80 values). Based on the above conventional choices, the zero degree term from Equation (10) yields the value
No = −0.665 m, which has been added to the geoid heights obtained from the spherical harmonic coefficients series expansions of all geopotential models.
6. Conclusions
The computation of a regional gravimetric geoid model with proper accuracy is a difficult task that needs special notice and full investigation in all the phases of computations to produce good results.
Using this high-quality data set, we have carried out the gravimetric quasigeoid computation using the Helmert condensation method. This procedure is shown as an efficient method to compute in high resolution the new quasigeoid in the area 52° < λ < 68° and 11° < ϕ < 30° in a grid 1.5’ × 3.0’. Three versions of zero-tide quasigeoid were computed from the gravity data described above and geopotential models: GOCO06, GOCE-DIR6, and EIGEN-6C4.
Twenty-nine GNSS/leveling points of the ASG-EUPOS network were chosen for evaluating the quasigeoid models. According to the analyses carried out, their accuracy is on the level of ±4 cm. For reasons beyond our control, we have access to GNSS/leveling points at the southern end of the tested area, so we suspect that the accuracy of the calculated quasigeoid should be slightly higher about ±2 cm.
The gravity data used in this calculation come from different sources and differ in measurement methods, accuracy, and resolution. In addition, these data refer to different systems, both horizontal and vertical. We also did not have accurate information on gravity data tidal systems and terrain corrections (except in Finland, Sweden, and Poland) introduced into gravity anomalies. It is assumed that these data produced in the last century are in the mean tidal system and contain terrain corrections. Therefore, to calculate the new quasigeoid model in the Baltic, it is necessary to improve the quality of gravity data.
This work uses data only from airborne measurements and measurements at sea performed between 1978 and 1980 by the teams of the Moscow Research Institute. Therefore, the impact of additional data, e.g., sea data from Hakon Mossby or altimetric data on the accuracy of the quasigeoids to be calculated, should be investigated.
Finally, quasigeoid separation at the Władysławowo, Rozewie, and Ustka tide gauge stations was calculated. Differences between each geopotential model are from few millimeters to 1 centimeter. Then, the mean quasigeoid separations at tide gauge stations were compared with the value obtained in the paper [
46], p. 83. Due to different methods in quasigeoid computations and different data sets used, the observed shift of 5 cm means a good agreement between these two solutions.