Earth Sci Inform (2012) 5:123–136

DOI 10.1007/s12145-012-0102-2

SOFTWARE ARTICLE

GGMCalc a software for calculation of the geoid

undulation and the height anomaly using the iteration
method, and classical gravity anomaly
Siamak Moazezi · Hossein Zomorrodian

Received: 31 October 2011 / Accepted: 25 May 2012 / Published online: 10 June 2012
© Springer-Verlag 2012

Abstract The Global Geopotential Models (GGMs) geodetic parameters such as geoid undulations, height
are very significant because of their usefulness in de- anomalies, gravity disturbances, classical gravity anom-
termination of the parameters like geoidal undulations, alies, deflection of the vertical, gravity gradients, dis-
height anomalies, gravity anomalies, and so on. In this turbance potential gradients, and so on (Rapp 1982;
paper, first, we review the calculation of such parame- Zhu 2007; Barthelms 2009). There are different meth-
ters and then present the GGMCalc software, which ods to determine the geodetic parameters from geopo-
is prepared in Fortran 95 under GNU/Linux Operat- tential models. Rapp (1982) proposed a software in
ing System. This software is capable of using the files Fortran to calculate the geodetic parameters using the
presented by the International Center for Global Earth geopotential models which can be used for low de-
Models (ICGEM) directly without need to apply any gree geopotential models. Tscherning presented the
changes on them. The software is completely structural GRAVSOFT program GEOCOL in Fortran which
and user friendly and could be used conveniently for can determine some geodetic parameters by using the
academic purposes. geopotential models with high degrees (Tscherning
et al. 1992; Nielsen et al. 2012). Casotto and Fantino
Keywords Geopotential model · Iteration method · (2007) evaluated the models for spherical harmonic
Geoid undulation · Height anomaly · Gravity anomaly synthesis of gravitational potential and its gradients
via comparison of four methods. Kiamehr and Eshagh
(2008) present the EGMlab software in Matlab which
Introduction determine the gravity and gradient components from
GGMs through evaluation of fully normalized asso-
Geopotential models are in fact the coefficients of ciated Legendre functions. Pavlis et al. (2008) pro-
the earth’s spherical harmonic expansion of gravity vide HARMONIC_SYNTH_WGS84 Fortran software
(Colombo 1981; Lemonie et al. 1998). Using the geopo- at http://earth-info.nga.mil/GandG/wgs84/gravitymod/
tential models, it is possible to calculate the gravity egm2008/egm08_wgs84.html which is written officially
potential of the earth. Thereby by obtaining the dis- for EGM2008 geopotential model to determine a wide
turbance potential, it is easier to determine different range of geodetic components. On the other hand,
Barthelms (2009) presented a sophisticated method
for determination the geoid undulation and the height
Communicated by: Hassan A. Babaie anomaly, using the iteration method, which was more
accurate than the previous ones. Here we introduce the
S. Moazezi (B) · H. Zomorrodian software, which is able to use directly all the files of the
Department of Geophysics, Science and Research Branch,
Islamic Azad University, Tehran, Iran GGMs that exist in ICGEM without any changes. In the
e-mail: [email protected] next section (Section “Theoretical fundaments”), we
H. Zomorrodian will discuss the fundamental concepts and mathemat-
e-mail: [email protected] ical formula necessary for preparing the software. In
124 Earth Sci Inform (2012) 5:123–136

Section “Design and implementation”, the description define the higher degrees of these coefficients as fol-
of the GGMCalc software is presented as design and lows (Moritz 1980):
implementation. Section “Comparing the results and √
3e2n 4n + 1
computing time of the program” contains the com- C̄2n = (−1)n

parison between the results obtained by the soft- (2n + 1) (2n + 3) (4n + 1)
ware and those obtained from the terrestrial data, √ C̄2
including the comparison of computing time and accu- × 1 − n − 5 5n 2 , (3)
e
racy of two software with the GGMCalc. The conclu-
sion is presented in Section “Conclusions”. The “soft- where e is the first eccentricity of the ellipsoid.
ware files, availability and requirements” section finally In the proposed Fortran program, the real and the
is presented in Section “Software files, availability normal potential functions for the heights above the
and requirements”. ellipsoid were designed. One can use
W(h, ϑ, λ)
 N
GM max
a n
n

Theoretical fundaments = 1+ C̄nm cos mλ + S̄nm sin mλ
r(h) n=2
r(h) m=0
Spherical harmonic expansion of the real 
and the normal potential × P̄nm (cos ϑ) , (4)

Gravity potential in terms of the spherical harmonic

for real potential and
expansion and the Legendre’s function is
GMellipsoid
U(h, ϑ) =
W(r, ϑ, λ) r(h)
 ∞     N /2 
GM  a n
n
  max 2n
= 1+ C̄nm cos mλ + S̄nm sin mλ a
r r m=0 × 1+ C̄2n P̄2n (cos ϑ) , (5)
n=2 n=1
r(h)

for the normal potential; In these latter equations, h is
× P̄nm (cos ϑ) , (1)
the height of the observation point over the ellipsoid,
r(h) and ϑ are as follows (Rapp 1982):
where GM is the geocentric gravitational constant, a is
the mean earth radius, r is the earth radius in obser- r(h) = X2 + Y2 + Z 2
vation point, ϑ and λ are the colatitude and longitude π Z
of the observation point respectively, C̄nm and S̄nm are ϑ= − tan−1 √
2 X + Y2
2
the fully normalized coefficients of the geopotential
models with degree n and order m, and P̄nm is the fully where X, Y, and Z are
normalized associated Legendre’s function.
The fully normalized spherical harmonic expansion X = (N + h) cos φ cos λ
of the normal potential is
Y = (N + h) cos φ sin λ
 ∞  
  
GMellipsoid  a 2n Z = N (1 − e2 ) + h sin φ,
U(r, ϑ) = 1+ C̄2n P̄2n (cos ϑ) ,
r n=1
r in which e is the first eccentricity of the ellipsoid, and
(2) N is the prime vertical radius of curvature:
a
N = .
where a is the semi-major axis of the ellipsoid, r is the 1 − sin2 φ
e2
radius of the ellipsoid in observation point, GMellipsoid
is the geocentric gravitational constant of the ellip- Normal gravity
soid, and C̄2n is the coefficient of the fully normalized
zonal harmonic of the normal potential. Knowing the According to the Bruns formulas ζ = T(h)/γ (h − ζ )
coefficients of the fully normalized zonal harmonics for calculation of the height anomalies and N =
of the normal potential of degree 2, then one can T(N)/γ (0) for calculation of the geoid undulation
Earth Sci Inform (2012) 5:123–136 125

(Hofmann-Wellenhof and Moritz 2006), one can deter- computer calculation, and the time consuming calcula-
mine the normal gravity in different heights over the tion of the whole coefficients, it seems not to be the
ellipsoid by using best approach for a software. Therefore the recursive
method presented by Colombo (1981), and also by
 
2  3 Holmes and Featherstone (2002), and corresponding
γ (h, φ) = γ0 1 − 1 + f + m − 2 f sin2 φ h + 2 h2 ,
a a Clenshaw formulas were applied. These formulas are
accurate enough and need the small amount of com-
(6)
puting time.
where h is the ellipsoidal height in observation point, The recursive formula for calculation of non-
flattening f is sectorial P̄nm (t) for n > m is

a−b P̄nm (t) = anm t P̄n−1,m (t) − b nm P̄n−2,m (t), ∀n > m,

f = ,
a (10)
ω 2 a2 b where,
m= ,
GM
t = cos ϑ
and γ0 is as follows (Hofmann-Wellenhof and Moritz 
2006): (2n − 1) (2n + 1)
anm =
(n − m) (n + m)
aγa cos2 β + b γb sin2 β 
γ0 = , (7)
a2 cos2 β + b 2 sin2 β (2n + 1) (n + m − 1) (n − m − 1)
b nm = , (11)
(n − m) (n + m) (2n − 3)
where a, b , γa and γb are the parameters of the con-
cerned reference ellipsoid, and β is reduced latitude. and the corresponding Clenshaw formula is (Tschern-
ing and Poder 1982; Holmes and Featherstone 2002)
Fully normalized associated Legendre’s function
snmα = an+1,m tqsn+1,m,α − b n+2,m q2 sn+2,m,α + C̄nmα (12)
For determination of the real and normal poten-
where, C̄nm1 = C̄nm , C̄nm2 = S̄nm , and q = (a/r). This
tials, the calculation of the fully normalized associ-
allows the recursive computation of all snmα , constant
ated Legendre’s function or Clenshaw summation is
m, and sequentially decreasing n, from s Nmax ,m,α to smmα ,
necessary. The definition of this function is as follows
where
(Hofmann-Wellenhof and Moritz 2006), for m = 0:
 
Nmax
√ r
qn C̄nmα P̄nm (t)
P̄n,0 (t) = 2n + 1 2−n (−1)k
n=m
k=0 smmα = smα = , (13)
qm P̄mm (t)
(2n − 2k)!
× tn−2k , (8)
k! (n − k)! (n − 2k)! the initial seed for Eq. 12 is s Nmax +1,m,α = s Nmax +2,m,α = 0.
The recursive formula of sectorial P̄nm (t) for n = m
and for m = 0:
is
 
(n − m)! −n  m 2m + 1
P̄nm (t) = 2 (2n + 1) 2 1 − t2 2 P̄mm (t) = u P̄m−1,m−1 (t), ∀n = m, (14)
(n + m)! 2m


r where,
(2n − 2k)!
· (−1)k
tn−m−2k ,
k! (n − k)! (n − m − 2k)! u = sin ϑ
k=0

(9) P̄00 (t) = 1

√
P̄11 (t) = 3 u, (15)
where r is the greatest integer, which is smaller or
equal to (n − m) /2. Although this definition provided and the corresponding Clenshaw formula is (Tschern-
accurate solutions for these functions, but considering ing and Poder 1982):
the fact that the calculation of the factorial functions,
resulted values out of the standard ranges of IEEE for sm = am+1 uqsm+1 + vm , (16)
126 Earth Sci Inform (2012) 5:123–136

where, use the following equation to calculate the undulation

⎧√ iteratively:
⎨ 3 m=1

am =
⎩ 2m + 1 m>1 Ni+1 (ϑ, λ) = Ni (ϑ, λ) +
1
[W(Ni , ϑ, λ) − U 0 ]
2m g(Ni , ϑ, λ)
vm = sm1 cos(mλ) + sm2 sin(mλ), (17) (22)
this allows the recursive computation of all sm , sequen- Replacing the gravity g(Ni ) by normal gravity γ (0)
tially decreasing m, from s Nmax to s0 , where in the Eq. 22, the behavior of the iteration does not
 N change, because each step will be scaled only by a factor

N max max

s0 = qn C̄nm1 P̄nm (t) cos(mλ) of (1 − g/γ ), which is in the order of 10−4 or smaller
m=0 n=m (Barthelms 2009), so one can write:


Nmax
1
+ q C̄nm2 P̄nm (t) sin(mλ) ,
n
(18) Ni+1 (ϑ, λ) = Ni (ϑ, λ) + [W(Ni , ϑ, λ) − U 0 ] .
n=m γ (0, ϑ)

the initial seed for Eq. 16 is s Nmax +1 = 0. For addi- (23)

tional information about the Clenshaw summation, see With i = 0 and N0 = 0 in Eq. 23 one obtains:
Tscherning and Poder (1982).
1
N1 (ϑ, λ) = [W(0, ϑ, λ) − U 0 ] , (24)
Calculation of the geoid undulation γ (0, ϑ)
using the recursive approach which is the starting value of iteration. Finally for
presenting an accurate and precise solution, one can
Barthelms (2009) proposed the following formula for continue the iterative calculation (Eq. 23), as long as
determination the geoid undulation, using the iterative one achieves an arbitrary calculation precision ε, so the
method. Starting with following equation holds:

. ∂ W  Ni − Ni−1 ≤ ε
W(N) − W(Ni ) = (N − Ni ) · , (19) (25)
∂h h=Ni
Figure 1 shows the flowchart of this approach. To
and considering that the partial derivative ∂ W/∂h can achieve the aforementioned accuracy in the ith step of
be represented by the normal component of the gradi- iteration (the last iteration step), we have
ent ∇W, which is the projection onto the normal plumb
line and the fact that the directions of ∇W and ∇U . 2π Gρ H 2 (ϑ, λ)
N(ϑ, λ) = Ni (ϑ, λ) − (26)
nearly coincide, one can write γ (0, ϑ)
  
∂ W . ∇W  Calculation of the height anomaly, using iteration
− = ∇W = |∇W| , (20)
∂h |∇W|  approach
where a | b denotes the scalar product of the vectors a
By using the method similar to that applied by
and b. If b has the unit length, its projection is in direc-
Barthelms (2009) for calculation of the height anomaly
tion of a. Now one can calculate the geoid undulation
ζ , one can write:
approximately by
.
W(ht ) − U(ht − ζi ) = γ (ht − ζi ) · (ζ − ζi ) (27)
. 1
N = Ni + [W(Ni ) − U 0 ] . (21)
g(Ni ) With similar procedure for determination of the geoid
. undulation, we have
The reasons for using “=” instead of “=” are the
linearization in Eq. 19 and the approximation in Eq. 20. 1  
ζi+1 = ζi + W(ht ) − U(ht − ζi ) (28)
It means, if the gravity potential W is known also inside γ (ht − ζi )
the topographic masses, one can use the Eq. 21 to cal-
culate the geoid iteratively for each point with desirable Considering the starting value of the iteration
accuracy. If we have appropriate starting value for the T(ht , ϑ, λ)
iteration, and the iteration converges, one can simply ζ1 (ϑ, λ) = , (29)
γ (ht , ϑ)
Earth Sci Inform (2012) 5:123–136 127

Fig. 1 Flowchart for

determination approach of
the geoid undulation and
the height anomaly

and by applying the true and the normal potentials, case, the geoid undulation N should be formerly deter-
one can continue the iteration procedure as long as one mined. One can calculate the classic gravity anomaly as
achieves an arbitrary precision and obtains follows:
ζi − ζi−1 ≤ ε, (30)
g(ϑ, λ)
where ε is the arbitrary calculation precision for the 
height anomaly. The flowchart of this approach is = [Wu (N, ϑ, λ)]2 + [Wv (N, ϑ, λ)]2 + [Ww (N, ϑ, λ)]2
shown in Fig. 1. 
− [U u (0, ϑ, λ)]2 + [U v (0, ϑ, λ)]2 + [U w (0, ϑ, λ)]2 ,
Calculation of the gravity anomaly
(32)
Gravity anomaly g is the magnitude of downward
continued gravity acceleration |∇W| on the geoid mi-
where Wu , Wv , Ww and U u , U v , U w are respectively the
nus the normal gravity |∇U| on the ellipsoid at the
partial derivative of the true gravity potential and the
same polar distance ϑ and the ellipsoidal longitude λ
partial derivative of the normal gravity in direction of
(Hofmann-Wellenhof and Moritz 2006):
u, v, w (Barthelms 2009).
g(ϑ, λ) = |∇W(N, ϑ, λ)| − |∇U(0, ϑ)| (31) Using following relation:

Having a mathematical statement of the potential 

W(h, ϑ, λ), the calculation of the classic gravity anom- . ∂ |∇U| 
|∇U(0)| = |∇U(N)| − N · , (33)
aly g(ϑ, λ) from the Eq. 31 is straight forward. In any ∂h h=0
128 Earth Sci Inform (2012) 5:123–136

one can write the Eq. 32 as follows: The software is able to use height, bathymetry, and ice
 thickness of each point, separately to apply the correc-
. ∂ |∇U(ϑ)|  tion, by Eq. 26 for calculation of the undulations. The
g(ϑ, λ) = |∇W(N, ϑ, λ)| − |∇U(N, ϑ)| + N ·  .
∂h h=0 output file contains the obtained results. In addition
(34) to the mentioned output data, the program produces
a report containing the used coefficients and the kind
Barthelms (2009) has justified that the downward- of input data (the ellipsoidal or orthometric heights).
continuation in Eq. 34 can be used safely with accept- These will be presented in a file with “.log” extension.
able errors for computation of the gravity potential at The software is completely structural and user friendly
zero level. and is simply utilizable for academic purposes. Input
In the second approach, the gravity disturbance at and output data files with standard ASCII formatting
the surface of the Earth can be used to obtain the are used for this software.
gravity values, then by using these gravity values one The Fortran subroutines and/or Fortran functions
can determine the classic gravity values at zero level as written in this program are as follows.
follows:
∂γ Fortran subroutine to calculate the true potential
g(ϑ, λ) = g(h, ϑ, λ) − (h − N) − γ (0, ϑ, λ), (35)
∂h and its derivatives
where,
According to the Eq. 4 and the Eq. 16 a subroutine for
g(h, ϑ, λ) = δg(h, ϑ, λ) + γ (h, ϑ, λ). (36) determination of the true gravity potential and its deriv-
In the third approach, Molodensky definition can be atives is written. Its input parameters are as follows:
used to obtain gravity anomalies using pre-obtained 1. λ, Geographic latitude of the studying point,
height anomalies as follows: 2. h, Height of the studying point,
g(ϑ, λ) = g(h, ϑ, λ) − γ (h − ζ, ϑ, λ), (37) 3. a, Mean radius of the earth obtained from geopo-
tential model,
4. Nmax , Maximum degrees of the geopotential model,
Design and implementation 5. vm obtained from Eq. 17

Design Fortran subroutine for calculation of the normal

potential and its derivatives
GGMCalc software is prepared in Fortran 95 us-
ing the OpenMP for parallel computations under the According to Eq. 5, a subroutine is written for deter-
GNU/Linux operating system. This software is de- mination of the normal potential and its derivatives
signed to use ellipsoidal parameters such as semi- due to fully normalized spherical harmonics. Its input
major and minor axis, fully normalized zonal harmon- parameters are as follows:
ics of the normal potential and so on, the geopoten- 1. ϑ, Co-latitude of the studying point,
tial models as well as the height of calculating points 2. h, Height of the studying point,
to determine some geodetic parameters without any 3. a, Semi-major axis of the used ellipsoid,
further information such as Bouguer gravity anomaly 4. b , Semi-minor axis of the used ellipsoid,
which may be used in Eq. 38. By using the geopo- 5. Nmax , Maximum degrees of the geopotential model,
tential models presented in the website of Interna- 6. C̄2n , Fully normalized zonal harmonics of the nor-
tional Center for Global Earth Models (ICGEM) at mal potential of even degree, which presented by
http://icgem.gfz-potsdam.de/ICGEM/, the software is the reference ellipsoid.
able to calculate the geoid undulations and the height
anomalies by iteration method, the classical gravity
Fortran function for calculation of the normal gravity
anomalies from the calculated undulations through two
approaches, Molodensky’s gravity anomalies from cal-
Using the Eqs. 6 and 7, a function is written to calculate
culated height anomalies, as well as the gravity distur-
the normal gravity. Its input parameters are as follows:
bance. The software has the ability to use the available
files of GGMs in the ICGEM website directly as an 1. ϑ, Co-latitude of the studying point,
input file without any changes. As another input file 2. h, Height of the studying point,
the coordinates of the calculating points were used. 3. a, Semi-major axis of the used ellipsoid,
Earth Sci Inform (2012) 5:123–136 129

4. b , Semi-minor axis of the used ellipsoid, 3. h, Height of the studying point,

5. γa , Normal gravity at the equator, 4. ε, Accuracy needed for the calculation.
6. γb , Normal gravity at the pole,
7. ω, Angular velocity of the earth.
Fortran function for calculation the classical gravity
Fortran subroutine for calculation of the fully anomaly
normalized associated Legendre’s function
and its derivative According to the Eqs. 34, 35 and 37, three functions are
prepared to calculate the gravity anomaly. Their input
Using the Eqs. 10, 11, 14 and 15 a subroutine is pre- parameters are as follows:
pared for calculation of the fully normalized associated
Legendre’s function and its derivative. Its input para- 1. ϑ, Co-latitude of the studying point,
meters are as follows: 2. λ, Geographic latitude of the studying point,
3. h, Height of the studying point.
1. n, Degrees of the fully normalized associated
4. N, Undulation (or ζ , height anomaly for Eq. 37)
Legendre’s function,
2. m, Orders of the fully normalized associated
Legendre’s function, Implementation
3. Nmax , Maximum degrees of the geopotential model,
4. ϑ, Co-latitudes of the studying point. For running this software one should use three different
input data files. The first one is the GGM file
Fortran subroutine for calculation of Clenshaw “coeff.dat”. The second one is “input.dat” file, which
summation of the fully normalized associated contains the coordinates of the input data points (lati-
Legendre’s function for constant order m tude, longitude, and hight). Finally the ellipsoidal para-
meters are in the third file “ellipsoid.dat”. By execution
Using the Eq. 12, a subroutine is prepared for calcula- of the software, the program reads “ellipsoid.dat” para-
tion of the Eq. 13. Its input parameters are as follows: meters first, then the data of the “coeff.dat” file, and
1. m, Orders of the fully normalized associated finally the “input.dat” file. In this stage the software
Legendre’s function, asks some questions as follows:
2. Nmax , Maximum degrees of the geopotential model,
3. ϑ, Co-latitudes of the studying point. 1. Is Height of Points Ellipsoidal(T) or Orthomet-
4. q = (a/r) ric(F)
5. C̄nm and S̄nm , the coefficient of the fully normalized 2. Are Topography, Bathymetry, and Ice Thickness
spherical harmonics of the model. Available Separately (T/F)
3. Please Enter Production Name
Fortran function for calculation of the geoid undulation 4. Please Enter Required Iterations for Calculation of
the Undulations and the Height Anomalies;
Using the Eqs. 23 to 26, a function is written to calcu- The software asks two additional questions, if one
late the geoid undulations. Its input parameters are as entered zero value as the required iteration:
follows:
(a) Please Enter Required Precision for Calcula-
1. ϑ, Co-latitude of the studying point, tion of the Undulations
2. λ, Geographic latitude of the studying point, (b) Please Enter Required Precision for Calcula-
3. h, Height of the studying point, tion of the Height Anomalies
4. ρ, Assumed density for the studying point,
Then the software begins to calculate the geodetic
5. ε, Accuracy needed for the calculation.
parameters with the desired precisions. When the cal-
culation is complete, the software writes the calcu-
Fortran function for calculation of the height anomaly
lated values for the undulations, the height anom-
alies, the gravity disturbances, the gravity anomalies
Using the Eqs. 28 to 30, a function is written to calculate
by the Eq. 34, the gravity anomalies by the Eq. 35,
the height anomaly. Its input parameters are as follows:
and the Molodensky gravity anomalies along with the
1. ϑ, Co-latitude of the studying point, coordinates and the heights of input points in an out-
2. λ, Geographic latitude of the studying point, put file “output_[GGM name]_[production name].dat”
130 Earth Sci Inform (2012) 5:123–136

Table 1 The degree and order of used models

with the more recent geopotential models like AIUB-
Model name Degree and order GRACE03S (Beutler et al. 2011), EIGEN-51C
AIUB-GRACE03S 160 (Bruinsma et al. 2010), GOCO02s (Goiginger et al.
GOCO02s 250
2011) were selected to be compared with the terrestrial
EIGEN-51C 359
EGM96 360
data. Furthermore, the EGM2008 model is used to
EGM2008 2,190 compare GGMCalc software with GEOCOL, and
HARMONIC_SYNTH_WGS84, cause of its highest
of degree and order among the models up to now,
for precision and computing time of calculation of
undulations. Table 1 summarized the degree and order
(e.g. “output_EGM96_test01.dat”). At last, another of used models in this investigation.
file is created by the software with “.log” exten-
sion (“output_[GGM name]_[production name].log”),
which contains the screen outputs of the software such Comparing of the gravity anomalies
as kind of the heights, total number of data points,
ellipsoid and GGM parameters, which are used for Gravity anomalies were calculated by using the terres-
calculations. trial gravity data obtained from the archive of the Bu-
reau Gravimetrique International (BGI) for this region
(Fig. 2), and compared with the results extracted from
Comparing the results and the computing time the written program.
of the program Table 2 shows statistical comparison between three
calculation methods of the gravity anomalies with the
Moazezi (2010) showed that among five geo- obtained gravity anomalies from terrestrial data. Ac-
potential models: EGM2008 (Pavlis et al. 2008), cording to Table 2, the Eq. 34 can be used safely
EGM2008upto360, EGM96 (Lemonie et al. 1998), to calculate the gravity anomalies through downward-
EIGEN-5C (Förste et al. 2008), and GGM03C (Tapley continuation.
et al. 2007), the EGM96 model proved to have the Figure 3 shows the results of GGMCalc for the
best fit with the terrestrial data in the territory of gravity anomalies and Fig. 4 shows the differences be-
Iran. In current research the EGM96 model together tween the gravity anomalies calculated by the program

Fig. 2 Gravity anomaly

values from the BGI
Earth Sci Inform (2012) 5:123–136 131

Table 2 Statistical comparison between three calculation method of the gravity anomalies with the obtained gravity anomalies from
terrestrial data (values in mGal)
Minimum Maximum Average Standard deviation
Classic gravity anomaly (Eq. 34) −153.528 187.625 17.943 28.013
Classic gravity anomaly (Eq. 35) −153.569 183.411 17.552 27.882
Molodensky gravity anomaly −154.413 183.354 17.393 27.907

Fig. 3 Gravity anomaly

values from the EGM96
geopotential model
determined by GGMCalc

Fig. 4 Differences between

gravity anomaly values from
the EGM96 geopotential
model and the terrestrial data
132 Earth Sci Inform (2012) 5:123–136

Table 3 Statistical comparison between the results of the gravity anomalies obtained from the geopotential models and those of the
terrestrial data (values in mGal)
Model name Minimum Maximum Average Standard deviation
EGM96 −187.733 153.6878 −18.0278 28.02054
AIUB-GRACE03S −193.880 178.9482 −18.8012 34.28408
EIGEN-51C −234.753 144.3773 −17.8473 29.41016
GOCO02s −206.034 156.6280 −17.9557 32.00708

Fig. 5 Undulation values

from the EGM96
geopotential model
determined by GGMCalc

Fig. 6 Differences between

the geoid undulations
obtained from EGM96
geopotential model and those
of the terrestrial data
Earth Sci Inform (2012) 5:123–136 133

Table 4 Statistical comparison between the results of the geoid undulations obtained from the geopotential models and those of the
terrestrial data (values in m)
Model name Minimum Maximum Average Standard deviation
EGM96 0.920843 8.857550 3.835147 2.075677
AIUB-GRACE03S −1.754430 9.365241 3.59820 2.213486
EIGEN-51C −1.173080 8.863399 3.635566 2.106982
GOCO02s −1.093840 9.125303 3.570279 2.198840

Fig. 7 Height anomaly values

from the EGM96
geopotential model
determined by GGMCalc

Fig. 8 Differences between

the height anomalies
obtained from EGM96
geopotential model and those
of the terrestrial data
134 Earth Sci Inform (2012) 5:123–136

Table 5 Statistical comparison between the results of the height anomalies obtained from the geopotential models and those of the
terrestrial data (values in m)
Model name Minimum Maximum Average Standard deviation
EGM96 0.812444 8.516727 3.602485 2.079347
AIUB-GRACE03S −1.845350 8.839260 3.364855 2.185882
EIGEN-51C −1.047610 8.512602 3.404861 2.106583
GOCO02s −1.196490 8.622317 3.338968 2.176817

(for instance, EGM96 model) and the results obtained gram and those from terrestrial measurements intro-
from the terrestrial data. Table 3 shows the maximum, duced in Table 4.
minimum, and the mean values as well as the stan-
dard deviations of the differences between the gravity
Comparing of the height anomalies
anomalies obtained from the program and those from
terrestrial measurements.
Height anomalies for the stations of the precise level-
ing network of the country are calculated as follows
Comparing of the undulations (Sjöberg 2010):

In case of the geoid undulations, we compared the g BO

p VgT − V PT δ ḡ BO − δg BO
ζ = N− H− − P
. (38)
results of the program with the values calculated by γ̄ γ̄ γ̄
Ardalan and Grafarend using the Remove-Restore
method and based on the inverse solution of the Abel- where, ζ and N are the height anomaly and the geoidal
Poisson Integral (Ardalan and Grafarend 2004). undulation, respectively, H is the orthometric height,
Figure 5 shows the results of GGMCalc for the geoid γ̄ is the mean normal gravity between the reference
undulations and Fig. 6 shows the differences between ellipsoid and telluroid, g BO
p and δg BO are the refined
the geoid undulations calculated by the program for Bouguer gravity anomaly and gravity disturbance, VgT
EGM96 model and the results obtained from the terres- and V PT are the topographic potentials at geoid and
trial data. The maximum, minimum, and the mean val- surface points, respectively, and δ ḡ BO is the mean value
ues as well as the standard deviations of the differences of the gravity disturbance between sea level and the
between the geoid undulations obtained from the pro- height of the computation point.

Table 6 Precision and computing time of obtained undulations by GGMCalc, GEOCOL, and HARMONIC_SYNTH_WGS84
software for the EGM2008 model
CPU Software name Time (sec) Minimum Maximum Average Std. dev.
Intel® Core™2Due T9300 2.5GHz GGMCalc (Precision = 1 mm) 1196.351 −1.842 8.283 2.854 2.107
GGMCalc (Iteration = 1) 411.209 −1.868 8.284 2.860 2.119
GGMCalc (Iteration = 2) 805.717 −1.842 8.283 2.854 2.107
GGMCalc (Iteration = 3) 1200.685 −1.842 8.283 2.854 2.107
GEOCOL 380.694 −1.628 8.282 2.831 2.121
H_S_WGS84 (GCC) 232.661 −1.104 8.682 3.286 2.112
H_S_WGS84 (Original) 2268.140 −22.511 24.946 1.119 9.958
Intel® Core™ i7–2600 K 3.4 GHz GGMCalc (Precision = 1 mm) 419.098 −1.842 8.283 2.854 2.107
GGMCalc (Iteration = 1) 145.903 −1.868 8.284 2.860 2.119
GGMCalc (Iteration = 2) 283.613 −1.842 8.283 2.854 2.107
GGMCalc (Iteration = 3) 421.353 −1.842 8.283 2.854 2.107
GEOCOL 431.286 −1.628 8.282 2.831 2.121
H_S_WGS84 (GCC) 139.981 −1.104 8.682 3.286 2.112
H_S_WGS84 (Original) 129 8.287 −22.511 24.946 1.119 9.958
Earth Sci Inform (2012) 5:123–136 135

Figure 7 shows the results of GGMCalc for the is the advantage and innovation of this program. The
height anomalies and Fig. 8 shows the differences computing time of the GGMCalc with three iterations
between the height anomalies calculated by the pro- for undulation showed that the software has a good
gram for EGM96 model and the results obtained from efficiency. Furthermore, the software is structural and
the terrestrial data. The maximum, minimum, and the user friendly, and can be simply applied for academic
mean values as well as the standard deviations of purposes.
the differences between the height anomalies obtained
from the program and those from terrestrial measure-
ments are introduced in Table 5.
Software files, availability and requirements
Comparing of the computing time
GGMCalc is a non-commercial software under the
GNU Public License, version 3 (GPL v.3) for academic
Table 6 shows computing time and precision of calcula-
purposes. All software files including sample input files
tion of obtained undulations by GGMCalc, GEOCOL,
are available at http://sourceforge.net/projects/xgravity.
and HARMONIC_SYNTH_WGS84 software for the
Source codes can be compiled by using gfortran com-
EGM2008 model. The precision of each software is
piler as a part of GCC under GNU/Linux. At least, a
shown by statistical comparison between the results
1GHz dual-core CPU is recommended for implemen-
of the software and those of the terrestrial data. In
tation. Needed memory depends on degree and order
this comparison, GGMCalc is running in four condi-
of the used GGM as well as the number of calculating
tions such as with 1mm precision, 1 iteration, 2, and
points, but in general, 512MB RAM is recommended
3 iterations. These three software are compiled by
for usual usages.
GCC in 64-bit GNU/Linux. Furthermore, the origi-
nal version of HARMONIC_SYNTH_WGS84 which
is presented in its official site is used in comparison. Acknowledgements We would like to express our apprecia-
The comparisons are performed through running of tion to Prof. Carl Christian Tscherning for kindly providing the
GRAVSOFT package for us. We are indebted to Prof. Hassan
each three mentioned software on two kinds of CPU A. Babaie for significant helps and three reviewers for valuable
(Intel® Core™2Due T9300 2.5GHz with 4GB RAM suggestions and comments to improve of this paper. We are
and Intel® Core™ i7-2600K 3.4GHz with 8GB RAM). grateful to the Bureau Gravimetrique International (BGI) for
providing the gravity data used in this study. Finally, we are
appreciative to Mrs. Farahnaz Moazezi for the English review of
this paper.
Conclusions

In current research, we first reviewed the calculation

of the geoid undulations and the height anomalies References
by the iteration method and also the gravity anomaly
through three approaches, using the geopotential mod- Ardalan AA, Grafarend EW (2004) High-resolution regional
geoid computation without applying Stokes’s formula case
els. A Fortran 95 program under GNU/Linux has then
study: high resolution geoid of Iran. J Geod 78:138–156
been presented for calculation of the aforementioned Barthelms F (2009) Definition of functional of the geopoten-
parameters. This program is able to use the models tial and their calculation from spherical harmonic models.
like EGM2008 up to degrees and orders of 2,190, and Scientific Technical Report STR09/02, Helmholtz Zentrum,
Potsdam
the latitudes from 90N to 90S. The software is tested Beutler G, Jäggi A, Meyer U, Prange L, Bock H, Dach R,
by using the EGM96 and four other more recently Mervart L (2011) Gravity Field Determination at AIUB:
proposed geopotential models. The output data of the From CHAMP and GRACE to GOCE. Presented at
software were compared with the results obtained by the 2011 General Assembly of the European Geosciences
Union, Vienna, Austria
the terrestrial data. The output results of GGMCalc in
Bruinsma SL, Marty JC, Balmino G, Biancale R, Foerste C,
comparison to the terrestrial components and the other Abrikosov O, Neumayer H (2010) GOCE gravity field re-
mentioned software showed to be somehow acceptable. covery by means of the direct numerical method. Presented
The proposed program can use the GGM files already at the ESA Living Planet Symposium 2010, Bergen, June 27–
July 2, Bergen, Noway
available in ICGEM website without any changes to
Casotto S, Fantino E (2007) Evaluation of methods for spheri-
determine the geoid undulations, height anomalies, cal harmonic synthesis of the gravitational potential and its
and the classical gravity anomalies accurately, which gradients. Adv Space Res 40(1):69–75
136 Earth Sci Inform (2012) 5:123–136

Colombo O (1981) Numerical methods for harmonic analysis on measurements to present an optimum geopotential model
the sphere, Report no 310. Dept. of Geodetic Science and for Iran. M.Sc. Thesis, Science and Research Branch, Islamic
Surveying, The Ohio State University Azad University, Tehran, Iran, p. 131
Förste C, Flechtner F, Schmidt R, Stubenvoll R, Rothacher Moritz H (1980) Geodetic reference system 1980. J Geod
M, Kusche J, Neumayer KH, Biancale R, Lemoine JM, 54(3):395–405
Barthelmes F, Bruinsma J, Koenig R, Meyer U (2008) Nielsen J, Tscherning CC, Jansson TRN, Forsberg R (2012)
EIGEN-GL05C—a new global combined high-resolution Development and User Testing of a Python Interface to
GRACE-based gravity field model of the GFZ-GRGS the GRAVSOFT Gravity Field Programs. In: Geodesy for
cooperration. Genral Assembln European Geosciences Planet Earth: Proceedings of the 2009 IAG Symposium,
Union Buenos Aires, Argentina, 31 August–4 September 2009,
Goiginger H, Höck E, Rieser D, Mayer-Gürr T, Maier A, Krauss pp 443–449. doi:10.1007/978-3-642-20338-1_53
S, Pail R, Fecher T, Gruber T, Brockmann JM, Krasbutter Pavlis NK, Holmes SA, Kenyon SC, Factor JK (2008) An earth
I, Schuh W-D, Jäggi A, Prange L, Hausleitner W, Baur O, gravitational model to degree 2160: EGM2008. Presented
Kusche J (2011) The combined satellite-only global gravity at the 2008 General Assmbly of the European Geosciences
field model GOCO02S. Presented at General Assembly of Union, Vienna, Austria
the European Geosciences Union, Vienna, Austria Rapp RH (1982) A fortran program for the computaion of gravi-
Hofmann-Wellenhof B, Moritz H (2006) Physical geodesy, 2nd metric quantities for high degree spherical harmonic expan-
Edition. Springer-Verlag, Wien, New York, p 403 sion, Report no 334. Dept. of Geodetic Science and Survey-
Holmes SA, Featherstone WE (2002) A unified approach to the ing, The Ohio State University, Columbus
Clenshaw summation and the recursive computation of very Sjöberg LE (2010) A strict formula for geoid-to-quasigeoid sepa-
high degree and order normalised associated legendre func- ration. J Geod 84:699–702
tions. J Geod 76(5):279–299 Tapley B, Ries J, Bettadpur S, Chambers D, Cheng M, Condi
Kiamehr, R., Eshagh, M. (2008) EGMlab, a scientific software F, Poole S (2007) The GGM03 mean earth gravity model
for determining the gravity and gradient components from from GRACE. Eos Trans. AGU, 88(52), Fall Meet. Suppl.,
global geopotential models. Earth Sci Inform 1:93–103 Abstract G42A-03.
Lemonie FG, Kenyon SC, Factor JK, Trimmer RG, Pavlis NK, Tscherning CC, Poder K (1982) Some geodetic applications of
Chinn DS, Cox CM, Klosko SM, Luthcke SB, Torrence clenshaw summation. Boll Geod Sci Affini 4:351–364
MH, Wang YM, Williamson RG, Pavlis EC, Rapp RH, Tscherning CC, Forsberg R, Knudsen P (1992) The GRAVSOFT
Olson TR (1998) The development of the joint NASA package for geoid determination. In: Proc. 1. continental
GSFC and the National Imagery and Mapping Agency workshop on the geoid in Europe, Prague, May 1992. Re-
(NIMA) geopotential model EGM96, NASA Technical search Institute of Geodesy, Topography and Cartography,
Paper NASA/TP1998206861. Goddard Space Flight Center, Prague, pp 327–334
Greenbelt, USA Zhu L (2007) Gradient modelling with gravity and DEM, Report
Moazezi S (2010) Comparison the components of five global no 483. Dept. of Geodetic Science and Surveying, The Ohio
geopotential models with those of the terrestrial gravity State University, Columbus