Determination of A New Gravimetric Geoid Modelling

IOP Conference Series: Earth and Environmental Science
PAPER • OPEN ACCESS
Determination of a new gravimetric geoid modelling for Sudan using the

least-squares collocation technique
To cite this article: Anas Sharafeldin Mohamed Osman and Ira Mutiara Anjasmara 2023 IOP Conf. Ser.: Earth Environ. Sci. 1127 012014
View the article online for updates and enhancements.
This content was downloaded from IP address 216.19.203.64 on 20/01/2023 at 12:38

Geomatics International Conference 2022 IOP Publishing
IOP Conf. Series: Earth and Environmental Science 1127 (2023) 012014 doi:10.1088/1755-1315/1127/1/012014
Determination of a new gravimetric geoid modelling for

Sudan using the least-squares collocation technique
Anas Sharafeldin Mohamed Osman1,2 and Ira Mutiara Anjasmara1, *

1
Department of Geomatics Engineering, Institut Teknologi Sepuluh Nopember,
Sukolilo, Kampus ITS Surabaya 60111, Surabaya, Indonesia.
2
Department of Surveying Engineering, Faculty of Engineering, Omdurman Islamic
University, Khartoum, Sudan
*E-mail : [email protected]
Abstract. The main purpose of this study is to compute a new gravimetric geoid model for
Sudan by using the least-square collocation technique (LSC method) and applying the remove-
compute-restore (RCR) technique. The computation of the model contains different datasets
which are the gravity contribution of the model GO_CONS_GCF_2_TIM_R6e degree/order
300, BGI free-air gravity dataset in Sudan, GPS/levelling data, and high-resolution topographic
information from ASTER digital elevation model. The “residual gravity anomalies” were run
through the GEOCOL program using the GRAVSOFT software package, and the effects were
restored to calculate the quasi-geoid surface (height anomalies). The gravimetric geoid was
computed by adding the (𝑁 − 𝜁ℎ ) separation term to the quasi-geoid and was fitted to the GPS
and levelling data provided by Sudan. The accuracy of our gravimetric geoid model SDN-
LSC-G22 of the area of Sudan and some areas of bordering countries has been investigated by
using geoid undulations computed from GPS and levelling data and by investigating the
differences between the geoids of the GGM models which are EGM2008 and SGG-UGM-2.
Our gravimetric geoid model (SDN-LSC-G22) has indicated an accuracy of 17.4 cm, in terms
of a standard deviation compared with 66 GPS and Leveling data distributed in the area of
Khartoum (most of these points are control points and benchmarks). Also, we evaluated our
gravimetric geoid model by using 19 points distributed in the area of Sudan, and they indicated
a standard deviation of 51.3 cm. The overall accuracy of SDN-LSC-G22 compared with the
geoid undulation of all GPS and levelling has indicated an STD of 34.1 cm. The model SDN-
LSC-G22 has shown better accuracy and significant differences compared with the GGM
models EGM2008 and SGG-UGM-2 in terms of the differences with the available GPS and
levelling data which have shown ~17 cm differences using (Abdalla, 2009) GPS and levelling
data. It has demonstrated STD of ~17 cm differences by using 66 GPS and levelling data.
Therefore, the model SDN-LSC-G22 provided better improvements and reliable geoid heights
over Sudan compared to EGM2008 and SGG-UGM-2 gravity field models.
Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
1. Introduction
The previous studies of Sudan geoid modelling used SRTM30 in their computations, while several
digital elevation models are available such as ASTER and GTOPO30. Evaluation of the elevation
contribution of several types of DEM in a region is quite essential in geoid modelling computation and
research. A precise local geoid model for Sudan is essential and widely needed for a wide range of
geodetic, surveying, and mapping applications. Even though various geoid models have been
developed in recent decades, their precision levels are still far from the civil engineering community's
high-accuracy expectations. This study discusses in detail the strategy of the processing and the results
in determining a gravimetric geoid modelling for Sudan (see Figure 3) to find the flowchart of the
applied (RCR) strategy in this work. In the geoid modelling computations, we have used the terrestrial
gravity dataset distributed in the country areas (see Figure 1 (a)). The geoid is a continuous, closed,
and not a regular surface, so it is unsuitable for mathematical computations. The geoid surface might
be approximated by an analytical surface like the ellipsoid [1]. Gravity measurements play an essential
role in precise geoid determination.
The method of measuring the Earth's gravity field is based on the fundamental acceleration
quantities and time by some methods such as Pendulum Method, Free-Fall methods, and gravimeters.
Though gravity data is collected from many parts of the world by traditional means, they vary
significantly in quality. They are also incomplete (i.e. heterogeneous), including several data gaps in
some areas such as mountainous and oceanic regions. Therefore, a demand for gravity observations
from space allows the geodetic community to determine a global map of the Earth’s gravity field. In
geodesy, precise geoid determination has been a major study topic. Accurate geoid models can be used
in a variety of engineering and scientific applications, such as orthometric heights vertical datum,
which allow ellipsoidal heights derived from GPS data to be converted into orthometric heights
(equation 1), which are related to the Earth's gravity field, without the requirement for time-consuming
and costly spirit-levelling [1], a levelling reference surface, Consolidation of regional and local
vertical datums, investigations of the ocean and the interior of the Earth, locating ore, gas, and oil
deposits, and hydrographic surveying and marine navigation [2]. Global Geopotential Models (GGMs)
can be used as a reference to help develop more detailed regional/local geoids, or they can be used to
provide geoid heights on their own. GGMs are obtained from satellite gravity data and a combination
of satellite model and gravity measurements, terrestrial gravimetric, altimeter-derived gravity data in
marine areas, and, more recently, airborne gravimetric. Over the last two decades, several attempts and
research have been conducted for geoid modelling for the territory of Sudan.
The first geoid modelling computation attempt in Sudan was made by [3] from Cornell University
in the USA; He computed the vertical deflections and the separation between the geoid and the
reference ellipsoid Clarke 1880 by using 46 Astro-geodetic sites. He found that the information was
insufficient to calculate accurate geoid unless more measurements were carried out over the un-
surveyed areas due to a lack of data from neighbouring countries and extensive un-surveyed regions of
the Northwest and Southwest parts of Sudan. [4,5] used astrogeodetic and satellite approaches to
determine the geoid shape in Sudan. The dataset for the astrogeodetic approach was obtained in 1937,
while the satellite methods utilized data from observations with a Doppler receiver conducted in 1975,
1978, and 1981. [6] utilized Transit Satellite Doppler observations collected over a decade (1975–
1986) to determine the geoid's shape in Sudan. Salih has concluded that the geoid in Sudan increases
from west to east and north to south in relation to the local (Adindan) datum, which is based on the
Clarke 1880 Spheroid; for more detail, you can read [6]. The following study was developed by [7]; a
gravimetric geoid model was created using terrestrial gravity data, topography data, modified Stokes
kernel, and the GEM-T1 in [7]. The gravimetric geoid heights were obtained at the same stations as
Doppler geoid heights. To verify the accuracy of the geoid model, a comparison was made between
Doppler geoid heights and gravimetric geoid heights, yielding the following statistics (mean = 0.43 m,
standard deviation = 1.88 m). [7] concludes that for a more accurate geoid model in Sudan,
densification of regional gravity data should be increased to at least 144 point values within an area of
(11) arc-deg.
2
ℎ =𝑁+𝐻 (1)
Where h is the ellipsoidal height, H is the orthometric height, and N is the undulation height (geoid
undulation).
2. Data Collection
This study uses several types of datasets for gravimetric geoid modelling computations which is
consists of land and sea free-air gravity anomalies, GPS/levelling data collected over the Sudan region,
TIM_R6e data, and high-resolution topographic data collected from the ASTER. The grid specification
of our study area is (2.5 ° < 𝜙 < 23.7°, 20.5° < 𝜆 < 40°) which is covered Sudan, South Sudan, and
some areas of the bordering countries. The land gravity data and see gravity data for this study have
been provided by International Gravimetric Bureau (BGI), the distribution of the free-air gravity
dataset is not homogeneous over Sudan, with significant gaps (see Figure 1 (a)). The shortage of
terrestrial observations can cause a significant limitation on the final geoid model accuracy, and it can
affect the accuracy of GGMs evaluation in order to know which model is more reliable in the region
nevertheless the accuracy of the applied computational method of geoid determination. Gravity
measurements have begun in Sudan in the second half of the last century [8]. Sudan had only a few
gravity observations before that period (except the Red Sea) [8]. Robert Research International (RRI)
carried out a major regional gravity survey of northeast Sudan, a remote desert region, with the help of
the Geological Research Authority of Sudan (GRAS) as part of a petroleum promotion project. The
total of the land gravity dataset in Sudan and a few points distributed in some parts of the neighboring
countries are 16564 points, and 14442 points are sea gravity datasets in the area of the red sea; both
have been combined. Therefore, the total gravity dataset in this study is 31006 points. The available
GPS/levelling dataset in this study has been taken from several references. The first reference was the
study [8]; Abdalla's dataset contains 19 points distributed in the area of Sudan, and the levelling data for
these stations have been derived from old vertical geodetic controls referred to as first, second and
third-order (orthometric heights based on normal gravity). At these stations, the accuracy of
GPS/levelling geoid heightsis estimated to be between 0.1 and 0.5 m [8]. It must be noted that the error
caused by using the normal orthometric height rather than the orthometric height has not been
considered. The GPS survey was done in 12 h observing sessions in the framework of multiple
commercial geodetic projects between 2005 and 2008. Ellipsoidal heights of those stations were
acquired from the GPS survey [8]. Also, there are some points have been collected from several
control points distributed in the city of Khartoum, some of these points were measured in the area of
Khartoumby the Ministry of Physical planning, which is the most reliable authority in Sudan; we have
added this data also in our dataset, so the total of the GPS/levelling data is going to be 85 points (see
Figure 1 (b)). The Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) data
have been used in this study. The data of the ASTER has been downloaded from NASA earth data
https://search.earthdata.nasa.gov/search?fi=ASTER. Figure 2 below shows the DEM ASTER model
for Sudan. The ASTER mission's primary scientific goal is to understand the local and regional-scale
processes that occur on or near the Earth's surface and lower atmosphere, particularly surface-
atmosphere interactions [9].
3
(a)
(b)
Figure 1. (a) Distribution of the gravity anomaly BGI dataset in Sudan and neighbouring
countries. And (b) Available GPS/levelling data distributed in Sudan.
4
Figure 2. Aster digital elevation model over Sudan.
3. Methodology
Least Square Collocation is most commonly used in geodesy for geoid determination, interpolation,
and spherical harmonic coefficients [10]. Collocation is useful for predicting and filtering gravity
anomalies from reduced gravity observations; the term "reduced" here refers to removing normal
gravity [11]. The method used to calculate the geoid in Sudan is the remove- compute-restore (RCR)
methodology. In this procedure, the long and short wavelength components of the gravity field
functional were computed from a TIM_R6e, and very short wavelengths of the gravity field functional
were calculated from a high-resolution digital terrain model. The steps are well explained in this
section. The primary purpose of the remove step is to make the gravity anomalies and height
5
anomalies smoother and easy to gridded by removing as much information as possible. These effects
have been removed from local/regional terrestrial gravity anomalies data ∆𝑔𝐹𝐴 and height anomalies 𝜁
providing residual gravity anomalies ∆𝑔𝑟𝑒𝑠 and residual height anomalies 𝜁𝑟𝑒𝑠 as follows
∆𝑔𝑟𝑒𝑠 = ∆𝑔𝐹𝐴 − ∆𝑔𝐺𝐺𝑀 − ∆𝑔𝑅𝑇𝑀 (2)
𝜁𝑟𝑒𝑠 = 𝜁 − 𝜁𝐺𝐺𝑀 − 𝜁𝑅𝑇𝑀 (3)
Where ∆𝑔𝐹𝐴 is representing the terrestrial gravity, 𝜁 represents the height anomalies, ∆𝑔𝐺𝐺𝑀 and
𝜁𝐺𝐺𝑀 are representing the long to short wavelength component computed from GGM, and ∆𝑔𝑅𝑇𝑀 and
𝜁𝑅𝑇𝑀 are represents the very short wavelengths component of the gravity and height anomalies,
respectively, caused by the topography of the area.
The gravimetric quasi-geoid heights (height anomalies) 𝜁ℎ have been computed using the RCR
procedures is expressed as follows
𝜁ℎ = 𝜁𝐺𝐺𝑀 + 𝜁𝑅𝑇𝑀 + 𝜁𝑟𝑒𝑠 (4)
For the computation of the gravimetric geoid model for Sudan, we have used the GRAVSOFT
software package, particularly the programs GEOEGM, TC, EMPCOV, COVFIT, and GEOCOL, to
complete the procedures of a gravimetric geoid determination.
The conversion of height anomalies 𝜁 to geoid heights N at altitude H is needed if orthometric
heights are used. We have applied the N2ZETA program in the Gravsoft for the converting the quasi-
geoid height anomalies to geoid heights, and also it can be converted to geoid heights by using geoid
quasi-geoid separation (read [12]):
∆𝑔𝐵
𝜁−𝑁 ≈ 𝐻 (5)
𝛾
Where N represents the geoid heights, 𝜁 is the quasi-geoid height anomalies, ∆𝑔𝐵 is the Bouguer
anomaly, 𝜁 is the mean normal gravity, and 𝐻 is the height above the mean sea level. It can also be
written as:
∆𝑔𝐵
𝑁≈𝜁+ 𝐻 (6)
𝛾
The simple Bouguer gravity anomalies (∆𝑔𝐵) can be given as [13]:
∆𝑔𝐵 = ∆𝑔𝑓 − 𝐴𝐵 + 𝐴 𝑇𝐶 (7)
We have applied the method for a remove-restore-compute technique [14–17] for the computation
of a gravimetric geoid modelling using LSC through the below main steps [18]:
a. Data Preparation
i. Computing the free-air gravity anomalies from the observed gravity and normal gravity
values.
ii. Computing the height anomalies values from the ellipsoidal heights and orthometric
heights.
iii. Prepare the coefficient file of the model TIM_R6e to make it ready to be used in
Gravsoft programs.
b. Remove step workflow
i. Remove the effect of a global Earth gravity field model (TIM_R6e, a spherical
harmonic expansion).
ii. Remove the effect of the topography from the data. These two steps will produce what
we will call residual data.
c. Compute step workflow
i. Estimating the empirical covariance function for the residual data after computing for
the terrain correction effect.
ii. Determining the parameters of analytic representation of the empirical covariance
function.
iii. Compute residual height anomalies 𝜁𝑟𝑒𝑠 (the quasigeoid surface) by using LSC (Geocol
6
program).
d. Restore step workflow
i. Restore the height anomalies from the GGM, which is TIM_R6e 𝜁𝐺𝐺𝑀.
ii. Restore the height anomalies from the RTM, which ASTER30 𝜁𝑅𝑇𝑀.
iii. Compute the quasi-geoid height anomalies by adding all the restored contributions
𝜁ℎ = 𝜁𝐺𝐺𝑀 + 𝜁𝑅𝑇𝑀 + 𝜁𝑟𝑒𝑠.
iv. Convert the height anomalies to geoid heights 𝑁 = 𝜁ℎ + 𝑆𝑒𝑝𝑎𝑟𝑎𝑡𝑖𝑜𝑛.
e. Validating the final result of the final geoid model against the geoid undulations of the
GPS/leveling dataset in order to check the accuracy of the final model.
In the least Square Collocation method, the residual height anomalies 𝜁𝑟𝑒𝑠 can be computed from
the residual gravity anomalies, which can be done by applying the LSC formula; it can be written as
follows [17]:
𝜁𝑟𝑒𝑠 = 𝐶𝜁𝑟𝑒𝑠∆𝑔𝑟𝑒𝑠 𝐶−1
∆𝑔𝑟𝑒𝑠 ∆𝑔𝑟𝑒𝑠 ∆𝑔𝑟𝑒𝑠 (8)
−1
Where 𝐶∆𝑔𝑟𝑒𝑠 ∆𝑔𝑟𝑒𝑠 represents the auto-covariance matrix of the residual gravity anomalies ∆𝑔𝑟𝑒𝑠 ,
𝐶𝜁𝑟𝑒𝑠∆𝑔𝑟𝑒𝑠 are represents the matrix of the cross-covariance between residual height anomalies 𝜁𝑟𝑒𝑠 and
the residual gravity anomalies ∆𝑔𝑟𝑒𝑠 .
The empirical covariance function can be computed as follows
1 𝑖𝑘
𝑐𝑜𝑣𝜁𝑟𝑒𝑠∆𝑔𝑟𝑒𝑠 = ∑ ∆𝑔𝑟𝑒𝑠 (𝜑𝑖 , 𝜆𝑖 )∆𝑔𝑟𝑒𝑠 (𝜑𝑘 , 𝜆𝑘 ) (9)
𝑁𝑖𝑘 𝑖,𝑘
Where 𝑁𝑖𝑘 is representing the number of pairs of gravity anomalies for each interval would be the
following condition
𝜓𝑖𝑘 − ∆< 𝜓𝑖𝑘 < 𝜓𝑖𝑘 + ∆ (10)
Where 𝜓𝑖𝑘 Represents the spherical distance between gravity anomalies in a pair [19]. Empirical
local covariances are calculated using the input data and then fitted to an analytical covariance model,
such as the analytical Tscherning-Rapp model. It's needed to perform the computations throughout the
LSC method. As a result, complete covariance is shown as follows
𝑛+1 𝑛+1
𝑁𝑚𝑎𝑥 𝑅𝐸2 ∞ 𝐴 𝑅𝐸2
𝑐𝑜𝑣(𝑇(𝐼), 𝑇(𝐽)) = 𝛼 ∑ 𝜎𝑛2 ( ) 𝑃𝑛 (𝑡) ∑ ( )( ) 𝑃𝑛 (𝑡) (11)
𝑛=2 𝑅𝐼 𝑅𝐽 𝑛=𝑁𝑚𝑎𝑥 +1 (𝑛 − 1)(𝑛 − 2)(𝑛 − 4) 𝑅𝐼 𝑅𝐽
Where 𝑅𝐸 represents the mean Earth’s radius, 𝛼 is the covariance parameters (or scale parameter),
𝑁𝑚𝑎𝑥 is the applied maximum degree expansion of the geopotential model, 𝜎 2 represents the error
degree variance for the anomalous potential, 𝑃𝑛 (𝑡) = 𝑃𝑛 (𝑐𝑜𝑠𝜓𝐼𝐽 ) is the Legendre polynomial of the
degree 𝑛 with the spherical distance 𝜓𝐼𝐽 between the points 𝐼 (𝑟𝐼 , 𝜑𝐼 , 𝜆𝐼 ) and 𝐽 (𝑟𝐽 , 𝜑𝐽 , 𝜆𝐽 ), 𝑅𝐵
represents the radius of the Bjerhammar sphere, 𝑟𝐼 and 𝑟𝐽 are the geocentric distances between the
points 𝐼 and 𝐽, and 𝐴 is a constant parameter in units of (𝑚/𝑠) 4. and the Parameters 𝛼, 𝐴 and 𝑅𝐵 are
obtained from the fitted of the empiricalmeasured covariance function
7
Figure 3. Flowchart of the procedures used in a gravimetric geoid computation for Sudan.
8
4. Results of remove-compute-restore technique

We have successfully applied the remove-compute-restore technique in order to compute a new
gravimetric geoid modelling in the area of Sudan and some parts of bordering countries. We have
explained the results in each procedure and the step of the geoid modelling computations in detail in
the following subsections.
4.1. Remove procedures results

Figure 4 is showing the free air gravity anomalies values that is going to be applied in the remove
procedures. We have removed the long and medium wavelengths component from the GGM (see part
(a) of Figure 5), and also we have removed the very short wavelengths component from the RTM (see
part (b) of Figure 5) the computations have been done by applying the (equation 2) in GEOEGM
program of Gravsoft (python interface). The statistics of the free-air gravity anomalies before applying
the removal process are the minimum value of -76.300 mGal, the maximum value of 82.600 mGal, the
mean of 0.654 mGal, and the standard deviation is 16.311 mGal, as well as the statistics of the free-air
gravity anomalies after removing long and medium wavelengths component from the GGM are the
minimum value of -92.455 mGal, the maximum value 76.457 mGal, the mean -2.800 mGal. The
standard deviation is 1024 mGal, and finally, the statistics of the free-air gravity anomalies after
removing very short wavelength component from the RTM are the minimum value -82.38 mGal, the
maximum value 184.04 mGal, the mean value is -1.210 mGal, and the standard deviation is 14.220
mGal (see Table 1).
Figure 4. The free gravity anomalies before applying remove procedures
9
(a) The differences between free-air gravity (b) The residuals between free-air gravity
anomalies and TIM_R6e d/o 300 ∆gFA − anomalies, TIM_R6e d/o 300 and ASTER30
GGM ∆gFA − GGM − RTM
Figure 5. The differences between the free-air gravity anomalies after removing long-medium
wavelengths from the TIM_R6e d/o 300, and after removing the very short wavelengths from
ASTER30.
Table 1. The statistics of the free-air gravity anomalies and the residuals gravity anomalies
after long-medium wavelengths, including very short wavelengths (mGal).
Anomaly Min. Max. Mean Standard deviation
∆𝒈𝑭𝑨 -76.300 82.600 0.654 16.311
∆𝒈𝑭𝑨 − 𝑮𝑮𝑴 -92.455 76.457 -2.800 15.024
∆𝒈𝑭𝑨 − 𝑮𝑮𝑴 − 𝑹𝑻𝑴 -82.38 184.04 -1.210 14.220
4.2. Compute procedures results

In the current study, the spherical distance ∆𝜓 was chosen to be 360 arc-second 6′ × 6′ gridded by 0.1°
× 0.1°. In the compute procedures, we have calculated the empirical and the analytical covariance
functions, the residuals height anomalies, and the DEM computations, as you can see in the following
subsections:
4.2.1 Empirical and Analytical Covariance Function Estimation and Representation. The
empirical covariance function was determined from the residuals gravity anomalies and then fitted to
the analytical 'modelled' covariance in the area bounded by parallels of 3°N and 23°N and meridians
of 21.15°E and 39°E for (d/o 300), the relationships between the computed empirical and analytical
fitted covariance is shown in Figure 6 and Table 2 are showing a good behaviour (agreement) between
the empirical and the analytical covariance’s up to 0.40° spherical distance, and there are a little few
differences in the rest of the spherical distances, which reflects the homogeneity of the applied residual
gravity anomalies' distribution in this investigation. In the study [20], the empirical and analytical
covariance functions have been estimated and fitted derived from residual gravity anomalies in the
area bounded by parallels of 11°N and 21°N and meridians 25°E and 35°E. The covariance parameters
for the modelled covariance were obtained and applied in our geoid collocation procedures which are:
The depth to the Bjerhammer sphere was 𝑅𝐸 − 𝑅𝐵 = −61.20 𝑘𝑚, the variance of gravity anomalies at
zero altitude of 813.06 𝑚𝐺𝑎𝑙2 the error degree variance scale factor of 0.999 with maximum degree of
300. All these predicted parameters have been used in the computations of the quasi- geoid height
anomalies on the 6′ × 6′ grid from the residual gravity anomalies.
10
Table 2. Final outputs data after using both EMPCOV and COVFIT programs.
PSI (ψ) (Spherical

Observed Model
distance) Error
Covariance Covariance
[𝑑𝑒𝑔] [𝑚𝐺𝑎𝑙2] [𝑚𝐺𝑎𝑙2] [𝑚𝐺𝑎𝑙2]
0.000 236.259 232.585 0.048
0.167 149.508 141.977 0.005
0.333 44.274 39.562 0.007
0.500 1.696 -21.152 0.010
0.667 3.181 -43.950 0.011
0.833 -6.363 -37.725 0.014
1.000 -262 -1056 0.017
1.167 -10.791 9.381 0.018
1.333 -10.269 23.530 0.021
1.500 -8.693 22.622 0.027
1.667 -7.103 9.837 0.031
1.833 9.528 -935 0.041
2.000 7.912 -1870 0.055
2.167 -8.648 -1368 0.070
2.333 -2.162 -6.381 0.092
2.500 -9.239 4.846 0.114
2.667 -12.605 11.556 0.121
2.833 -22.736 10.579 0.140
3.000 -1447 3.596 0.155
3.167 -13.524 -4.441 0.153
Figure 6. The empirical and the analytical covariance functions for the residual
gravity anomalies of the area bounded by parallels of 3°N and 23°N and meridians of
21.15°E and 39°Efor (d/o 300).
4.2.2 Residuals height anomalies computations. We have estimated the reduced height anomalies
(𝜁𝑟𝑒𝑠 ) from the reduced gravity anomalies ∆𝑔𝑟𝑒𝑠 at GPS/levelling stations. The differences between
predicted and observed height anomalies are shown in Table 3 The findings have shown that we may
estimate the reduced height anomalies with an error (in terms of STD) of 8.22 cm.
11
Table 3. The statistics of predicted and observed height anomalies [m] at GPS/levelling stations.
Anomaly OBSERVATIONS PREDICTIONS DIFFERENCE ERROR ESTIMATES
MEAN 0.984 0.237 0.747 0.105
ST.DEVI. 0.478 0.361 0.611 0.082
MAX 2.598 1.017 4.121 0.272
MIN -0.221 -2.230 -0.273 0.014
The residual height anomalies (𝜁𝑟𝑒𝑠 ) have been estimated one more time by taking the predicted
observation error into account during geoid collocation computations (which was STD of 8.22 cm),
Table 4 is showing the estimated height anomalies after removing the bias, where the standard
deviations of the estimated error have been reduced from 8.22 cm to 1.8 cm. Also, the mean of the
differences has been reduced from 0.747 m to 0.000 m (see Table 3 and Table 4). Figure 7 shows the
residual height anomalies 𝛿𝜁𝑟𝑒𝑠 computed from GEOCOL program long- medium wavelengths.
Table 4. The statistics of predicted and observed height anomalies [m] at GPS/levelling stations after
reducing the bias between both height systems.
Anomaly OBSERVATIONS PREDICTIONS DIFFERENCE ERROR ESTIMATES
MEAN 0.984 0.984 0.000 0.052
ST.DEVI. 0.478 0.326 0.245 0.018
MAX 2.598 1.940 1.200 0.088
MIN -0.221 0.236 -0.690 0.025
Figure 7. The residual height anomalieswere Figure 8. Residual terrain effects over
computed from the GEOCOL program long- Sudan from ASTER30.
medium wavelengths.
12
Figure 9. Terrain correction over Sudan Figure 10. Topographic affect over Sudan
and some areas of bordering countries. andsome areas of bordering countries.
4.2.3 DEM computations results. We have extracted the grid points from the Gravsoft grid format
file (GRI file) of the digital elevation model (ASTER30) to compute needed computations that will be
used in the restore stage (the calculations should be in terms of height anomalies). We have calculated
the residual terrain effect over Sudan (see Figure 8) and computed the terrain corrections over the
study area, as shown in Figure 9 These results will be used in the Bouguer anomalies computations
that are needed to convert the quasi-geoid heights to the geoid heights (read equation 6 and equation
7). Figure 10 shows the topographic effect on Sudan.
4.3. Restore procedures result

The gravimetric quasi-geoid model (SDN-LSC-QG22) has been computed by applying the restore
procedures (see Figure 3). The restore procedures have been used for the outputs of compute strategies
which are the reference height anomalies calculated from the GGM 𝜁𝐺𝐺𝑀 , the corresponding ones from
topography affects 𝜁𝑅𝑇𝑀 . And the residual height anomalies, both quantities were added to the residual
height anomalies δζ (see equation 5). Finally, we have converted the gravimetric quasi-geoid model
SDN-LSC-QG22 to the gravimetric geoid model SDN-LSC-G22 by applying geoid quasi-geoid
separation (see equation 6). Table 5 shows the statistics of the quasi-geoid model (SDN-LSC-QG22),
which has shown the minimum value is -24.552 m, the maximum value is 21.891 m, the mean is 0.654
m, and the standard deviation is 9.209 m, geoid to quasi-geoid separation which has shown the
minimum value is -2.038 m, the maximum value is 0.470 m, the mean is 0.080 m, and the standard
deviation is 0.154 m. The final geoid model (SDN-LSC-G22) has shown a minimum value of -24.582
m, a maximum value of 21.857 m, a mean of 0.577 m, and a standard deviation of 9.238 m. The
figures in this subsection are shown the different estimated components in the restore stage during the
determination of a gravimetric geoid model and quasi-geoid model. Figure 11 shows the height
anomalies contribution of the TIMR6e compared with the residual terrain effect over Sudan.
Figure 12 shows the combination of the residuals height anomalies δ𝜁𝑅𝐸𝑆 With residual terrain
effects. Figure 13 shows the gravimetric quasi-geoid model SDN-LSC-QG22, which is combined with
the residuals height anomalies δ𝜁𝑅𝐸𝑆, residuals terrain effects, and the height anomalies of the TIMR6e
13
ζGGM. In order to convert the height anomalies of the quasi-geoid, we have applied N2ZETA
program in the Gravsoft, and we have successfully computed the complete Bouguer anomalies (see
Figure 14) as well as we have calculated the geoid-to-quasi- geoid correction (separation) see Figure
15. Finally, we have computed the final gravimetric geoid model over Sudan and some parts of
neighbouring countries (SDN-LSC-G22) (see Figure 16).
Table 5. Statistics of the Quasi-geoid model (SDN-LSC-QG22), geoid-to-quasi geoid

correction, and Final geoid model (SDN-LSC-G22) [m].
Statistics Min. Max. Mean Standard
deviation
Quasi-geoid model (SDN-LSC-QG22) -24.552 21.891 0.654 9.209
Geoid to Quasi-geoid correction -2.038 0.470 0.080 0.154
Final geoid model (SDN-LSC-G22) -24.582 21.857 0.577 9.238
EGM2008 -24.122 22.070 0.466 9.186
Tongji-GMMG2021S -23.955 21.709 0.494 9.179
Figure 11. The height anomalies contribution Figure 12. The restored residual height
ofthe TIMR6e compared against residuals anomalies combined with the residual terrain
terrain effect over the study area (6′ × 6′). effects.
14
Figure 13. The gravimetric quasi-geoid model Figure 14. Estimated Bouguer anomalies using
SDN-LSC-QG22 in [m]. Gravsoft (N2ZETA program)
Figure 15. The geoid-to-quasi-geoid Figure 16. The gravimetric geoid model
correction (separation) [m] at a 6′ × 6′grid. (SDN-LSC-G22) in [m].
15
5. Validation of the final geoid model against available GPS/leveling data

Presently, the most reliable technique to determine a gravimetric geoid's real potential is to compare its
result with the computed geoidal heights from GPS/levelling data. In order to evaluate our new
gravimetric geoid model (SDN-LSC-G22), the geoid heights obtained from our model have been
compared against the corresponding ones computed from the GPS/levelling data (see equation 12).
Figure 17 are showing the differences in the geoid heights between the new gravimetric geoid model
and 66 GPS/levelling data distributed in the area of Khartoum, the standard deviation of the
differences between our gravimetric geoid model (SDN-LSC-G22) and these points are 17.4 cm.
Correspondingly, we have evaluated our model (SDN-LSC-G22) against [8] GPS/Levelling (see
Figure 18), and the standard deviation of the differences is 51.3 cm as is shown in Table 6 As we have
mentioned before, the levelling data of [8] GPS and levelling data have been extracted from old
vertical geodetic controls based on the first, second and third-order (levelling data based on normal
gravity), their accuracy is estimated between 0.1 to 0.5 meters, and that was the reason of having a
standard deviation of 51.3 cm in these points compared against SDN-LSC-G22. Table 7 and Table 8
show the statistical differences between the models EGM2008 and SGG-UGM-2 and the available
GPS/levelling data, respectively, in terms of comparing their accuracies with the SDN-LSC- G22. The
differences between the EGM2008 model and [8] GPS/levelling data is 68.8 cm; it has shown 23.9 cm
against the differences with the rest of the GPS/levelling dataset. Finally, the model has a standard
deviation of 50.3 cm as a difference between all available GPS/levelling data. As well as, the
difference between the SGG-UGM-2 model and [8] GPS/levelling data is 67.6 cm, it has shown 24.5
cm against the differences with the rest of the GPS/levelling dataset. Finally, the model has a standard
deviation of 47.4 cm as a difference between all available GPS/levelling data.
𝛥𝑁 = 𝑁𝑇ℎ𝑒 𝑛𝑒𝑤 𝑔𝑟𝑎𝑣𝑖𝑚𝑒𝑡𝑟𝑖𝑐 𝐺𝑒𝑜𝑖𝑑 − 𝑁𝐴𝑣𝑎𝑖𝑙𝑎𝑏𝑙𝑒 𝐺𝑃𝑆/𝑙𝑒𝑣𝑒𝑙𝑖𝑛𝑔 (12)
Table 6. Statistics of the differences between the available GPS/Levelling data compared
against the new gravimetric Geoid model (SDN-LSC-G22) [m].
No. of the
Statistics Min. Max. Mean STD
points
𝑵𝐭𝐡𝐞 𝐧𝐞𝐰 𝐠𝐫𝐚𝐯𝐢𝐦𝐞𝐭𝐫𝐢𝐜 𝐆𝐞𝐨𝐢𝐝−𝑨𝒃𝒅𝒂𝒍𝒍𝒂𝘍𝒔 𝑮𝑷𝑺/𝒍𝒆𝒗𝒆𝒍𝒊𝒏𝒈 19 -0.718 0.983 0.017 0.513
𝑵𝐭𝐡𝐞 𝐧𝐞𝐰 𝐠𝐫𝐚𝐯𝐢𝐦𝐞𝐭𝐫𝐢𝐜 𝐆𝐞𝐨𝐢𝐝−𝑻𝒉𝒆 𝒓𝒆𝒔𝒕 𝒐𝒇 𝒕𝒉𝒆 𝑮𝑷𝑺/𝒍𝒆𝒗𝒆𝒍𝒊𝒏𝒈 66 -0.556 0.380 -0.002 0.174
𝑵𝐭𝐡𝐞 𝐧𝐞𝐰 𝐠𝐫𝐚𝐯𝐢𝐦𝐞𝐭𝐫𝐢𝐜 𝐆𝐞𝐨𝐢𝐝−𝐚𝐥𝐥 𝑨𝒗𝒂𝒊𝒍𝒂𝒃𝒍𝒆 𝑮𝑷𝑺/𝒍𝒆𝒗𝒆𝒍𝒊𝒈 85 -0.614 1.414 0.033 0.341
against EGM2008 [m].
No. of the
points
𝑵𝐄𝐆𝐌𝟐𝟎𝟎𝟖−𝑨𝒃𝒅𝒂𝒍𝒍𝒂𝘍𝒔 𝑮𝑷𝑺/𝒍𝒆𝒗𝒆𝒍𝒊𝒏𝒈 19 -1.707 0.351 0.637 0.688
𝑵𝐄𝐆𝐌𝟐𝟎𝟎𝟖−𝑻𝒉𝒆 𝒓𝒆𝒔𝒕 𝒐𝒇 𝒕𝒉𝒆 𝑮𝑷𝑺/𝒍𝒆𝒗𝒆𝒍𝒊𝒏𝒈 66 -0.257 0.965 0.183 0.239
𝑵𝐄𝐆𝐌𝟐𝟎𝟎𝟖−𝐚𝐥𝐥 𝑨𝒗𝒂𝒊𝒍𝒂𝒃𝒍𝒆 𝑮𝑷𝑺/𝒍𝒆𝒗𝒆𝒍𝒊𝒏𝒈 85 -1.707 0.965 0.001 0.503
against SGG-UGM-2 [m].
No. of
the points
𝑵𝐒𝐆𝐆𝐔𝐆𝐌𝟐−𝑨𝒃𝒅𝒂𝒍𝒍𝒂𝘍𝒔 𝑮𝑷𝑺/𝒍𝒆𝒗𝒆𝒍𝒊𝒏𝒈 19 -1.612 0.514 0.613 0.676
𝑵𝐒𝐆𝐆𝐔𝐆𝐌𝟐−𝑻𝒉𝒆 𝒓𝒆𝒔𝒕 𝒐𝒇 𝒕𝒉𝒆 𝑮𝑷𝑺/𝒍𝒆𝒗𝒆𝒍𝒊𝒏𝒈 66 -0.324 1.116 0.080 0.245
𝑵𝐒𝐆𝐆𝐔𝐆𝐌𝟐−𝑨𝒗𝒂𝒊𝒍𝒂𝒃𝒍𝒆 𝑮𝑷𝑺/𝒍𝒆𝒗𝒆𝒍𝒊𝒏𝒈 85 -1.612 1.116 0.075 0.474
16
Figure 17. The differences between the geoid heights from GPS/levelling data and the
new gravimetric geoid model [m].
Figure 18. The differences in the geoid heights from [8] GPS/levelling data and the
new gravimetric geoid model [m].
Table 9 shows the comparison between the SDN-LSC-G22 and the model KTH-SDG08 which has
been computed by [8] and the model SUD-GM2014 which has been calculated by [20]. The
comparison is going to be in terms of data used, applied methods for geoid modelling computations
and validations, and finally the models accuracies.
17
Table 9. The comparison between SDN-LSC-G22 and the latest computed geoid models.
Item Current study KTH-SDG08 geoid model SUD-GM2014 geoid model
SDN-LSC-G22 geoid model [8] [20]
Datasets Used GGM model Used GGM model Used GGM model
GO_CONS_GCF_2_TIM_R6e EIGEN-GL04C GO_CONS_GCF_2_TIM_R4
Degree/order=300 Degree/order=360 Degree/order=250
DEM: DEM: DEM:

ASTER SRTM SRTM30_PLUS
Gravity dataset Gravity dataset Gravity dataset

BGI gravity dataset GETECH gravity data GETECH gravity data
GPS and leveling data GPS and leveling data GPS and leveling data
85 points 19 points 19 points
Applied Geoid modeling method Geoid modeling method Geoid modeling method
Methods Least Squares Collocation least-squares modification (LSM) Least Squares Collocation (LSC)
(LSC)
Applied programs Applied programs Applied programs

Gravsoft programs (RCR LSMSSOFT (LSMS) Gravsoft programs (RCR
technique) technique)
Validation method Validation method Validation method

Validation the model by using Validation the model by using Validation with the geometric data
geometric data and comparing geometric data and Fitting with 7 and Fitting with 7 parameters
the accuracies with EGM2008 parameters model. model.
and SGG-UGM-2 in the same
geometric points.
Accuracies Standard deviation using 19 Standard deviation using 19 Standard deviation using 19 GPS
in terms of GPS and leveling points from GPS and leveling points from and leveling points from [8]:
STD (cm) [8]: (before fitting) [8]: (before fitting) (before fitting)
STD = 51.3 cm
STD = 58.9 cm STD = 64.4 cm
Standard deviation using new Standard deviation using 19 Standard deviation using 19 GPS
66 GPS and leveling points: GPS and leveling points from and leveling points from [8]:
[8]: (after fitting with 7 (after fitting with 7 parameters
parameters model) model)
STD = 17.4 cm STD = 29 cm STD = 30.6 cm
Overall validation with

available GPS and leveling
data (85 points)
STD = 34.1 cm
We have validated the

differences between the models
EGM2008 and SGG-UGM-2
and the available GPS and
levelling data and our model
has shown us better accuracies
compared with these high-
resolution models (see Table 6,
Table 7, and Table 8)
18
6. Conclusion
The precision of the new gravimetric geoid model SDN-LSC-G22 of Sudan and some areas of
bordering countries has been investigated by using geoid undulations from the ground-based GPS and
levelling data and the GGM models EGM2008 and SGG-UGM-2. Our gravimetric geoid model
(SDN- LSC-G22) has an accuracy of 17.4 𝑐𝑚, in terms of a standard deviation compared to 66 GPS
and Leveling data distributed in the area of Khartoum (most of these points are control points and
benchmarks). Also, we have evaluated our gravimetric geoid model by using 19 points distributed in
the area of Sudan, and they have shown a standard deviation of 51.3 𝑐𝑚. The overall accuracy of
SDN- LSC-G22 compared with the geoid undulation of all GPS and levelling has indicated an STD of
34.1 𝑐𝑚. The model SDN-LSC-G22 has shown better accuracy and significant differences compared
with the GGM models EGM2008 and SGG-UGM-2 in terms of the differences with the available GPS
and levelling data which have shown ~17 𝑐𝑚 differences using [8] GPS and levelling data. It has
demonstrated STD of ~7 𝑐𝑚 differences by using 66 GPS and levelling data. Therefore, the model
SDN-LSC-G22 provided better improvements and reliable geoid heights over Sudan compared to
EGM2008 and SGG-UGM-2 gravity field models.
7. Acknowledgments
The calculations of the new gravimetric geoid model (SDN-LSC-G22) for the area of Sudan and some
areas of bordering countries has been done using GRAVSOFT modules (Python Interface) of Rene
Forsberg and C.C.Tscherning. The spatial representations of the results have been plotted by using the
GMT6 (Generic Mapping Tools 6) software [21].
8. References
[1] Pinon D A 2016 Development of a Precise Gravimetric Geoid Model for Argentina 182
[2] Saadon 2019 On the Computation of Gravimetric Geoid of Egypt using Different Estimation
Techniques By Eng. Ahmed Saadon Mohamed Lotfy Zagazig Univ. Fac. Eng. Constr. Eng.
&Utilities Dept.
[3] Adam M O 1967 A Geodetic Datum for the Sudan
[4] Salih A B 1983 The shape of the geoid in the Sudan Aust. J. Geod. Photo. Surv. 39, 27–38.
[5] Salih A B 1985 The shape of the geoid by astrogeodetic and Doppler methods in the Sudan
Aust. J. Geod. Photo. Surv. 43, 71–83.
[6] Salih A B, Fashir H H and Latif K A 1990 Doppler geoid in Sudan Surv. Rev. 30 319–22
[7] Fashir H H 1991 The gravimetric geoid of Sudan J. Geodyn. 14 19–36
[8] Abdalla A 2009 Determination of a Gravimetric Geoid Model of Sudan Using the KTH Method
J. Phys. F Met. Phys. 4 1908–15
[9] Yamaguchi Y, Kahle A B, Tsu H, Kawakami T and Pniel M 1998 Overview of advanced
spaceborne thermal emission and reflection radiometer (ASTER) IEEE Trans. Geosci.
Remote Sens. 36 1062–71
[10] Moritz 1980 Geodetic reference system 1980 Bull. Géodésique 54 395–405
[11] Ruffhead A 1987 AN INTRODUCTION TO LEAST-SQUARES COLLOCATION 29 85–94
[12] Heiskanen and Moritz 1967 Physical Geodesy
[13] Wellenhof & Moritz 2005 Physical Geodesy by Bernhard Hofmann-Wellenhof and Helmut
Moritz (Institut für Navigation und Satellitengeodäsie Technische Universität Graz, Graz,
Austria This)
[14] El-Ashquer M, Elsaka B and El-Fiky G 2017 EGY-HGM2016: an improved hybrid local geoid
model for Egypt based on the combination of GOCE-based geopotential model with
gravimetric and GNSS/levelling measurements Arab. J. Geosci. 10 53115
[15] Forsberg R and Tscherning C C 1981 The use of height data in gravity field approximation by
collocation. J. Geophys. Res. 86 7843–54
[16] Schwarz K P, Sideris M G and Forsberg R 1990 The use of FFT techniques in physical geodesy
Geophys. J. Int. 100 485–514
19
[17] Saadon A, El-Ashquer M, Elsaka B and El-Fiky G 2021 Determination of local gravimetric
geoid model over Egypt using LSC and FFT estimation techniques based on different
satellite- and ground-based datasets Surv. Rev.
[18] Tscherning C C 2013 Geoid determination by 3D least-squares collocation Lect. Notes Earth
Syst. Sci. 110 311–36
[19] Sans` F and Sideris M G 2013 Geoid Determination Theory and Methods ed M P. Blondel, Bath
J. Reitner, G¨ottingen K. St¨uwe, Graz M.H. Trauth, Potsdam D. Yuen (WH Freeman and
Company. San Francisco, CA)
[20] Godah W 2015 A New Gravimetric Geoid Model for the Area of Sudan Using the Least
Squares Collocation and a GOCE-Based GGM Int. Assoc. Geod. Symp. 144 145–53
[21] Wessel P, Luis J F, Uieda L, Scharroo R, Wobbe F, Smith W H F and Tian D 2019 The Generic
Mapping Tools Version 6 Geochemistry, Geophys. Geosystems 20 5556–64
20

Determination of A New Gravimetric Geoid Modelling

Uploaded by

Copyright:

Available Formats

Determination of A New Gravimetric Geoid Modelling

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Determination of A New Gravimetric Geoid Modelling

Uploaded by

Copyright:

Available Formats

IOP Conference Series: Earth and Environmental Science

PAPER • OPEN ACCESS

Determination of a new gravimetric geoid modelling for Sudan using the

View the article online for updates and enhancements.

This content was downloaded from IP address 216.19.203.64 on 20/01/2023 at 12:38

Determination of a new gravimetric geoid modelling for

Anas Sharafeldin Mohamed Osman1,2 and Ira Mutiara Anjasmara1, *

Figure 2. Aster digital elevation model over Sudan.

4. Results of remove-compute-restore technique

4.1. Remove procedures results

Figure 4. The free gravity anomalies before applying remove procedures

4.2. Compute procedures results

PSI (ψ) (Spherical

4.3. Restore procedures result

Table 5. Statistics of the Quasi-geoid model (SDN-LSC-QG22), geoid-to-quasi geoid

5. Validation of the final geoid model against available GPS/leveling data

DEM: DEM: DEM:

Gravity dataset Gravity dataset Gravity dataset

Applied programs Applied programs Applied programs

Validation method Validation method Validation method

Overall validation with

We have validated the

You might also like