You are on page 1of 13

Available online at www.sciencedirect.

com

ScienceDirect
Advances in Space Research 72 (2023) 3933–3945
www.elsevier.com/locate/asr

DCB estimation and analysis using the single receiver


GPS/GLONASS observations under various seasons and
geomagnetic activities
Ahmed Wageeh a, Mohamed Doma a,b, Ahmed Sedeek c, Alaa Elghazouly a,⇑
a
Faculty of Engineering, Menoufia University, Egypt
b
Pyramids Higher Institute for Engineering and Technology, Egypt
c
Faculty of Petroleum and Mining Engineering, Suez University, Egypt

Received 16 January 2023; received in revised form 22 July 2023; accepted 28 July 2023
Available online 3 August 2023

Abstract

Global Navigation Satellite Systems (GNSS) have many applications in surveying, geodesy, and everyday life, so the accuracy of
observations is very important. To have accurate results, many errors should be corrected, including ionospheric errors. Estimating
the Differential Code Bias (DCB) and Total Electron Content (TEC) is essential to reduce ionospheric errors. This study’s main aim
is to determine how the combination of GPS (Global Positioning System) and GLONASS (GLObal NAvigation Satellite System) obser-
vations affect the estimation of DCB of satellites (SDCB) and receivers (RDCB) under the effects of geomagnetic storms and seasonal
changes. The MATLAB-based software GR_DCB (GPS and GLONASS DCB estimation) was developed to estimate the SDCB and
RDCB based on GPS and GLONASS observations. The results were tested using data from the International GNSS Service (IGS) net-
work from nine stations and compared to previous studies to assess the effectiveness of the developed code. The findings are consistent
with IGS products and more precise than previous studies that relied solely on GPS observations. An analysis of the effects of the geo-
magnetic storm on DCB variance is considered for June 22–23, 2015. During the geomagnetic storm, all GPS and GLONASS stations
exhibited a fall in RDCB, led by a modest increase in some stations. The SDCB had no observed change on storm days compared with
quiet days. The seasonal change affected GPS and GLONASS stations RDCB. Winter, which has the highest RDCB mean values, is also
the season when a deviation in RDCB is the greatest. Autumn and Summer had the lowest dispersion in RDCB over the season and the
lowest RDCB values.
Ó 2023 COSPAR. Published by Elsevier B.V. All rights reserved.

Keywords: GNSS; GPS; GLONASS; Ionosphere; DCB; Single Receiver

1. Introduction the results become. The ionosphere affects any device that
uses navigation signals, including GNSS. The ionosphere
The increase in GNSS reference stations receiving sig- is the main contributor to signal propagation error when
nals from GLONASS has made it necessary to include a signal is being transmitted from a satellite to a ground
GLONASS observations and GPS in studying the iono- receiver. Dual-frequency GNSS receivers can calculate
sphere. The more observations grow, the more accurate ionospheric delays by carrier phases and pseudo-range
observation for two frequencies (Jin et al., 2004).
While measuring the total ionospheric delay, a satellite
⇑ Corresponding author.
and receiver-dependent systematic error known as DCB
E-mail address: alaa_elghazouly@sh-eng.menofia.edu.eg (A. Elgha-
zouly).
is found; the DCB is an inaccuracy in measuring the iono-

https://doi.org/10.1016/j.asr.2023.07.063
0273-1177/Ó 2023 COSPAR. Published by Elsevier B.V. All rights reserved.
A. Wageeh et al. Advances in Space Research 72 (2023) 3933–3945

sphere TEC using GNSS dual-frequency pseudo-range One of the most significant ionosphere-affecting phe-
readings that can lead to non-physical TEC negative val- nomena is the geomagnetic storm. A significant distur-
ues. For illustration, in TEC estimation, one nanosecond bance of the Earth’s magnetosphere is caused by solar
(ns) DCB results in 2.85 TEC unit (TECU) for GPS fre- wind and Heliospheric Magnetic Field (HMF) distur-
quencies and 2.92 TECU for GLONASS frequencies bances. In a geomagnetic storm, the high-charge electrons
(Mylnikova et al., 2015). DCB values vary between one will rise substantially. Furthermore, Geomagnetic-
GNSS satellite to the other, and the ground receivers. Induced Currents (GIC) caused by geomagnetic field dis-
Yasyukevich et al. (2015) studied the DCB values for turbances will affect high-voltage Earth devices (Ansari
SDCB and RDCB of GPS and GLONASS from 2000 to et al., 2018; Sedeek, 2022). The storm can lead to a high
2014 over 270 stations. It was noticed that the GPS SDCB increase in ionospheric error, reaching 265% compared
values change by  0.35 ns per year, while the SDCB dis- with quiet days (Sedeek, 2020). Li et al. (2021) studied
crepancies for GLONASS are more significant, ranging the storm on 28 May 2017 to better understand the factors
from 1.71 to 3.42 ns per year. The advantage of using that affect DCB estimation during geomagnetic storms. It
GLONASS in addition to GPS (besides adding more was found that RDCB is affected by the ionospheric distur-
observations that increase accuracy) is that GLONASS bance; during the storm, the standard deviation of the
satellites can effectively cover some areas that GPS satel- RDCB abruptly rises, and it then returns to its previous
lites cannot reach at a particular time (Choi et al., 2018). level after the storm has passed.
GLONASS satellites have the largest orbit inclination The SDCB values show slight variations over several
angle among all GNSS constellations (about 10 degrees lar- months (Choi et al., 2011) with day-to-day variations of
ger than GPS), which is more helpful in covering high- less than 0.2 ns. On the other hand, RDCB values have sig-
latitude areas (Fig. 1). GLONASS has an equivalent or nificant variations, even within one-hour intervals, due to
even better positioning accuracy than GPS in high latitudes many factors (Coco et al., 1991). According to Liu et al.
because of its greater orbit inclination angle, and using (2020), there is a notable association between the ambient
both GPS and GLONASS results in the best accuracy for temperature and the long-term RDCB changes recom-
coordinates (Zheng et al., 2022). mending 15 degrees to have stable RDCB values. Coster
There are various methods for calculating the iono- et al. (2013) investigated how the GPS receiver bias varies
spheric delay, such as Otsuka et al. (2002) and Liu et al. with temperature, including indoor and outdoor tempera-
(2004). These studies modeled the ionosphere as a sphere tures. They demonstrated that temperature-dependent bias
with an infinitesimally thin layer. It was assumed that the is evident and that this temperature dependence varies
TEC is uniform over small areas during specific time inter- from one receiver to another. Yasyukevich et al. (2015)
vals and that the DCB values of satellites and receivers studied the DCB values variation of GPS and GLONASS
remain constant over a day. Jin et al. (2012) developed devices and the consequential TEC error from 2000 to 2014
MATLAB code (M_DCB) to calculate instrumental biases over 270 stations to find that the systematic variation in
using dual-frequency GPS observations with two options RDCB and considerable seasonal variations caused by
of analysis: single-station and multi-station. However, the the environmental state (temperature and humidity).
software considered only Codeless Tracking receivers. In the current paper, more attention is considered to the
Sedeek et al. (2017) presented another MATLAB software satellite’s and receiver’s DCB values of GPS and GLO-
(ZDDCBE). ZDDCBE used weighted least-square to NASS and the variation in DCB under different conditions.
reduce DCB estimation error. Choi et al. (2018) added The calculations are made using a weighted least square
the GLONASS observations to GPS to calculate the algorithm developed on MATLAB and assessed using data
DCB values and TEC using nine South Korean stations. from IGS reference stations, and that will be discussed in
On average, there was a 0.64 TECU difference between sections (2 & 3). The algorithm’s results are validated in
GPS-only TEC and GPS + GLO TEC. section (4). Section (5) shows the case studies of geomag-

Fig. 1. Ground tracks of GPS satellites on the left and GLONASS satellites on the right on day 37 of year 2021 (Huang, 2022).
3934
A. Wageeh et al. Advances in Space Research 72 (2023) 3933–3945

netic storm and the seasonal variation and their effect on Where: b is the geocentric latitude of IPPs, s ¼ kIPP  k0
DCB values. Section (6) discusses the results and presents is the sun-fixed longitude of the IPPs, kIPP ; k0 are the geo-
the conclusions. centric longitude of IPPs and the apparent.
Solar time, n is the degree of the spherical function, m is
the order of spherical harmonic function; fourth order is used,
2. Mathematical model P mn normalized Legendre polynomials, amn , bmn are the esti-
mated spherical harmonics coefficients, z is the satellite eleva-
Two types of waves can be received for any GNSS satel- tion angle, and Mp(z) is a mapping function to convert Slant
lite: pseudo-range and carrier phase. Carrier phase observa- TEC (STEC) into Vertical TEC (VTEC) since STEC signifi-
tions are more accurate as they have slight measurement cantly varies depending on a satellite elevation angle
noise compared to pseudo-range observations, which have (Klobuchar 1987; Mannucci et al. 1993), c is the speed of light
significant noise (but unambiguous) measurements. Still, in vacuum = 299,792,458 m/s, f 1 and f 2 refer to frequencies
the integer ambiguity in the carrier phase makes it ambigu- of the GNSS navigation signals. The frequencies are constant
ous. Merging the two types of observations is the key to for all satellites in the case of GPS signals, f 1 and f 2 are
having the best results. The Receiver INdependent 1575.42 MHz and 1227.60 MHz, respectively. On the other
EXchange (RINEX) file contains the observations from hand, GLONASS has a frequency division method to distin-
GPS and GLONASS satellites for each station. These obser- guish between individual satellites (Hoffmann-Wellenhof
vations include, in addition to ionospheric delay, many et al., 2008). The two frequencies are determined to f 1 =
other types of errors, such as troposphere delay, instrumen- (1602 + 9/16 k) MHz and f 2 = (1246 + 7/16 k) MHz, respec-
tal clock errors, multipath errors, etc. tively, where k refers to the frequency channel number, which
To estimate the ionospheric error accurately geometry- can be checked in the satellite navigation file, P 4;sm is the
free linear combination is used. Only ionospheric delays smoothed P4 by its corresponding carrier phase observation
are frequency dependent; thus, we can eliminate all other L4 to reduce noise, R is the radius of the Earth = 6,378,137
errors (clock errors, troposphere delay, multipath errors, m, H is the altitude of the ionosphere electrons
etc.) using this combination given by (Sardón et al., 1994). concentration = 506,700 m, and a is a correction fac-
P4 ¼ Psr;1 ðiÞ  Psr;2 ðiÞ tor = 0.9782. H and a values are evaluated as the Modified
Single Layer Model (MSLM) (Schaer, 1999), which is used
¼ Is1  Is2 þ c  DCBr þ c  DCBs ð1Þ by most of the IAACs, e.g., CODE, ESA, and NRCan.
Equation (3) exhibits a rank deficiency since a bias com-
L4 ¼ Lsr;1 ðiÞ  Lsr;2 ðiÞ mon to all satellites cannot be distinguished from a corre-
¼ Is2  Is1 þ kðN2  N1 Þ þ kðbr;2  bs;2 Þ  kðbr;1 sponding bias common to all receivers. Zero-mean
reference is added to separate the DCBs of satellites and
 bs;1 Þ ð2Þ
receivers (Montenbruck et al., 2014).
Spherical harmonics provide a complete and orthogonal
With r, s, j, and i are the receiver, satellite, frequency,
basis on the entire surface of a sphere, but their basis func-
and epoch indices, and where: Psr;j ðiÞ, Lsr;j ðiÞ Pseudo-range
tions may not remain orthogonal in a limited local region,
and carrier-phase measurements, respectively, in meters, which can affect the determination of the model’s coeffi-
Isj the ionospheric delay, in meters, Nj carrier-phase integer cients. Moreover, the boundary conditions for spherical
ambiguities, in cycles, k carrier-phase wavelength, in harmonics may need to be adjusted for local regions, such
meters, bs,i, br,i satellite and receiver instrument biases as those near the poles, to account for the sphere’s curva-
phase advance, respectively, in cycles, and DCBs ; DCBr ture and ensure an accurate representation of the region.
are the satellite and receiver differential code biases, respec- To address these issues, various techniques have been pro-
tively, in time units. posed to improve the accuracy of the computation of
The ionospheric spherical harmonic function is used to the spherical harmonic coefficients in the local region
calculate satellite and receiver DCBs for GPS and GLO- (Razin & Voosoghi, 2017). Our research aims to evaluate
NASS at any station from the smoothed P4 observations the impact of GLONASS observation on GPS observa-
following the technique described by Jin et al. (2012). tion in estimating DCB. We compare our results with pre-
The resulting equation is given by: vious research that uses GPS only and use the same model.
X
N X
n The least squares approach is commonly used to solve
P mn sinðbÞðamn cosðmsÞ þ bmn sinðmsÞÞ the spherical harmonic function, as there are fewer vari-
n¼0 m¼0
" # ables than observations (Ma and Maruyama 2003; Jin
f 21 f 22 et al. 2008). The following equation illustrates the structure
¼ MpðzÞ  ðP 4;sm  c  DCB r  c  DCB s Þ ð3Þ
40:3ðf 21  f 22 Þ of the current algorithm’s unknown parameters, coefficient
   matrix, and observation vector. The two conditions of
R zero-mean reference are shown in the last two rows as
MpðzÞ ¼ cos arcsin sinðaz ð4Þ
RþH the sum of all SDCB for the same constellation equals zero.

3935
A. Wageeh et al. Advances in Space Research 72 (2023) 3933–3945

2 3 2
DCBGr ðf þ dÞ
6 DCBR 7
6 r 7
x¼h i2 ð9Þ
2 3 6 7
F 0 F  0 0  0 Eðb; sÞ  Eðb; sÞ
6
6
7
7 0:05 þ sinðzÞ
0:02
2
6 . 6 G 7
6 . .. .. .. .. .. .. .. .. .. 7 7 6 DCBs1 7
6 7
6 . . . F . . . . . . . 7 6 .. 7
6 7 6 7
6F 0 0  F 0  0  Eðb; sÞ 7 6 . 7 Where X is the unknown parameters (variables) vector,
6 Eðb; sÞ 7 6
6 7 G 7
6 0 F  F   Eðb; sÞ 7 6 DCBs32 7 A is the coefficient matrix, L is the observation vector (val-
6
0 0 0 Eðb; sÞ
7 6 7
6 DCBR 7
6 .
6 .. .. .. .. .. ... .. .. 7 6 s1 7 ues of P 4;sm ), x is the weight matrix and f, d are two con-
6 . . . . 0 F 0 . . 7 7 6
6 ..
7
7
6 7 6 . 7 stants equal to 5 and 2 cm, respectively.
6 0 F 0  0 0  F Eðb; sÞ  Eðb; sÞ 7 6 7
6 7 6 DCBR 7
6 0    0 7 6 s24 7
4 0 1 1 0 0 0 5 6 7
6 I coef ð1Þ 7
0 0 0  0 1  1 0  0 6 7
6 .. 7 3. Software details
6 7
4 . 5
I coef ðkÞ
2 3 The GR_DCB software was developed in MATLAB.
P 4 ð1Þ
6 P 4 ð2Þ 7 The inputs for this software are shown in Fig. 2.
6 7
6 . 7 The study passes through three steps. The first step is to
6 . 7
6 . 7 show the value of adding GLONASS observations to GPS;
6 7
6 .. 7
6 . 7 A comparison is made between GR_DCB and two soft-
6 7
6 . 7
6 .. 7 ware that use single station analysis of GPS observations
6 7
6 . 7
6 ..
6
7
7
only: ZDDCBE and S_DCB. The comparison data
6 . 7 includes nine stations in mid-latitudes from 1 Jan to 31
6 . 7
¼F6 . 7 ð5Þ
6 7 Jan 2015. In the second step, the effect of the geomagnetic
6 .. 7
6 . 7
6
6 ..
7
7
storm is studied by calculating the variations between quiet
6 . 7 and stormy days in satellites and receivers’ DCBs. The
6 7
6 . 7
6 .. 7 study includes data from June 2015 during a severe G4-
6 7
6 . 7
6 .. 7 class geomagnetic storm of seven IGS reference stations.
6 7
6 P ðnÞ 7 The last step introduces the variation of DCB during differ-
6 4 7
6 7
4 0 5 ent seasons. Eleven days are chosen from each month dur-
0 ing 2015 to estimate instrumental biases. The mean DCB of
each season (Winter, Spring, Summer, and Autumn) is cal-
Where DCBG r ; DCBr are the receiver DCB for GPS and
R
culated to study the seasonal change effect on DCB.
GLONASS, respectively, DCBG R
si , DCBsi are the satellites
DCB for GPS and GLONASS, respectively, Icoef is the
ionospheric coefficient that can be used to calculate VTEC, 4. Validation of the algorithm
n, and k are the number of observations, and the number
of ionospheric coefficients, The observations of nine GNSS reference stations,
namely: BOGI, BRUX, GANP, GRAS, HERT, ONSA,
f 21 f 22 POTS, SOFI, and WTZZ, shown in Fig. 3 during the per-
F ¼ Mp ðzÞ ð6Þ
40:3ðf 21  f 22 Þ iod from 1 Jan to 31 Jan 2015 (station GANP was out of
service and do not have RINEX files on 8 and 9 January),
and are used as an input data to make a comparison between
the results from the three algorithms: GR_DCB,
X
N X
n
ZDDCBE, and S_DCB. These stations include the stations
Eðb; sÞ ¼ n sinðbÞðan cosðmsÞ þ bn sinðmsÞÞ
Pm m m
ð7Þ Jin et al. (2012) used that received GLONASS observations
n¼0 m¼0
and others from the same region. This study aims to verify
As the number of satellites observed at different stations how employing weight functions and adding GLONASS
may differ, the DCB for the satellites with no observations observations affects the accuracy of DCB estimates, which
are read from the IONosphere map EXchange (IONEX) is supposed to improve results accuracy. The year 2015 was
file, and their sum is added with inverse signal replacing selected since it was the year of the geomagnetic storm
zero in observation vector (L) to keep zero difference the under study and because all GLONASS satellites are avail-
same for all stations. able throughout this year.
A weight function is used to reduce observations noise Absolute mean error and Root Mean Square (RMS) are
in addition. The weight function depends on the satellite computed for each software result to assess its accuracy as
elevation angle at each observation. The form of the follows:
weighted least square is defined by the following equation Pn  
software 
m¼1 DCBm  DCBm
 IGS
(Ghilani and Wolf, 2012): n¼
n
X ¼ ðAT xAÞ  1AT xL ð8Þ ð10Þ
3936
A. Wageeh et al. Advances in Space Research 72 (2023) 3933–3945

Fig. 2. The flowchart of the GR_DCB software.

Fig. 3. The location of the used IGS reference stations.

3937
A. Wageeh et al. Advances in Space Research 72 (2023) 3933–3945
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
uP Table 2
u n software 2
u ðDCBIGS m  DCBm Þ Mean GLONASS RDCB values from IGS, GR_DCB, absolute mean
tm¼1 error between them from January 1–31, 2015, and RMS for BRUX,
RMS ¼ GANP, GRAS, HERT, ONSA, POTS, SOFI, and WTZZ.
n 
ð11Þ Station IGS RDCB (ns) GR_DCB RDCB (ns) n(ns) RMS (ns)
BRUX 2.23 2.21 0.17 0.20
 GANP 10.43 10.32 0.19 0.23
Where n is an absolute mean error, is DCB DCBIGS
m GRAS 16.07 16.36 0.31 0.36
from solutions of the IGS Analysis Centers, DCBm software
is HERT 12.18 12.17 0.15 0.20
ONSA 9.08 8.95 0.21 0.29
the calculated DCB from the used software (GR_DCB,
POTS 15.88 15.60 0.35 0.43
ZDDCBE, or S_DCB), and n is the total number of study SOFI 4.62 5.44 0.82 0.85
days. WTZZ 8.99 9.26 0.30 0.37
Table 1 shows the absolute mean error and RMS of
GPS RDCB values calculated by GR_DCB, ZDDCBE,
and S_DCB from 1 to 31 Jan 2015. The results show that GLONASS. The results show that the GLONASS contri-
applying the weight function in GR_DCB and ZDDCBE bution reduced the mean error in all stations, although
made their RDCB results closer to IGS solutions than the impact varies. For instance, BOGI and ONSA stations
S_DCB in BOGI, BRUX, GANP, GRAS, HERT, POTS, experienced a modest improvement of about 0.03 ns, while
and WTZZ stations. Adding GLONASS observations in BRUX, GRAS, and HERT SDCB results were more con-
the GR_DCB enhanced the accuracy of RDCB against sistent in the GR_DCB results, and the absolute mean
ZDDCBE in all stations except BOGI. GR_DCB gives errors were relatively small. Notably, RDCB and SDCB
RDCB results with only 0.17 ns absolute mean error in results from ZDDCBE and GR_DCB at GRAS had
GANP and HERT with RMS at GANP of 0.20 ns. Even high error rates of 1.31 and 1.43 ns on average, respec-
with inaccurate observations in GRAS, the RDCB error tively, and the inclusion of GLONASS in GR_DCB
is only 0.49 ns from IGS compared with 2.26 ns from reduced the error to 0.43 ns only.
ZDDCBE and 2.65 ns from S_DCB, which gives a strong To ensure the consistency of SDCB calculated from dif-
advantage of adding GLONASS to GPS in reducing inac- ferent stations on the same day, the daily standard devia-
curate DCB results. Compared to other solutions, the low tion for each GPS SDCB is estimated across the nine
RDCB values RMS from the GR_DCB solution show the stations in January. The results, presented in Fig. 5, indi-
low dispersion in software results. GR_DCB max absolute cate that the variations in SDCB across stations range
mean error from IGS is 0.71 ns in SOFI with RMS 0.75 ns. between 0.2 and 0.6 ns, with a few outliers showing a high
The GLONASS RDCB values are compared to the corre- correlation between the results. Notably, satellite G 26
sponding IGS product in Table 2 to validate GLONASS had limited observations, with data available for only the
results (station BOGI has no GLONASS DCB values in first five days of the month. Overall, the results are accept-
IGS IONEX files; it could not be compared). The absolute able for the single-receiver analysis adopted in this study.
mean error and RMS are between 0.15 and 0.2 ns in HERT
and 0.35 and 0.43 ns in POTS, except for SOFI, with 0.82 5. DCB variation under varied factors
and 0.85 ns, respectively.
GPS SDCB values were obtained at each station using The DCB is known to be device (satellite or receiver)
GR_DCB, ZDDCBE, and S_DCB. Fig. 4 displays the dependent. Still, many researchers show that any change
average of all GPS satellites’ absolute mean error between in ionospheric total electron content reflects on it, as illus-
IGS and the computed SDCB values by these methods to trated in the introduction. This part discusses two main
indicate the GPS SDCB estimation accuracy after adding parameters affecting the ionosphere and the DCB estima-

Table 1
Absolute mean error and RMS of the obtained GPS RDCB values from GR_DCB, ZDDCBE, and S_DCB using the values from IGS as a reference from
January 1–31, 2015 for BOGI, BRUX, GANP, GRAS, HERT, ONSA, POTS, SOFI, and WTZZ.

Station n of GPS RDCB (ns) RMS of GPS RDCB (ns)
GR_DCB ZDDCBE S_DCB GR_DCB ZDDCBE S_DCB
BOGI 0.27 0.26 0.36 0.35 0.34 0.46
BRUX 0.19 0.23 0.28 0.23 0.31 0.36
GANP 0.17 0.22 0.27 0.20 0.27 0.33
GRAS 0.49 2.26 2.65 0.53 2.94 2.81
HERT 0.17 0.34 0.44 0.22 0.45 0.58
ONSA 0.29 0.24 0.23 0.35 0.32 0.29
POTS 0.25 0.31 0.35 0.32 0.40 0.47
SOFI 0.71 0.88 0.46 0.75 0.91 0.52
WTZZ 0.29 0.31 0.31 0.36 0.39 0.41

3938
A. Wageeh et al. Advances in Space Research 72 (2023) 3933–3945


Fig. 4. The average of the 32 GPS satellites’ absolute mean error (n) GR_DCB, ZDDCBE, and S_DCB computed DCB from 1 to 31 January 2015 for
BOGI, BRUX, GANP, GRAS, HERT, ONSA, POTS, SOFI, and WTZZ.

Fig. 5. Daily station-to-station variations of GPS SDCB across BOGI, BRUX, GANP, GRAS, HERT, ONSA, POTS, SOFI, and WTZZ for each
satellite in January.

tion: the geomagnetic storms and different environmental ceded by a slight rise in some. In BRUX, GANP, GRAS,
states over various seasons. GPS and GLONASS instru- POTS, and SOFI, the GPS RDCB values increase by a
mental DCB are calculated under different conditions to range from 0.02 to 0.88 ns in BRUX and SOFI, respec-
show the influence of each of these parameters on DCB. tively, on 22-jun from the previous day before they
decrease on the next day. The same is true about GLO-
5.1. Geomagnetic storm effect on DCB estimation NASS RDCB in POTS and SOFI. This attitude can be
explained by the fact that the geomagnetic storm has two
One of the largest solar activities of solar cycle 24 was phases: the positive phase, which caused an increase in
on 22–23 June 2015 (Augusto et al., 2018). Several Coronal some stations, and the negative phase, which was more
Mass Ejections (CMEs) were produced, resulting in a mod- robust and caused a decrease the next day. The proposed
erate to severe G4-class geomagnetic storm on 22 June with explanation agrees with Zhang et al. (2009) test results of
an interplanetary 3-hour magnetic index (Kp) greater than DCB and TEC on the three geomagnetic storms they stud-
eight between 18:00 and 21:00 UT (as shown in Fig. 6). In ied in 2004, 2005, and 2006. In GRAS, a significant change
this section, the effect of the geomagnetic storm on DCB is in GLONASS RDCB is recorded (more than 3.5 ns)
considered in this study. because the receiver was changed from TRIMBLE NETR5
The DCB values are calculated during the two storm  4.70 to TRIMBLE NETR5  4.87 on 23 June (can be
days, two days before and two days after the storm (20– checked on the IGS website ‘‘https://igs.org/imaps/sta-
26 June 2015), to show the variation in DCB for seven sta- tion.php?id=gras00fra”).
tions: BRUX, GANP, GOPE, GRAS, HERT, POTS, and To discuss the accuracy of GR_DCB SDCB results dur-
SOFI. The results for receivers obtained using GR_DCB in ing the geomagnetic storm, the SDCB results difference
Table 3 show a decrease in RDCB during the geomagnetic from IGS is introduces in Fig. 7 on June 21, 22, and 23.
storm in all GPS and GLONASS stations and may be pre- For GPS, the SDCB estimation error ranges from 1.2
3939
A. Wageeh et al. Advances in Space Research 72 (2023) 3933–3945

Fig. 6. Geomagnetic indexes from 20 Jun to 24, 2015 (universal time). (a) The interplanetary magnetic field component Bz in (nT), (b) The solar wind
speed Vp in (km/s), and (c) the Kp index.

Fig. 7. SDCB differences (ns) between GR_DCB results at BRUX and IGS IONEX for GPS (A) and GLONASS (B) satellites on 21, 22, and 23 June.

3940
A. Wageeh et al. Advances in Space Research 72 (2023) 3933–3945

Table 3
The RDCB in ns for GPS on the left and GLONASS on the right before, during, and after the geomagnetic storm (22,23 June 2015 shaded in gray) for
BRUX, GANP, GOPE, GRAS, HERT, POTS, and SOFI.
Station GPS RDCB (ns) GLONASS RDCB (ns)
20-Jun 21-Jun 22-Jun 23-Jun 24-Jun 25-Jun 20-Jun 21-Jun 22-Jun 23-Jun 24-Jun 25-Jun
BRUX 7.80 7.76 7.78 7.21 7.41 7.77 2.46 2.46 2.53 3.03 2.86 2.51
GANP 10.18 9.15 8.98 10.14 9.89 10.81 10.68 9.98 10.77 11.30 10.22 10.45
GOPE 5.01 5.47 5.14 4.96 5.14 5.35 3.18 3.92 3.35 3.38 3.24 3.48
GRAS 18.28 18.20 17.73 18.52 17.87 18.05 16.38 16.52 16.02 14.78 13.14 13.01
HERT 8.26 8.26 8.17 7.71 7.72 7.99 12.55 12.54 12.62 13.14 13.08 12.89
POTS 5.28 5.33 5.48 4.49 4.84 5.53 15.58 15.40 15.33 16.38 16.27 15.45
SOFI 14.42 14.05 14.93 14.35 14.70 13.98 5.90 6.43 5.57 5.99 5.77 6.36

Table 4
The difference in RDCB between GR_DCB and IGS IONEX in ns for GPS on the left and GLONASS on the right, before, during, and after the
geomagnetic storm (shaded in gray) for BRUX, GANP, GOPE, GRAS, HERT, POTS, and SOFI.
Station GPS RDCB difference (ns) GLONASS RDCB difference (ns)
20-Jun 21-Jun 22-Jun 23-Jun 24-Jun 25-Jun 20-Jun 21-Jun 22-Jun 23-Jun 24-Jun 25-Jun
BRUX 0.14 0.14 0.13 0.01 0.01 0.29 0.05 0.01 0.11 0.14 0.12 0.22
GANP 0.31 0.22 0.06 0.39 0.33 0.49 0.13 0.48 0.00 0.25 0.36 0.25
GOPE 0.37 0.00 0.13 0.09 0.04 0.12 0.76 0.06 0.41 0.07 0.36 0.18
GRAS 0.41 0.27 0.29 0.30 0.07 0.16 0.43 0.24 0.78 0.18 0.44 0.40
HERT 0.02 0.01 0.15 0.10 0.35 0.08 0.04 0.08 0.12 0.14 0.29 0.11
POTS 0.04 0.08 0.26 0.34 0.32 0.33 0.63 0.70 0.76 0.12 0.06 0.82
SOFI 0.84 1.27 0.30 0.46 0.34 0.96 0.75 1.25 0.50 0.28 0.29 0.74

Fig. 8. The IONEX DCB deviation for the duration of 17 to 26 June 2015 from the first day (16 June) for six GPS satellites (A) and six GLONASS
satellites (B); the stormy days are shaded in gray.

3941
A. Wageeh et al. Advances in Space Research 72 (2023) 3933–3945

to 1 ns, , while for GLONASS, it ranges from 1.4 to 2.2 behavior on stormy days except a bit greater scattering
ns. The error oscillates during the three days with no dis- for two GPS satellites agreeing with Zhang et al. (2009)
cernible trend indicating that the storm decreases the results that the deviation of SDCB is more significant on
SDCB estimation accuracy. stormy days than quiet days.
The difference in RDCB between GR_DCB and IGS
products is listed in Table 4 to consider the impact of geo- 5.2. Seasonal change effect on DCB estimation
magnetic storms on the proposed software estimation accu-
racy. The table shows no noticeable increase in stormy days A systematic variation in DCB occurs during different
compared to quiet days, as the error gets less in some sta- seasons caused by the environmental state, including tem-
tions and greater in others for both constellations. The perature and humidity (Mylnikova et al., 2015). The
maximum absolute difference on stormy days is 0.46 and RDCB of five stations: BRUX, GOPE, HERT, POTS,
0.78 ns for GPS and GLONASS, respectively. The absolute and SOFI, are calculated during the four seasons in 2015
error for stormy days RDCB is 0.22 and 0.28 ns compared to study the DCB variation from one season to another
to 0.28 and 0.37 ns for quiet days for GPS and GLONASS, and if it is the same for GLONASS as GPS or not. These
respectively. The table indicates that the introduced model five stations are chosen because they have continuous
is reliable on stormy days, and solar activities do not affect observations throughout the year without any breaks. In
results accuracy. addition to the 31 days in January, eleven days are chosen
Fig. 8 indicates a sample of the IONEX SDCBs devia- systematically from each month (one every three days)
tion from the first day (DCB of each day minus 17 June) from February to December to represent the entire year.
from 17 to 26 June for GPS and GLONASS. GPS SDCB The calculated GPS and GLONASS receiver DCB all
values are very stable with a max daily change of 0.12 ns over the year 2015 for stations BRUX, GOPE, HERT,
compared with GLONASS SDCB, which looks very diver- POTS, and SOFI are shown in Fig. 9. The maximum differ-
gent and sparse, reaching 0.2 ns daily change. No trend can ences observed between GPS receiver DCB throughout the
be established from the Figure to describe the SDCB year were 1.88, 1.58, 1.89, 3.83, and 2.44 ns for BRUX,

Fig. 9. The RDCB for GPS on the top (continuous lines) and GLONASS on the bottom (dash lines) stations computed by GR_DCB for BRUX, GOPE,
HERT, POTS, and SOFI over 2015.
3942
A. Wageeh et al. Advances in Space Research 72 (2023) 3933–3945

Table 5
The mean GPS and GLONASS RDCB in ns for seasons Spring, Summer, Autumn, and Winter during 2015 for BRUX, GOPE, HERT, POTS, and SOFI.
Station Mean GPS RDCB (ns) Mean GLONASS RDCB (ns)
Winter Spring Summer Autumn Winter Spring Summer Autumn
BRUX 7.90 7.77 7.39 7.15 2.28 2.44 2.69 2.67
GOPE 5.25 5.10 4.74 4.72 3.60 3.58 3.23 3.35
HERT 8.40 8.13 7.76 7.63 12.34 12.70 12.83 12.72
POTS 5.22 5.08 4.74 4.44 15.58 15.82 15.96 15.97
SOFI 14.74 14.52 14.13 13.90 5.53 5.79 6.08 5.99

Table 6
The GPS and GLONASS RDCB RMS in ns for seasons Spring, Summer, Autumn, and Winter during 2015 for BRUX, GOPE, HERT, POTS, and SOFI.
Station GPS RDCB RMS (ns) GLONASS RDCB RMS (ns)
Winter Spring Summer Autumn Winter Spring Summer Autumn
BRUX 0.30 0.31 0.28 0.20 0.25 0.30 0.25 0.14
GOPE 0.23 0.29 0.31 0.26 0.23 0.30 0.32 0.28
HERT 0.30 0.30 0.26 0.17 0.28 0.32 0.23 0.16
POTS 0.55 0.42 0.41 0.41 0.46 0.47 0.37 0.43
SOFI 0.42 0.36 0.26 0.34 0.30 0.45 0.22 0.27

GOPE, HERT, POTS, and SOFI, respectively. The maxi- et al. (2015) hypothesis that the RDCB seasonal variation
mum differences for GLONASS receivers in these same sta- is associated with the receiver environment. The high tem-
tions were 1.56, 1.66, 1.89, 3.16, and 2.25 ns. It is perature (Summer and Autumn) matches the minimum
noteworthy that the maximum GLONASS DCB difference seasonal DCB variation.
was very close to GPS DCB in all stations, which can be
attributed to the fact that both systems are affected by sim- 6. Conclusion
ilar temperature and humidity conditions.
The mean RDCB for each season is calculated from the This study presents a new software GR_DCB to esti-
obtained results using GR_DCB, as shown in Table 5. All mate instrumental biases Using single-station data analysis
GPS station’s mean DCB results have the lowest values in and dual-frequency observations from GPS and GLO-
Autumn and the greatest in Winter, with a difference of NASS by incorporating observations from 24 GLONASS
0.53 ns in GOPE compared to 0.84 ns in SOFI. After Win- satellites. The software is validated by analyzing data from
ter, Spring mean RDCB values are greater than Summer. 9 stations from 1 to 31 Jan 2015, and the results demon-
Same as GPS, GLONASS in Winter also gives the greatest strate a high level of agreement with IGS combined prod-
RDCB values in all stations, but it has the lowest RDCB ucts. GR_DCB is compared with two other algorithms
values in Summer in all stations except POTS. The differ- (ZDDCBE and S_DCB) for calculating both satellites
ence between Summer and Autumn in GLONASS mean and receivers using the same duration and station, giving
RDCB is between 0.37 and 0.54 ns in GOPE and SOFI, better results. Moreover, when some stations have high
respectively. The results show that GLONASS RDCB is errors in GPS DCB estimation, adding GLONASS obser-
less affected by seasonal change than GPS RDCB in the vations concretely reduces the errors and enhances the
study area for these stations; however, further analysis is accuracy.
needed to draw definitive conclusions. GR_DCB is used to investigate the influence of geomag-
It is also important to study the deviation in results dur- netic storms on the instrumental bias during June 22–23,
ing each season. Therefore, the RMS for the RDCB during 2015. The common behavior of RDCB during the geomag-
the four seasons is calculated in Table 6. The RMS is netic storm is to decrease by 0.3 to 1.0 ns for GPS and 0.5
between 0.17 and 0.55 ns for GPS in Autumn for HERT to 1.3 ns for GLONASS, with some stations noticing a
and Winter for POTS, respectively, and 0.14 and 0.47 ns slight increase in RDCB before the decrease. However,
for GLONASS in Autumn for BRUX and Spring for the results from GR_DCB and IONEX products show that
POTS. Table 6 also indicates that Autumn and Summer the storm does not affect SDCB for GPS and GLONASS.
have the least dispersion in RDCB values for both constel- The RDCB is calculated using the proposed software for
lations except GOPE, which may be affected by any other five stations for eleven days each month in 2015 to study
reason. the effect of environmental changes (humidity and temper-
The results show a significant correlation between the ature) on RDCB. Hence, the mean and RMS are estimated
ambient temperature and the seasonal RDCB variations, for each season. The mean values for GPS and GLONASS
and it is the key to the RDCB stability matching Liu RDCB are lowest in Autumn and Summer, respectively,
et al. (2020) conclusions. They also agree with Mylnikova when the temperature is high. The highest RDCB mean
3943
A. Wageeh et al. Advances in Space Research 72 (2023) 3933–3945

values for the two constellations are in Winter when the Huang, W. (2022). Enhancing GNSS by integrating low Earth orbiters.
temperature is low. The RMS for both systems is propor- September 2021, ORCID: 0000-0001-9721-8978.
Jin S.G., Luo O.F., & Park P., (2008). GPS observations of the
tional to the DCB value, with Winter having the highest ionospheric F2-layer behavior during the 20 Nov 2003 geomagnetic
RMS and conversely for Summer and Autumn. storm over South Korea. J Geod 82(12):883–892. http://doi.org/
10.1007/s00190-008-0217-x, 2008.
Declaration of Competing Interest Jin, R., Jin, S., Feng, G., 2012. M_DCB: Matlab code for estimating
GNSS satellite and receiver differential code biases. GPS Solutions 16
(4), 541–548. https://doi.org/10.1007/s10291-012-0279-3.
The authors declare that they have no known competing Jin, S.G., Wang, J., Zhang, H., Zhu, W., 2004. Real-time monitoring and
financial interests or personal relationships that could have prediction of the total ionospheric electron content by means of GPS
appeared to influence the work reported in this paper. observations. Chin. Astron. Astrophy 28 (3), 331–337. https://doi.org/
10.1016/j.chinastron.2004.07.008.
Acknowledgment Klobuchar, J.A., 1987. Ionospheric time-delay algorithm for single-
frequency GPS users. IEEE Trans. Aerosp. Electron. Syst. 23, 325–
331. https://doi.org/10.1109/TAES.1987.310829.
The RINEX data, precise orbits products, and IONEX Li, J., Huang, D., Zhao, Y., Hassan, A., 2021. Receiver DCB analysis and
products were obtained through the online archives of the calibration in geomagnetic storm-time using IGS products. Surv. Rev.
Crustal Dynamics Data Information System (CDDIS), 53 (377), 122–135. https://doi.org/10.1080/00396265.2019.1702369.
NASA Goddard Space Flight Center, Greenbelt, MD, Liu, Z., Gao, Y., 2004. Ionospheric TEC predictions over a local area GPS
reference network. GPS Solut. 8, 23–29. https://doi.org/10.1007/
USA. RINEX data: https://cddis.nasa.gov/archive/ s10291-004-0082-x.
gnss/data/daily/ GPS precise orbits (sp3): https://cd- Liu, A., Li, Z., Wang, N., Yuan, C., Yuan, H., 2020. Analysis of the short-
dis.nasa.gov/archive/gnss/products/ GLONASS precise term temporal variation of differential code bias in GNSS receiver.
orbits (sp3): https://cddis.nasa.gov/archive/glonass/prod- Measurement 153. https://doi.org/10.1016/j.measurement.2019.107448
ucts/ IONEX products: https://cddis.nasa.gov/archive/ 107448.
Ma, G., Maruyama, T., 2003. Derivation of TEC and estimation of
gnss/products/ionex/ instrumental biases from GEONET in Japan. Ann. Geophys. 21,
2083–2093. https://doi.org/10.5194/angeo-21-2083-2003.
Appendix A. Supplementary material Mannucci, A.J., Wilson, B.D., & Edwards, C.D., (1993). A new method
for monitoring the Earth’s ionospheric total electron content using the
Supplementary data to this article can be found online GPS global network. In: Proceeding of ION GPS-93. Salt Lake City:
Institute of Navigation, 1323–1332. https://hdl.handle.net/2014/36277.
at https://doi.org/10.1016/j.asr.2023.07.063. Montenbruck, O., Hauschild, A., & Steigenberger, P., (2014). Differential
Code Bias Estimation using Multi-GNSS Observations and Global
References Ionosphere Maps. Navigation, Journal of the Institute of Navigation,
61(3), 191–201. https://doi.org/10.1002/navi.64
Ansari, K., Panda, S.K., Corumluoglu, O., 2018. Mathematical modelling Mylnikova, A.A., Yasyukevich, Y.V., Kunitsyn, V.E., Padokhin, A.M.,
of ionospheric TEC from Turkish permanent GNSS Network (TPGN) 2015. Variability of GPS/GLONASS differential code biases. Results
observables during 2009–2017 and predictability of NeQuick and Phys. 5, 9–10. https://doi.org/10.1016/J.RINP.2014.11.002.
Kriging models. Astrophys. Space Sci. 363 (3), 1–13. https://doi.org/ Otsuka, Y., Ogawa, T., Saito, A., Tsugawa, T., Fukao, S., Miyazaki, S.,
10.1007/s10509-018-3261-x. 2002. A new technique for mapping of total electron content using
Augusto, C. R. A., Navia, C. E., De Oliveira, M. N., Nepomuceno, A. A., GPS network in Japan. Earth Planets Space 63–70. https://doi.org/
Raulin, J. P., Tueros, E., De Mendonça, R. R. S., Fauth, A. C., Vieira 10.1186/BF03352422.
De Souza, H., Kopenkin, V., & Sinzi, T. (2018). The 2015 Summer Razin, M.R.G., Voosoghi, B., 2017. Regional ionosphere modeling using
Solstice Storm: One of the Major Geomagnetic Storms of Solar Cycle spherical cap harmonics and empirical orthogonal functions over Iran.
24 Observed at Ground Level. Sol Phys, 293, 84. https://doi.org/ Acta Geod. Geophys. 52 (1), 19–33. https://doi.org/10.1007/s40328-
10.1007/s11207-018-1303-8 016-0162-8.
Choi, B.K., Cho, J.H., Lee, S.J., 2011. Estimation and analysis of GPS Sardón, E., Rius, A., Zarraoa, N., 1994. Estimation of the transmitter and
receiver differential code biases using KGN in Korean Peninsula. Adv. receiver differential biases and the ionospheric total electron content
Space Res. 47, 1590–1599. https://doi.org/10.1016/j.asr.2010.12.021. from Global Positioning System observations. Radio Sci. 29 (3), 577–
Choi, B., Yoon, H.S., Lee, S.J., 2018. Combined GPS/GLONASS relative 586. https://doi.org/10.1029/94RS00449.
receiver DCB estimation using the LSQ method and ionospheric TEC Schaer, S., 1999. Mapping and predicting the earth’s ionosphere using the
changes over South Korea. J. Position. Navig. Timing 7 (3), 175–181. global positioning system. Astronomical Institute, University of Berne,
https://doi.org/10.11003/JPNT.2018.7.3.175. Berne, Switzerland, Ph.D. Thesis.
Coco, D., Coker, C., Dahlke, S., Clynch, J., 1991. Variability of GPS Sedeek, A., 2020. Ionosphere delay remote sensing during geomagnetic
satellite differential group delay biases. IEEE Trans. Aero. Elec. Sys. storms over Egypt using GPS phase observations. Arab. J. Geosci. 13
27, 931–938. https://doi.org/10.1109/7.104264. (16), 1–15. https://doi.org/10.1007/s12517-020-05817-6.
Coster, A., Williams, J., Weatherwax, A., Rideout, W., Herne, D., 2013. Sedeek, A.A., Doma, M.I., Rabah, M., Hamama, M.A., 2017. Determi-
Accuracy of GPS total electron content: GPS receiver bias temperature nation of zero difference GPS differential code biases for satellites and
dependence. Radio Sci. 48 (2), 190–196. https://doi.org/10.1002/ prominent receiver types. Arab. J. Geosci. 10 (3). https://doi.org/
rds.20011. 10.1007/s12517-017-2835-1.
Ghilani, C., & Wolf, P., (2012). Elementary surveying: an introduction to Sedeek, A., (2022). Validation of regional and global ionosphere maps
geomatics-13th ed, Pearson; 14th Edition (17 Jan, 2014), ISBN-13: from GNSS measurements versus IRI2016 during different magnetic
978-0-13-375888-7. activity. Journal of Applied Geodesy. https://doi.org/10.1515/jag-
Hofmann-Wellenhof B., Lichtenegger H., and Wasle, E., (2008). GNSS – 2021-0046.
Global Navigation Satellite Systems – GPS, GLONASS, Galileo & Yasyukevich, Y.V., Mylnikova, A.A., Kunitsyn, V.E., Padokhin, A.M.,
more, Springer-Verlag, Wien New York. ISBN 978-3-211-73012-6. 2015. Influence of GPS/GLONASS differential code biases on the

3944
A. Wageeh et al. Advances in Space Research 72 (2023) 3933–3945

determination accuracy of the absolute total electron content in the Zheng, Y., Zheng, F., Yang, C., & Nie, G., (2022). Analyses of
ionosphere. Geomag. Aeron. 55 (6), 763–769. https://doi.org/10.1134/ GLONASS and GPS + GLONASS Precise Positioning Performance
S001679321506016X. in Different Latitude Regions. 1–21. http://doi.org/10.3390/
Zhang, W., Zhang, D.H., Xiao, Z., 2009. The influence of geomagnetic rs14184640.
storms on the estimation of GPS instrumental biases. Ann. Geophys.
27 (4), 1613–1623. https://doi.org/10.5194/angeo-27-1613-2009.

3945

You might also like