CC BY
64
ТЕХНИЧЕСКИЕ НАУКИ
COMPARATIVE ASSESSMENT OF TROPOSPHERIC MODELS ON BASELINE PRECISION IN A _LOCAL GPS NETWORK_
DOI: 10.31618/ESU.2413-9335.2019.1.62.87" Bazarzagd Lkhagvasuren1, Munkh-Ireedui Khasbazar1, Zorigtbaatar Amarsaikhan2
1- Department of Land Management, Mongolian University of Life Sciences,
Zaisan, 17024, Ulaanbaatar city, Mongolia 2- Department of Geodesy, Agency for Land Administration and Management,
Geodesy and Cartography, 15170, Ulaanbaatar city, Mongolia
ABSTRACT.
Satellite based positioning systems are widely used for various applications. There are several error sources that causes the estimated position and time measurements from satellite based positioning systems' receivers such as GPS and GNSS to be inaccurate. The one of the significant error sources is tropospheric delay. The troposphere is the lowest layer of the atmosphere near the earth that causes the largest error source of pseudo-lite signal as it propagates through the lower troposphere, and the thickness of it is 9-16 km. The GPS signal path through troposphere depends on temperature, pressure, height, latitude, and humidity (Gurtner, et al. 1989). It is necessarily to correct the signal delay caused by the troposphere. This paper reveals the importance of the tropospheric delay modeling for precise baseline parameter estimation by assessment of tropospheric delay models, and mapping function used to model the tropospheric delay in Mongolia at different elevation angles (0°, 5°, 10°, 15°, and 20°). The paper also discusses the advantages and limitations of these models. Trimble Geomatic Office (TGO) software is used to analyze the data in the current study.
Keywords: tropospheric delay models, baseline precision, mapping function, GPS, error
INTRODUCTION
The indispensability of the Global Positioning System (GPS), or generally the Global Navigation Satellite System (GNSS), in modern geodetic techniques and other disciplines has necessitated investigations into the various errors affecting the system. Following the emission of signals by GNSS satellites, their propagation through the atmosphere to the receiver is delayed by the different refractive indices of the various atmospheric layers (Shrestha, et al. 2003). Two main atmospheric subdivisions can be distinguished based on the way radio waves are propagated. These are: the ionosphere and the troposphere.
The ionosphere is the upper part of the atmosphere and is a region of charged particles with a large number of free electrons. It is a dispersive medium with a frequency-dependent propagation delay. The ionospheric bias can be mitigated using dual frequency GPS receivers. The troposphere however, is the lowest part of the electrically-neutral region of the atmosphere spanning from the earth's surface to about 8km at the poles and 16 km over the equator (Rizos, et.al 1997). It is non-dispersive in nature and thus delays signals in a manner completely independent of their frequencies. The troposphere is an unstable layer, with significant atmospheric turbulence due to vertical convection currents,
particularly within its boundary layer i.e. the lowest 2 km of the troposphere (Kleijer, et.al 2004). Tropospheric delays can be separated into two main components (Hofmann-Wellenhof, et.al 2008).
Unless tropospheric delay effects are corrected, the height component of positions would be inaccurate especially in space-geodesy applications including sea-level monitoring, post-glacial rebound measurements, earthquake-hazard mitigation, and crustal motion studies (Shrestha, 2003). These considerations are critical reasons for tropospheric delay modelling. It also gives credence to the incorporation of tropospheric models in most commercial GNSS processing software. The availability of different tropospheric models is a further premise for probing into the impact of each of these models on baseline processing. STUDY AREA
We used the result of dual frequency static measurement of nine GPS stations which are located Ulaanbaatar, Darkhan, Erdenet, Khovd and Arvaikheer in Mongolia. The network consists of 9 stations and various different length of 36 baselines as shown in Figure 1. The DA01, ER01, UB01, UB02, and 5026 stations are permanent GPS stations. This network gives an opportunity to analyze various baseline lengths ranging from approximately 1km to 1300km.
96* 100' 104* 108' 112'
Figure 1. Location of GPS stations
METHODOLOGY The workflow is showed in Figure 2: the first stage is to derive the tropospheric error in the reference stations from continuous observations; and second is to build proper error interpolation models; finally, the interpolated tropospheric error would be analyzed and accordingly the performance of interpolation models could be evaluated. Each baseline was treated as an individual session and processed using the following tropospheric models:
- Saastamoinen model /SAAS/
- Hopfield model /HOP/
- Niell model /NIE/
- No model applied /NONE/
These models have a variation in accuracy. The effect of an error in all models of the tropospheric delay increasing with the elevation angles.DATA
The location of stations, dual frequency receivers and their antenna types are shown in Table 2.
Table 2.
Characteristics of GPS data used in this research
Station name Station code Lattitude, deg (N) Longitude, deg (E) Height, m Receiver type Antenna type Antenna height, m
Arvaikheer ARVA 46o 07' 17.48" 103o01' 16.05" 1590.276 LEICA GX1230GG LEIAX1202GG 1.420
Darkhan DA01 49o 28' 52.75" 105o56' 36.60" 692.655 Trimble 4000SSI TRM29659.00 0.000
Erdenet ER01 49o 01' 46.40" 104o03' 44.39" 1301.849 Trimble 4000SSI TRM29659.00 0.000
GR35 GR35 49o 33' 43.29" 105o 59' 12.48" 677.508 LEICA GX1230GG LEIAX1202GG 1.384
Khovd HOVD 47o 57' 46.78" 91o 37' 28.69" 1407.966 LEICA ATX1230 LEIATX1230GG 1.040
MONM MONM 47o 54' 58.58" 106o53' 58.88" 1288.681 LEICA GRX1200+ LEIAR10 0.076
UBO1 UB01 47o 55' 16.83" 106o54' 31.67" 1278.199 Trimble 4000SSI TRM29659.00 0.000
UBZUUN UB02 47o 54' 55.70" 106o 58' 19.05" 1290.395 Trimble 4000SSI TRM29659.00 0.000
5026 UB04 47o53' 06.82" 106o20' 02.30" 1681.398 Trimble 4000SSI TRM29659.00 0.000
ANALYSIS AND RESULTS
According to the Mongolian National Geodetic Accuracy Standards and Specifications for GPS/GLONASS Relative Positioning Technique, regarding as its lengths we classified the baselines into five categories illustrates in Table 1. For used network, there are 36 baselines in which 8 are AA class, 7 are A class, 14 are B class, 3 are C-1 class and 4 are C-2 class.
Table 1.
Baseline classification
Baseline classification Baseline lenght
АА longer than 500км
А 250-500 km
В 100-250 km
С-1 20-100 km
С-2 shorter than 20 km
We used Trimble Geomatics Office (TGO) at the processing stage, a popular Commercial GNSS processing software, which is one of the various Trimble products used in over 100 countries around the world (Neal, 2008). General flowchart of data processing in TGO is shown as Fig.2.
Figure 2. Flowchart of data processing
In the following analyses, the discrepancies in the three coordinate components compared to the reference coordinates were firstly calculated. The performance of each standard tropospheric model can be characterized by the Root Mean Square Error (RMSE). It can be seen from Table 3 that by applying any standard tropo-
spheric model in the baseline estimation step, accuracies of coordinates in both horizontal and vertical components are improved. In addition, all RMSE values indicate that the Saastamoinen and the Hopfield models tend to produce more reliable baseline results than the Simplified Hopfield model.
Table. 3.
Accuracy assessment in used models
Baseline classes RMSE, м
HOP SAAS NIE None
АА 0.0218 0.0229 0.0229 0.4364
А 0.0115 0.0113 0.0115 0.2120
В 0.0263 0.0266 0.0276 0.0925
С-1 0.0147 0.0093 0.0093 0.0243
С-2 0.0188 0.0188 0.0188 0.0225
0,5 0,4
s 0,3
LU
W
1 0,2 0,1 0
Хопфилд I Саастамойнен IНейлл I NONE
АА А В С-1 С-2 Baseline Class
Figure 3. Error in baseline lengths
Table 4.
Elevation angle and its accuracy__
Elavation angle SAAS HOP NIE
0 0.022 0.690 0.453
5 0.021 0.021 0.416
10 0.017 0.017 0.407
15 0.018 0.018 0.416
20 0.021 0.021 0.423
References
1. B.Hofmann-Wellenhof, H.Lichtenegger, J.Collins, GPS theory and practice, 5th edition, New York, pp.106-115, 2009
2. Rizos.M, Businger.ST, GPS Meteorology: Mapping zenith wet delays onto precipitable water, Journal of Applied Meteorology, Vol.33, pp.379-386, 1997
3. Celestino.C.C, Sousa.C.T, Yamaguti.W, Kuga.H.W, Evaluation of Tropospheric and Ionospheric effects on the Geographic localization of data collection platforms, Hindawi Publishing, Vol.2007, No.32514, pp.1155-1166, 2007
4. Shrestha.S, Prapod.C, Impact of different
tropospheric models on GPS baseline accuracy: Case study in Thailand, Journal of Global Positioning Systems, Vol.4, No.1, pp.36-40, 2003
5. Chang. C. C, Neal.C.L, Testing on tropospheric modeling for GPS tracking stations in Taiwan, Bulletin Geodesique, Vol.64, pp.127-159, ROC, 2008
6. Kleijer.K, Improvements in GPS tropospheric delay estimation with numerical weather prediction, PhD thesis, University of New Brunswick, Canada, 2004
7. Davis.J.L, Herring.T.A, Shapiro.I.I, Effects of atmospheric modeling errors on estimates of baseline length, Radio Science, Vol.20, pp.1593-1607, 1985
Table 5.
Station heights derived from adjustment
№ Station name NONE HOP SAAS NIE
1 ERDENET 12.93.727" 1293.792 1293.73B 1293.721
DARKHAN 69Q.039 639.366 6 39.353 690.015
3 5Q2 6 ie73.424 1673.6 OB 1673-бОЭ 1673.432
4 GR35 674.77 674.705 674.693 674.75
5 MON M 1235.731 1235.776 1235.77 5 1235.732
& UBOl 1275.403 1275.403 1275.40 3 1275.403
7 UBZUUN 1237.5 S4 1237.591 12S7.59 1237.5B4
3 ARVA 1537.501 1537.525 1537.52 3 1537.502
