International Journal

on Marine Navigation

and Safety of Sea Transportation

Volume 5

Number 1

March 2011

1 INTRODUCTION

Global Navigation Satellite System is designed for

positioning, navigation, amongst other possible ap-

plications it can also be used to derive information

about the state of the atmosphere, what is now rec-

ognized as GNSS meteorology. Particularly GNSS

meteorology is the remote sensing of the atmosphere

from satellite platform (GNSS radio occultation me-

teorology) (Pavelyev et al. 2010) and ground perma-

nent stations (ground based GNSS meteorology)

(Bender et al. 2010). Continuous observations from

GNSS receivers provide an excellent tool for study-

ing the earth atmosphere. There are many GNSS me-

teorology applications: climatology, nowcasting and

4D monitoring.

The ground based GNSS meteorology is based on

the tropospheric delay, one of the results of GNSS

data processing . The tropospheric delay is repre-

sented by the Zenith Total Delay ZTD. The ZTD can

be split into hydrostatic ZHD and wet ZWD compo-

nent of the delay:

ZWDZHDZTD +=

(1)

The wet component of Zenith Tropospheric De-

lay ZWD is the foundation for computing of water

vapor content in the atmosphere. The relation be-

tween ZWD and the water vapor content in atmos-

phere is expressed by IWV (Integrated Water Vapor)

and given by the equation (Kleijer 2004):

−













′

⋅

ZWD

IWV

(2)

where R

is the specific gas constant for water

vapor, k’

, k

are refraction constants (Boudouris

1963) and T

is weighted mean water vapor temper-

ature of the atmosphere (Kleijer 2004).

The IPWV (Integrated Precipitable Water Vapor)

is computed IPWV according to relation:

IWV

IPWV

(3)

where ρ

is the water density (Mendes 1999).

The IPWV is delivered according to equations (2

and 3) from ZWD and gives the information about

contents of water vapor (2D model) above GNSS

stations. The EUREF Permanent Network (EPN:

www.epncb.oma.be) is the base of determination of

IPWV in Europe (Vedel and Huang 2004). Since

2005 EPN analysis centres ASI, BKG, GOP and

LPT delivers Near Real Time ZTD for meteorologi-

cal applications in the frame of international project

E-GVAP (EUMETNET GPS Water Vapour Pro-

gramme) (Dousa 2010).

The spatial structure and temporal behavior of the

water vapor in the troposphere (4D model) can be

modeled by using the GNSS tomography method.

The input data of GNSS tomography are: the signal

GNSS Meteorology

J. Bosy, W. Rohm, J. Sierny & J. Kaplon

Wroclaw University of Environmental and Life Sciences

ABSTRACT: GNSS meteorology is the remote sensing of the atmosphere (troposphere) using Global Navi-

tion Satellite Systems (GNSS) to derive information about its state. The most interesting information is a

delay of the signal propagation due to the water vapor content - the Slant Wet Delay (SWD). The inverse

modeling technique being concern here is the tomography. It is the transformation of the slant integrated ob-

servation of state of the atmosphere (SWD), to the three dimensional distribution of the water vapor. Over

past six years the studies on GNSS tomography were performed in the Wroclaw University of Environmental

and Life Sciences on the GNSS tomography. Since 2008 the new national permanent GNSS network ASG-

EUPOS (about 130 GNSS reference stations) has been established in Poland (www.asgeupos.pl). This paper

presents the issues of the Near Real Time troposphere model construction, characteristic of GNSS and mete-

orological data and the building of the required IT infrastructure.

Slant Wet Delays SWD, which are the results of the

GNSS data processing, the meteorological observa-

tions from synoptic stations and the Numerical

Weather Prediction (NWP) models data. The NWP

models data are also used for GNSS data verification

and calibration of the tomography model (Rohm and

Bosy 2010). The STD can be separated like (1) into

hydrostatic SHD and wet SWD components and rep-

resented by the well known relation:

( ) ( )

ZWDmZHDmSWDSHDSTD

εε

+=+=

(4)

where

is the satellite elevation angle and m

(

) and

(

) are the mapping functions (Niell 1996; Boehm

et al. 2006).

In the GNSS tomography SWD extracted from (4)

is linked with the wet refractivity N

by the given

equation:

NASWD ⋅=

(5)

where A is the design matrix.

Currently several methods exist to solve the

GNSS tomography model. The first is to add hori-

zontal and vertical constraints into the system of

equations (5) and then solve it (Hirahara 2000), the

second is to use a Kalman filter with the same equa-

tion system (Flores et al. 2000), the third is to find

the solution directly from the GNSS phase meas-

urement equation (Nilsson and Gradinarsky 2006)

and another is Algebraic Reconstruction Technique

(ART) developed by Kaczmarz (Bender et al. 2009).

The method presented in this paper uses the mini-

mum constraint conditions imposed on the system of

observation equations (5) (Rohm and Bosy 2009;

Rohm and Bosy 2010).

The wet refractivity N

is estimated from equa-

tion (5) and finally the water vapour distribution in

the troposphere (4D) represented by the water va-

pour partial pressure e and the temperature T is ex-

tracted from the formula:

−













′

(6)

where Z

-1

is an inverse empirical compressibility

factor (Owens 1967).

The new Polish national permanent GNSS net-

work (Ground Base Augmentation System) ASG-

EUPOS has been established since 2008. 17 Polish

stations equipped with GNSS receivers and uniform

meteorological sensors work currently in the frame

of the European Permanent Network (Bosy et al.

2007; Bosy et al. 2008). The ASG-EUPOS network

consists (including foreign stations) of about 130

GNSS reference stations located evenly on the coun-

try area and build network of greater density than

EPN network. This guarantees that the 4D tropo-

sphere delay and water vapor models will be more

representative for the territory of Poland.

Since 2010 the idea of integrated researches

based on the GNSS and meteorological observations

from ASG-EUPOS stations is realized in the frame

of research project entitled Near Real Time atmos-

phere model based on the GNSS and the meteoro-

logical data from the ASG-EUPOS reference sta-

tions on the territory of Poland. The paper presents

in the second section the methodology of NRT at-

mosphere models construction procedures. The se-

cond section encloses proposal of the method of wa-

ter vapor distribution in space and time (4DWVD)

using GNSS tomography technique. The third sec-

tion includes the ASGEUPOS system description

and sources of GNSS and meteorological data, local-

ization and accuracies. The fourth section contains

the specification of IT infrastructure for NRT data

streaming and processing. The paper is closed in

fifth section with conclusions.

2 NEAR REAL TIME ATMOSPHERE MODEL

The GNSS and meteorological observations form

ASG-EUPOS stations are the base of near real time

models of tropospheric delay and water vapor (NRT

ZTD and NRT ZWD) in atmosphere. Figure 1 shows

the diagram of NRT ZTD and NRT ZWD models

construction (Bosy et al. 2010).

Figure 1: The diagram of NRT ZTD, ZWD and IPWV models

construction on the base of GNSS and meteorological data

from ASG-EUPOS reference stations

The NRT ZTD will be obtained from the NRT so-

lution of ASG-EUPOS stations network. The strate-

gy of NRT solution will be realized according to

standards used for global IGS and regional EPN

permanent GNSS networks and NRT solution strat-

egy created in the frame of COST Action 716 (Eu-

ropean Cooperation in the field of Scientific Tech-

nical Research-exploitation of ground-based GPS for

climate and numerical weather prediction applica-

tions, 1998-2004), TOUGH (Targeting Optimal Use

of GPS Humidity Data in Meteorology,

http://tough.dmi.dk/, 2003-2006) and E-GVAP (The

EUMETNET GPS Water Vapour Programme,

http://egvap.dmi.dk, 2004-2008) projects (Dousa

2004; Dousa 2010). The ZHD for all ASG-EUPOS

stations will be estimated in NRT mode on the base

of meteorological observation of Polish EPN sta-

tions equipped with meteorological sensors. Next

according to relation (1) the values of ZWD will be

computed. The IWV and IPWV values above all

ASG-EUPOS stations will be calculated from equa-

tions (2) and (3) and finally NRT ZWD and NRT

IPWV models for Poland territory will be construct-

ed (Bosy et al. 2010).

The spatial structure and temporal behavior of the

water vapour in the troposphere (4D) can be mod-

eled using the GNSS tomography method. The

GNSS signal delays due to the water vapour are

evaluated for a large number of different views

through the atmosphere (Bender and Raabe 2007).

The idea of GNSS tomography for Poland is pre-

sented in the figure 2 (Bosy et al. 2010).

Figure 2: The ray path in consecutive voxels. Two cases are

considered, the first when the ray is coming out of the model’s

side face (sf), and the second, when ray is comming out of the

model top bounduary (tb)

The input data of GNSS tomography are: the sig-

nal Slant Wet Delays SWD, which are the results of

the GNSS data processing, the meteorological ob-

servations from ASG-EUPOS meteo stations. The

quality and quantity of the GNSS observations is

strongly correlated with the GNSS satellites constel-

lation, the number of the ground stations and inter-

station distances. As a result of the GNSS data pro-

cessing the cut off angle of SWD observations are set

to 3 – 5 degrees of elevation. Moreover, the empiri-

cal results show that the angles are ranging from 5 to

85 degrees, but most of the observations are clus-

tered between 10 and 15 degrees. In case of GNSS

troposphere tomography it is typical condition

(Bender and Raabe 2007). In the GNSS tomography

SWD is linked with wet refractivity N

by the equa-

tion (5). One of the method to resolution of equation

(5) is the authors method (Rohm and Bosy 2009;

Rohm and Bosy 2010) presented briefly below.

To find the voxels’ (Fig. 2) refractivities one

needs to invert the equation (5), which in theory

might be solved by the means of the least squares

method:

( )

TTT

SWDPAAPAN ⋅⋅⋅⋅⋅=

−1

(7)

where the A is design matrix (5) and P is a weighting

matrix. The weighting matrix P is constructed as an

inversion of covarince matrix of observations SWD

given by the formula:

1−

SWD

(8)

To get 3D picture of the wet refractivity N

the

Singular Value Decomposition (SVD) technique is

used. The SVD technique is the pseudo inverting of

system (5) on the base of factorization of the vari-

ancecovariance matrix in the equation (7).

( )

USVAPA ⋅⋅=⋅⋅

(9)

where U is a n×n orthogonal matrix of left-singular

vectors, V is a m×m orthogonal matrix of right sin-

gular vectors, S is a n×m diagonal matrix of singular

values sorted in descending order (Anderson et al.

1999) and S

is a pseudoinverse of the matrix S.

3 ASG-EUPOS GNSS AND

METEOROLOGICAL DATA

The GNSS data are currently available from the

GNSS permanent stations operated in the frame of

national networks. In Poland the national permanent

GNSS network ASG-EUPOS has been established

since 2008. The receiving segment (ground control

segment) consists of a network of GNSS reference

stations located evenly on the whole territory of Po-

land. Comply with EUPOS and project of the ASG-

EUPOS system standards distances between neigh-

boring reference stations should be 70 km what

gives number of stations 98 (3). According to rules

of EUPOS organization (in the frame of cross-border

data exchange) 3 reference stations from Lithuania

(LITPOS), 6 stations from Germany (SAPOS), 7 sta-

tions from Czech Republic (CZEPOS) and 6 stations

from Slovakia (SKPOS) were added (Fig. 3) (Bosy

et al. 2008).

Figure 3: Reference stations included of ASG-EUPOS system

(www.asgeupos.pl)

The reference stations of ASG-EUPOS system

(Fig. 3) are equipped with themodern GNSS receiv-

ers and the antennas with absolute calibrations. In

the 14 localizations of the EPN stations the new uni-

form meteorological infrastructure Paroscientic, Inc.

MET4A sensors were installed. In the EPN/IGS sta-

tion Borowiec (BOR1) the equivalent meteorologi-

cal sensors: NAVI Ltd. HPTL.3A and Skye Instru-

ments Ltd. are installed.

Meteorological observations from ASG-EUPOS

stations (Bosy et al. 2010) will be used for ZHD ex-

traction from equation (1), to compute ZWD and fi-

nally SWD (4). The external meteorological observa-

tions from Institute of Meteorology and Water

Management (IMGW) synoptic stations, radiosound-

ings observations and NWP COAMPS model out-

puts will be used also for verification of GNSS to-

mography model (Rohm and Bosy 2010).

4 IT INFRASTRUCTURE

Meteorological data from stations dispersed on the

area of study are collected with GNSS data and put

into ASG-EUPOS caster. Then using the NTRIP

(Networked Transport of RTCM via Internet Proto-

col), data are transferred to the center that deals with

NRT (Near Real Time) processing. The Internet is

particularly well suited to transmit data between dif-

ferent providers over long distances. However, the

servers required must be tied to the Internet via in-

terconnected broadcasters with sufficient bandwidth.

In order to secure resources the infrastructure must

be equipped with firewall and IDS (Intrusion Detec-

tion System) / IPS (Intrusion Prevention System) el-

ements (Fig. 4).

Figure 4: IT infrastructure for data streaming and processing

The GNSS and meteorological data from Man-

agement Centre (MC) of ASG-EUPOS are decoded

and sent both to compute clusters and database

(Fig. 4). Despite the effectiveness of the algorithms

computations require compute clusters. That is why

one needs such computer cluster, which enables par-

allel computations. NRT processing should be

equipped with back-up system, which connects with

other devices through dedicated high-speed fiber

channel network. The above presented conception is

still very general and will be detailed during tests

and the project realization.

5 CONCLUSIONS

Ground based GNSS meteorology currently utilizes

the height resolution GBAS networks like AS-

GEUPOS, where reference stations are equipped

with GNSS and meteorological sensors. The NRT

troposphere model based on the GNSS and meteoro-

logical data of GBAS system could be used as well

as in meteorological applications, in the real-time

and post-processing positioning services of AS-

GEUPOS system. The model created from meteoro-

logical and GNSS data, could be competitive to

Numerical Weather Prediction models, especially

for nowcasting. The improvement in positioning is

that tropospheric delays will be calculated directly

from observations, not like now from deterministic

models.

ACKNOWLEDGEMENTS

This work has been supported by the Polish Ministry

of Science and Higher Education: research project

No N N526 197238.

REFERENCES

Anderson, E., Z. Bai, C. Bischof, S. Blackford, J. Demmel,

J. Dongarra, J. Du Croz, A. Greenbaum, S. Hammarling,

A. McKenney, and D. Sorensen (1999). LAPACK Users’

Guide, Third Edition. Society for Industrial and Applied

Mathematics.

Bender, M., G. Dick, J. Wickert, M. Ramatschi, M. Ge,

G. Gendt, M. Rothacher, A. Raabe, and G. Tetzlaff (2009).

Estimates of the information provided by GPS slant data

observed in Germany regarding tomographic applications.

J. Geophys. Res. 114, D06303.

Bender, M. and A. Raabe (2007). Preconditions to ground

based GPS water vapour tomography. Annales Geophysicae

25(8), 1727–1734.

Bender, M., R. Stosius, F. Zus, G. Dick, J. Wickert, and A.

Raabe (2010). Gnss water vapour tomography – expected

improvements by combining gps, glonass and galileo ob-

servations. Advances in Space Research In Press, Correct-

ed Proof, –.

Boehm, J., A. Niell, P. Tregoning, and H. Schuh (2006). Glob-

al Mapping Function (GMF): A new empirical mapping

function based on numerical weather model data. Geophys.

Res. Lett. 33, 15–26.

Bosy, J., W. Graszka, and M. Leonczyk (2007). ASGEUPOS.

A Multifunctional Precise Satellite Positioning System in

Poland. European Journal of Navigation 5(4), 2–6.

Bosy, J., A. Oruba, W. Graszka, M. Leonczyk, and

M. Ryczywolski (2008). ASG-EUPOS densification of

EUREF Permanent Network on the territory of Poland. Re-

ports on Geodesy 2(85), 105–112.

Bosy, J., W. Rohm, and J. Sierny (2010). The concept of the

near real time atmosphere model based on the GNSS and

the meteorological data from the ASG-EUPOS reference

stations. Acta Geodyn. Geomater. 7(1(157)), 1–9.

Boudouris, G. (1963). On the index of refraction of air, the ab-

sorption and dispersion of centimeter waves by gases.

Journal of Research of the National Bureau of Standards

67D(6), 631684.

Dousa, J. (2004). Evaluation of tropospheric parameters esti-

mated in various routine GPS analysis. Physics and Chem-

istry of the Earth, Parts A/B/C 29(2-3), 167 – 175. Probing

the Atmosphere with Geodetic Techniques.

Dousa, J. (2010). The impact of errors in predicted GPS orbits

on zenith troposphere delay estimation. GPS Solutions.

Flores, A., G. Ruffini, and A. Rius (2000). 4D tropospheric

tomography using GPS slant wet delays. Annales Geophys-

icae 18(2), 223–234.

Hirahara, K. (2000). Local GPS tropospheric tomography.

Earth Planets Space 52(11), 935–939.

Kleijer, F. (2004). Troposphere Modeling and Filtering for

Precise GPS Leveling. Ph. D. thesis, Department of Math-

ematical Geodesy and Positioning, Delft University of

Technology, Kluyverweg 1, P.O. Box 5058, 2600 GB

DELFT, the Netherlands. 260 pp.

Mendes, V. B. (1999). Modeling the neutral-atmosphere prop-

agation delay in radiometric space techniques. Ph. D. the-

sis, Deparment of Geodesy and Geomatics Engineering

Technical Reort No. 199, University of New Brunswick,

Fredericton, New Brunswick, Canada.

Niell, A. E. (1996). Global mapping functions for the atmos-

phere delay at radio wavelenghs. J. Geophys. Res. 101(B2),

3227–3246.

Nilsson, T. and L. Gradinarsky (2006). Water Vapor Tomogra-

phy Using GPS Phase Observations: Simulation Results.

IEEE Trans. Geosci. Remote Sens. 44(10 Part 2), 2927–

2941.

Owens, J. (1967). Optical refractive index of air: dependence

on pressure, temperature and composition. Appl. Opt. 6(1),

51–59.

Pavelyev, A., Y. Liou, J.Wickert, T. Schmidt, and A. Pavelyev

(2010). Phase acceleration: a new important parameter in

gps occultation technology. GPS Solutions 14, 3–11.

10.1007/s10291-009-0128-1.

Rohm, W. and J. Bosy (2009). Local tomography troposphere

model over mountains area. Atmospheric Research 93(4),

777 – 783.

Rohm, W. and J. Bosy (2010). The verification of gnss tropo-

spheric tomography model in a mountainous area. Advanc-

es in Space Research In Press, Corrected Proof, –.

Vedel, H. and X. Huang (2004). Impact of ground based GPS

data on Numerical Weather Prediction. J. Meteor. Soc. Ja-

pan 82(1B), 459–472.