Weather Report
Meteorological Applications of GNSS from Space and on the Ground Authors from France’s national meteorological agency and the French space agency describe the techniques and applications that use the behavior of GNSS signal propagation through the Earth’s atmosphere to predict the weather and monitor climate change.
Paul Poli, Jean Pailleux, Veronique he year 2007 marked the 50th using data collected by GPS receivers Ducrocq, Patrick Moll, Florence Rabier anniversary of the first man- located on the ground or in space. Météo-France, Centre National de made artificial satellite carrying Experiments exploiting these Recherches Météorologiques (CNRS-GAME) T a radio transmitter. Launched ground- or space-based GPS techniques Michel Mauprivez, Sylvie Dufour, by the former Soviet Union in 1957, the were first reported as far back as 1992 Météo-France, Direction des Systèmes Sputnik satellite preceded the first con- and 1995, respectively. (For further dis- d’Observation, DOA/DA stellation of GPS satellites by more than cussion of these pioneering efforts, see three decades. the articles by M. Bevis et alia [1992] Michel Grondin, Françoise Carvalho, Beginning with launch of its first sat- and E. R. Kursinki et alia [1995] in the Jean-Luc Issler, Antoine de Latour, ellite in 1978 through declaration of full Additional Resources section at the end Lionel Ries operational capability in 1995, the U.S. of this article.) Similar measurements Centre National d’Etudes Spatiales (CNES), Global Positioning System has broadcast are now routinely available to national Service Technique de Transmissions et accurate and stable radio signals. These meteorological services worldwide. Traitement du Signal GPS signals have now become a source of The measurements collected by these information for exploratory and routine two techniques can be useful to meteo- monitoring of the Earth’s atmosphere, rologists provided that ancillary (exter-
30 InsideGNSS november/december 2008 www.insidegnss.com nal) information is available with which The ZTD can be expressed as ZTD information comes from the part to retrieve such data as integrated water of the atmosphere located below three vapor (for ground-based GPS measure- kilometers altitude. ments) or profiles of temperature and The inversion of a measured ZTD humidity (for space-borne GPS mea- where n(h) denotes the atmospheric into retrieval of an integrated water surements). refractive index as a function of height. vapor (IWV) content requires indepen-
However, the addition of such infor- The atmospheric refractive index n at dent estimates of surface pressure (Ps) mation convolves the error patterns in radio frequencies is usually expressed and mean atmospheric column temper- the meteorological retrievals, making in terms of the radio refractivity N = ature (Tm). The surface pressure enables them more difficult to use in an error- (n-1)·106. In the neutral atmosphere, the us to estimate the so-called hydrostatic reduction scheme such as those com- Smith and Weintraub formula derived in contribution to the ZTD (about 2.3 mil- monly employed in numerical weather 1953 relates radio refractivity N to pres- limeters of delay per one hectopascal of prediction (NWP) and assimilation sys- sure P, temperature T, and water vapor surface pressure), while the remaining tems. For that reason, NWP assimilation partial pressure e as follows: “wet” delay contribution can be scaled makes direct use of data closer to the raw via a conversion factor into IWV. measurements, for which the error pat- The conversion factor is Tm-depen- terns are better characterized. dent but is typically about 6.5 millime- -1 4 This article presents the method- where b1 = 77.6 K.hPa and b2 = 37.3·10 ters of delay per one millimeter of IWV. 2 -1 ology for assimilating the GPS-based K .hPa . More recent research has Although two pieces of information (Ps, measurements into the NWP system reevaluated that relationship, taking into Tm) are thus required to yield the meteo- of the French national weather service account the increase of the CO2 concen- rological product (IWV) from the ZTD, (Météo-France). We describe the error tration observed in the atmosphere. this technique is valid in all weather reductions obtained by this method in The temporal resolution of ZTD esti- conditions and can be useful to moni- estimating atmospheric conditions and mates is typically between 5–60 minutes tor in near real-time maps of IWV when making subsequent weather forecasts. but it should be noted that the errors in a network of GPS receivers is available. The article further considers the ZTD are correlated in time. The area Such maps can, for example, be used by application of these observations for cli- sensed by a “ZTD measurement” extends weather forecasters in conjunction with mate monitoring. Finally, we discuss the in the horizontal up to about a 17-kilo- radar and satellite measurements to help perspectives on these questions that will meter radius around the station. Note, track severe weather events. be allowed by the future Galileo system this assumes that the ZTD is obtained For example, in the United States, (as, of course, the perspectives allowed from raw measurements from various IWV data are derived in near real-time by Compass, GLONASS, IRNSS, etc., GPS satellites at equally distributed from ZTD data collected by a vast net- could have been similarly discussed). azimuth angles and down to 10 degrees work of ground-based GPS stations cov- elevation angle, and that most of the ering the continental United States. In Ground-Based GPS Measurements (a) (b)b) As mentioned earlier, ground-based sta- GPS3 φ φ ξ tions monitoring the signals broadcast GPS2 1 2 GPSGPS Vertical S A dξξ receiverrecec iviv by the GPS transmitters have been used line-of-sight GPS4 GPSPS dl B hmin h dhd since 1992 for meteorological purposes. transmittermitteer p f The ionosphere and the neutral atmo- dff sphere both induce a delay in the propa- GPS1 a surfacesurface gation of the GPS signals as compared to propagation in a vacuum. Antenna Because of the dispersive nature of GPS Receiver the ionosphere, measurements made 0 on the two GPS frequencies (L1 and FIGURE 1 Geometry of GPS meteorological observations (not to scale): (a) with a ground-based GPS L2) enable observers to assess a neutral receiver, the zenith total delay (ZTD) observation characterizes the total atmospheric delay at atmosphere-only delay. Several GPS the vertical of the GPS receiving station, but several GPS links are in fact required, as illustrated links (see Figure 1a) are required to here with several hypothetical GPS satellites viewed under different azimuth and zenith angles; (b) with a spaceborne GPS receiver in LEO orbit (between 400-1600 km altitude), a GPS radio retrieve a zenith total delay (ZTD) above occultation involves a GPS transmitter located in MEO orbit at about 20,000-kilometer altitude; the receiving station, using to this end the radio link is incident with an asymptotic ray-miss distance p with respect to the local centre of mapping functions such as described in curvature of the Earth O. Because of refraction this ray is bent towards the Earth’s surface and the distance of closest approach is the tangent height hmin. The contribution of an atmospheric slab of the article by A. E. Niell cited in Addi- thickness dh at altitude h to the bending angle ξ for that ray is noted d ξ. tional Resources. www.insidegnss.com november/december 2008 InsideGNSS 31 weather report
Japan, a dense network of GPS stations was established, originally From the receiver point of view, the wave fronts arrive for geodetic purposes, but may now also serve meteorology. with increasing timing separations (the result of atmospheric- In Europe, several initiatives have helped gather exper- induced Doppler shift on the received frequencies). Provided tise and methods so that ZTD data from European GPS sta- the kinematic Doppler shift caused by the differential motion of tions belonging to various networks can be made available in the platforms can be calculated with the help of accurate trans- near real-time for meteorological applications and national mitter and receiver positions and velocities, the atmospheric meteorological services’ use worldwide. These include Action contribution can be assessed from the measured carrier Dop- 716 (“Exploitation of Ground-Based GPS for Climate And pler shifts at the receiver. Note that in practice single-differenc- Numerical Weather Prediction Applications”) of the European ing at the spaceborne receiver may be required to correct for Cooperation in the Field of Scientific and Technical Research receiver clock errors. (COST-716) along with the following projects: Meteorological In 1995, the first GPS radio occultation proof-of-concept Applications of GPS Integrated Column Water Vapour Mea- experiment — GPS/MET — was conducted, followed by simi- surements in the Western Mediterranean (MAGIC), Target- lar experiments onboard satellites, e.g., CHAMP (Challenging ing Optimal Use of GPS Humidity measurements in meteo- Minisatellite Payload), SAC-C (Satellite Argentina Científicas rology (TOUGH), and most recently the EUMETNET GPS – C), GRACE (Gravity Recovery and Climate Experiment), and water vapor program (E-GVAP, on the web:
32 InsideGNSS november/december 2008 www.insidegnss.com FIGURE 2 Maps of GPS observations received at Météo-France in near real-time on 8 July 2007 00 UTC (+/- 3 hours) : (left) European ground-based GPS ZTD observations (analysis centers indicated at the top along with the number of stations processed by each – a few stations fall outside the map area) ; (right) GPS radio occultation from several satellites (total counts per satellite indicated at the top ; courtesy of Météo-France DPREVI/COMPAS
Assuming spherical symmetry, the integration of Bouguer’s appears more obvious for the occultations in the polar regions, formula enables us to relate the profile of atmospheric refractive but this is an artefact of the projection. index to the bending angle ξ observed for a ray with impact parameter p and tangent height hmin, noting a the local radius NWP Applications and Methodology of curvature of the Earth: Let us now turn to the practical application of these methods in numerical weather prediction, where we begin with a closer look at specific methodologies, including the assimilation of data into NWP models and its application by the French mete- To the first order, spherical symmetry is a reasonable orological agency, Météo-France. We will then more closely approximation for modelling the refraction of the radio path examine separate GPS ground-based and space-based (radio within the Earth’s atmosphere. However, this approximation occultation) approaches. is known to lead to important errors when the radio signal Operational NWP requires a regular update of the atmo- encounters strong horizontal gradients of refractive index. spheric state model estimate. This process (so-called analysis) Within an occultation plane, for example, using a global atmo- is typically conducted four times a day (00, 06, 12, 18 UTC) spheric model that featured a horizontal resolution of 1x1.25 with the help of meteorological observations and an a priori degree (latitude, longitude), this error has been estimated to estimate. The a priori estimate is usually a forecast generated average less than 1.5 percent RMS of the bending angle. by a circulation model run from the previous analysis. This finding is consistent with the error reduction obtained Such meteorological observations have a diverse nature. in the fit to observed SAC-C and CHAMP bending angles when They include, for example, in situ measurements from radio- a two-dimensional ray-tracing scheme was used to carry the sondes, aircraft, buoys, ground and ship stations, as well as refraction calculations from that same global model. We should satellite-based remote sensing measurements such as infrared point out that larger effects may be detected with a higher and microwave radiometers, cloud-track winds from imagery, resolution model (closer to real atmospheric gradients). Other and surface winds from scatterometers. research, using simulations from a meso-scale atmospheric The method employed at Météo-France for performing the model (horizontal resolution of 12 kilometers) over the Unit- analysis in its global NWP model and assimilation system is ed Kingdom, has shown that out-of-plane bending can also a four-dimensional variational (4DVAR) assimilation. More occur. information about the 4DVAR methodology is given in the Another aspect of the retrieval process that can lead to article by F. Rabier et alia (see Additional Resources). Briefly, errors because of departure from spherical symmetry is the the 3D model state (noted x) at analysis time is estimated from drift of the tangent point during an occultation. This can a variety of meteorological observations (noted y0) by minimiz- lead to larger errors in the modelling of bending angles, up ing the following cost function: to 15 percent RMS in terms of bending angle. Figure 2b shows the individual tangent point locations for which a retrieved bending angle was measured by the CHAMP, GRACE, and where B denotes the a priori error covariance matrix, R denotes FORMOSAT-3/COSMIC experiments. The tangent point drift the observation measurement and representativeness error www.insidegnss.com november/december 2008 InsideGNSS 33 weather report
covariance matrix, and H is an observa- France’s NWP system is detailed in investigated previously. Over three dif- tion operator used to calculate simulated the article by P. Poli et alia (Additional ferent seasons the effect was most appar- observations from the a priori informa- Resources). ent in improved geopotential heights tion x. The standard deviation of the differ- and, especially, in improved quantita- This operator H involves, for exam- ences between model calculations and tive precipitation forecast scores over ple, atmospheric radiative transfer the- observations is the sum of the a priori France. ory in order to model brightness tem- error variance and observation error Meanwhile, data collected by the perature observations. The additional variance, if correlations in time are dis- GPS radio occultation technique have constraint Jc serves to avoid non-physi- regarded. This standard deviation exhib- been available in near real-time to cal analysis solutions. In 4DVAR, the its large seasonal variations with stron- national meteorological services since observation operator involves a space ger standard deviations in the summer February 2007. Several impact studies interpolation from the 3D model fields, when more water vapor is present. of these data have been conducted at with the model state propagated to the The issue of time correlations is par- Météo-France. proper observation time with the help ticularly important, as GPS solutions For example, data presented in Figure of the forecast model. are not derived from individual mea- 3 show that the greatest positive impact The NWP analysis is essentially a surements but usually with a sliding can be found in analyses of the Southern filtering process by which one tries to window. The assimilation scheme used hemisphere’s troposphere. Note that in extract information from the obser- at Météo-France assumes observation situ measurements are less dense and vations in order to reduce the errors error variances that are station-depen- forecast error is typically greater for in the a priori state. The underlying dent and were derived using the method Météo-France’s global model in that assumption is that weather forecasting described in the article by G. Desroziers region. is an initial condition problem; hence, et alia (Additional Resources). Overall, these demonstrated benefits for a given weather forecast model, the Space-Based GPS: Radio Occultation. have resulted in Météo-France using the improvement of the initial conditions The observation operator for simulat- ZTD data and GPS radio occultations should improve the forecast results. ing GPS radio occultation bending for its NWP operations since September The minimization of the cost func- angles derives from work conducted at 2006 and September 2007, respectively. tion in model space relies on the follow- the European Centre for Medium-range ing key assumptions: all error sources Weather Forecasts. (Additional Resourc- Applications for Climate present a zero-mean Gaussian distri- es: article by S. B. Healy and J. N. Thé- Atmospheric observations can serve bution and are uncorrelated between paut). It involves the integration of equa- climate-monitoring purposes provided one another (except in 3D space for a tions (3) and (2) using the model fields, that they feature a proven high accuracy given a priori variable type), only small without applying a bias correction. and a long-term stability. With GPS- departures from linearity are allowed, The observation error variance based measurements, the raw observ- and the error variances are correctly estimates were obtained using the G. able is a number integer (and fraction) specified in the matrices B and R. Any Desroziers et alia method mentioned of the GPS wavelength, whose stabil- departure from these assumptions may previously. This method has also helped ity is guaranteed by the atomic clocks lead to sub-optimal solutions and the show that vertical error correlations are onboard the GPS satellites, which are misinterpretation of noise as signal (or present, possibly because of the smooth- themselves calibrated daily with atom- vice versa). ing process at Doppler shift level as sug- ic clocks on the ground. Hence, GPS- Ground-Based GPS. The observation gested earlier on by simulations. Because based measurements of atmospheric operator for simulating ground-based the data assimilation system used here delay are calibrated indirectly with GPS ZTD observations involves the assumes uncorrelated observation errors, atomic clocks. integration of equations (1) and (2) using we revert to a thinning (data selection) At this stage of development in the the model temperature, humidity, and procedure for the bending angle obser- measurement process, no other meteo- pressure above the vertical of each GPS vations, retaining only one observation rological metric available today can station location. per model level and per occultation event claim such a calibration feature. How- For each GPS station, the mean dif- (that is, ~20 observations per occultation ever, these atmospheric-delay measure- ferences between the model calculations up to 18 kilometers altitude). ments (whether from space or from the and observation are usually non-zero, ground) do not represent a quantity that indicating that either the model or the Effect of GPS-Derived Data can be used directly to characterize the observations (or both) present non-zero on NWPs climate. mean errors. A bias correction is applied As mentioned earlier, the influence of Each step of the inversion process using a mean of 10 days differences. The European ground-based GPS ZTD data beyond this point of measuring signal- methodology and results of assimilation introduced into the Météo-France global propagation delay requires particular of ground-based GPS ZTD in Météo- forecasting and assimilation system was scrutiny in order to demonstrate that the
34 InsideGNSS november/december 2008 www.insidegnss.com FIGURE 3 Forecast score differences for 41 forecasts (March-April 2007) to four days: RMS forecast errors from the baseline Météo-France global forecasting system (which did not assimilate any GPS radio occultation data at the time) minus similar RMS forecast errors (but when assimilating GPS radio occultation bending angles), using radiosondes as a verification, for geopotential height (in geopotential meters), wind speed (in meters/ second), and temperature (in Kelvin). Solid (dotted, dashed) lines indicate a positive (zero, negative, respectively) effect, i.e., an RMS forecast error reduction (no difference, increase, respectively). The Northern hemisphere (Tropics, Southern Hemisphere) geographical domain relates to the area located between 20oN–90oN latitude (20oS–20oN, 20oS–90oS, respectively). atomic-calibration feature is not lost to they require external meteorological the long-term stability of ZTD retrievals. greater error sources. observations, such as surface pressure Yet, the remaining advantage of ground- Readers should note that, in the fol- and mean temperature. The stability of based GPS for climate monitoring is that lowing discussion, we do not raise the these external data can only be achieved the raw data collection method “only” questions of (1) whether the trends that through a separate process (in particular needs to care about properly saving are observed with ground-based and by means of periodic calibration of the metadata (receiver, clock, and antenna radio occultation GPS are meaningful associated sensors). information at each receiving station, or consistent with other measurement Focusing thus on ZTD only, this assuming that the GPS satellite visibil- sources, or (2) whether the projected ground-based method of IWV retriev- ity mask remains unchanged at the sta- trends of climate change would incur al features several sources of error, tions). signals on GPS-based measurements amongst which is the ionospheric Consequently, a retrieval process that would be within today’s estimated contribution correction whose errors (analysis method) conducted after the GPS measurement accuracy. Instead, we depend on the 11-year solar cycle. The fact —for example, after a long time discuss the reverse problem of assessing homogeneity and stability (number series of raw data has been collected — the error sources that need to be quan- and position of the sites) of the overall would need a particular strategy in order tified in order to guarantee that GPS- network used for performing the ZTD to extract a climatologically meaningful based measurements present irrefut- retrievals could also influence the sta- time series of ZTDs measured above a able stability and can be used for climate bility of the particular ZTDs retrieved selection of GPS stations featuring con- monitoring. from above a given site. The accuracy sistent equipment over time. From the ground, obtaining a meteo- of the GPS orbits would also need to be A spaceborne GPS approach to cli- rologically meaningful ZTD requires a stable over time. mate monitoring removes some of the network of GPS receiving stations. More- Obviously all these errors sources constraints associated with possible over, IWV retrievals represent a step up would need to be taken into account in environmental influences on the mea- in difficulty for this application because a comprehensive study in order to assess surement, only to be replaced by oth- www.insidegnss.com november/december 2008 InsideGNSS 35 weather report
View of ESA PROBA2 microsatellite
(right) L1-L2C receiver delivered for PROBA2 ers. For example, the variation of sat- not be tracked all the way during an This receiver will be used for the ellite orbits and of the number of GPS occultation because of its low signal-to- first time in orbit in 2009 on board the transmitters over time can influence the noise ratio (thus requiring an extrapola- European Space Agency’s PROBA-2 mean local time (sampling) conditions tion). LEO microsatellite. The receiver’s mass for a given location. Quantitative upper bounds for cli- is close to one kilogram, making it easy Also, although the hardware on a mate-observation candidates have been to use on board small platforms. given receiving satellite will not change determined in accordance with the lev- The CNES R&D program enabled over time, studies of current missions els of atmospheric signal one attempts preparation of a detailed specification have shown that the on-board software to detect for various parameters. (See for the atmospheric radio occultation used to track occultations can influence article by G. Ohring et alia cited in software to be loaded or uploaded later the quantity of measured events. This Additional Resources.) For example, the onto this type of receiver. Météo-France affects event sampling and can cause an accuracy of temperature measurements assisted development of this specifica- artificial sampling bias change. for climate should be better than 0.5 K tion by supplying an atmospheric-delay Assuming that a receiver’s orbit- and their stability should be better than dynamic model appropriate for radio determination accuracy is not a limit- 0.04 K per decade in the troposphere occultation, to properly tune the high ing factor (below 0.05 mm/s), the first (i.e., less than 4·10-3 K/year). The article rate, phase locked loop tracking the source of error that would need to be by A. J. Mannucci et alia (and references GPS carriers. The realization of this evaluated more carefully is the removal therein) cited in Additional Resources radio-occultation software will start in of the ionospheric contribution (signal reports on first efforts to characterize early 2009. delays). GPS radio occultation measurement sta- The deployment of the GPS satellites This source of error (whose effects bility and accuracy with a traceability to broadcasting L2C signals is quite rapid, would become apparent over the course the raw measurement. with 22 such spacecraft expected to be in of 11-year solar cycles) can influence the orbit within five years. The building up quality of atmospheric refraction mea- New Receivers and Signals of a full constellation of L2C-transmit- surements for three reasons: because a To improve GPS radio occultation mea- ting satellites is important, because the combination of signals on two frequen- surements and precise orbit determi- U.S. government has announced plans cies may only correct for the first term nation with GPS (among other tasks), to discontinue guaranteed access to the of the expansion of the Appleton-Har- the European Space Agency developed legacy L2 signals for use in so-called tree equation for ionosphere dispersion with CNES support a qualified, second- codeless or semi-codeless techniques. (thus leaving higher-order, frequency- generation, dual-frequency spaceborne These techniques have been widely used dependent effects uncorrected), because GPS receiver based on commercial for the types of meteorological applica- the two radio paths at L1 and L2 in the equipment, which uses the new L2 civil tions discussed in this article. ionosphere may differ significantly in signal (L2C) in addition to GPS L1. The On September 23, 2008, the office of a strongly perturbed ionosphere (and receiver is also compatible with ground- the U.S. Assistant Secretary of Defense thus combining the two cannot help based C/A-code pseudolites having an for Networks and Information Integra- determine completely the ionospheric RNSS uplink frequency (Lp signal = tion (ASD-NII) published a notice in the signal), and because the L2 signal can- 1340 MHz). U.S. Federal Register confirming the U.S.
36 InsideGNSS november/december 2008 www.insidegnss.com government’s plan to phase out support is not so critical for the ground-based observed in measurements collected of codeless and semi-codeless access to GPS measurements (except for a real- from either system. the legacy GPS L2 signals by December time meteorological application) as it is Fourth, the surge in commercial 31, 2020. The U.S. government wants for the GPS radio occultation. applications from Galileo may contrib- to encourage the civilian user commu- Indeed, that latter technique cur- ute to making GPS and Galileo receivers’ nity to migrate to new GPS equipment rently relies on ad hoc receivers (semi- basic components more affordable and that can use the next-generation signals codeless) or complex antenna schemes supporting improved baseline designs known as L2C and L5, as well as the cur- in order to be able to collect L2 measure- for scientific receivers. rent L1 C/A code. ments, while extrapolation is needed for Fifth, the generalized deployment of This policy decision means that tracking in the lower atmosphere. With several ground (or spaceborne) networks L1-L2C spaceborne GPS receivers will Galileo the measurement process should (or constellations) of GPS and Galileo become the basic equipment for GPS- be simplified in that respect. receivers could enable meteorologists only precise orbit determination and/or Second, the ability for a future dual- to collect a wealth of measurements of radio occultation for any mission that constellation receiver to track both GPS opportunity. This would be possible if could potentially continue after 2020. and Galileo satellites will mean that the GPS receivers were able to conduct their number of links (raw measurements) is primary functions of positioning (orbit Prospects for Galileo multiplied by about two, thus yielding determination) and timing while at the The future Galileo system should bring potentially twice as many independent same time collecting scientific measure- about a number of improvements to the observations. ments with a common baseline receiver GPS-based observing systems discussed Third, the future climate records design. The amount of data collected in this article so far. from ground-based GPS or GPS radio (especially for spaceborne receivers) First, for the second frequency occultation measurements on the GPS could be tremendous and achieved at a required for ionospheric correction system itself would not rely on a single very little additional cost. Galileo will present a higher signal-to- GNSS system. Galileo will make it pos- The more promising frequencies noise ratio than the current GPS L2. This sible to compare the potential drifts to be used in the future for multiple
www.insidegnss.com november/december 2008 InsideGNSS 37 weather report
Yen, and Z. Zen, “The COSMIC/FORMOSAT-3 Mis- constellation radio occultation receiv- with GPS radio occultation receivers for sion: Early Results”, Bull. Amer. Meteorol. Soc., ers will be the ones commonly used by updating its operational weather predic- 89, pp. 313-333, 2008 several systems. For instance, the case tion analyses. This decision was made of two worldwide GNSS constellations after it was shown that these respective [2] Bevis, M., and S. Businger, T. Herring, C. with two confirmed common central observations helped improve quantita- Rocken, R. Anthes, and R. Ware, “GPS meteorol- ogy: Remote sensing of atmospheric water vapor frequencies is today unique, with GPS tive precipitation forecasts, and predic- using the Global Positioning System.” J. Geophys. and Galileo and E1/L1 + E5a/L5 fre- tions of temperature and wind. More Res., 97, pp. 15787-15820, 1992 quencies. progress is expected as the number of Of course, for high-end, heavy radio such observations may increase; the [3] Desroziers, G., and L. Berre, B. Chapnik, and P. occultation receivers, “all frequencies ground network of GPS receivers is due Poli, “Diagnosis of observation, background and analysis error statistics in observation space”, in view” or “all constellations in view” to expand worldwide, and more satellites Quart. J. Roy. Meteorol. Soc., 131, pp. 3385-3396, (GPS/Galileo/Compass/GLONASS/ may carry GPS receivers of radio occul- 2005 IRNSS/QZSS, for instance) are better tation grade. from a radio science perspective. In Overall, the use of the GPS signals [4] Healy, S. B., and J. N. Thépaut, “Assimilation particular, simultaneous use of the four for meteorological observations is a experiments with CHAMP GPS radio occultation measurements”, Quart. J. Roy. Meteorol. Soc., Galileo frequency bands — E5a, E5b, good example of a measurement of 132, pp. 605-623, 2006 E6, and E1 — offers the possibility to opportunity in which large number of directly solve the high-order terms of the transmitters and all-weather availabil- [5] Issler, J.-L., A. DeLatour, L. Ries, and M. Gron- ionospheric delays, to improve carrier ity have enabled meteorologists to gather din (2008), J., et al., “Lessons Learned from the phase ambiguity resolution and related ever more measurements of our atmo- Use of GPS in Space; Application to the Orbital use of GALILEO,” Proceedings of ION GNSS 2008, robustness, and to obtain more accurate sphere in near real-time. The potential Savannah, Georgia, September 16-19, 2008 measurements. The use of Compass and of these observations for climate studies Galileo E5b signals already being trans- is significant, but further detailed sen- [6] Kirk-Davidoff, D. B., and R. M. Goody, and mitted on a common frequency also sitivity studies are required in order to J. G. Anderson, “Analysis of Sampling Errors for looks promising, like the use of com- evaluate the effects of possible sources Climate Monitoring Satellites,” J. Climate, 18, pp. 810-822, 2005 mon E6 central frequency by QZSS and of interruptions in measurements as well Galileo and probably at least another as trends in the stability of time-series [7] Kursinski, E. R., and G. A. Hajj, K. Hardy, L. GNSS system. measurements being collected. Romans, and J. Schofield, “Observing tropospheric The prospects offered by the Gali- water vapor by radio occultation using the global Conclusions and leo system to meteorologists using positioning system,” Geophys. Res. Lett., 22, pp. 2365-2368, 1995 Perspectives GPS-based measurement may include Within only 10 years of the initial improving ionospheric characterization [8] Mannucci, A. J., and C. O. Ao, T. P. Yunck, L. E. deployment of the GPS system, opera- (and correction), increasing the number Young, G. A. Hajj, B. A. Iijima, D. Kuang, T. L. Mee- tional ground-observation networks of collected measurements, avoiding the han, and S. S. Leroy, “Generating climate bench- have been set up over Europe, Japan, and reliance of such measurements on one mark atmospheric soundings using GPS occulta- tion data,” Proc. SPIE, 6301, 2006 the United States. At first these typical- system only (GPS), and (perhaps most lywere for geodetic purposes, but now importantly) collecting a large number [9] Niell, A. E., “Global mapping functions for the they enable the collecting, processing, of measurements of opportunity at very atmospheric delay at radio wavelengths,” J. Geo- and distributing of GPS atmospheric little cost. phys. Res.¸ 101, pp. 3227-3246, 1996 delay measurements to national meteo- [10] Ohring, G., and B. Wielicki, R. Spencer, B. rological services. Manufacturers Emery, and R. Datla, “Satellite instrument calibra- Measurements of atmospheric The dual-frequency GPS receiver to tion for measuring global climate change,” Bull. refraction can also be collected via the be used on the PROBA-2 satellite is a Amer. Meteorol. Soc., 86, pp. 1303-1313, 2005 GPS radio occultation technique, which modified version of the TOPSTAR 3000 [11] Poli, P., and P. Moll, F. Rabier, G. Desroziers, involves GPS receivers in space. Recent from Thales Alenia Space, France, with B. Chapnik, L. Berre, S. B. Healy, E. Andersson, and examples include the six-satellite con- involvement of Sideral, Switzerland, F.-Z. El Guelai, “Forecast impact studies of zenith stellation FORMOSAT-3/COSMIC and and Nordspace, Norway. total delay data from European near real-time the GRAS (GNSS Receiver for Atmo- GPS stations in Météo-France 4DVAR,” J. Geophys. spheric Sounding) instrument on-board Additional Resources Res., 112, D06114, 2007 the MetOp satellite. [12] Rabier, F., and H. Järvinen, E. Klinker, J.-F. [1] Anthes, R. A., and P. A. Bernhardt, Y. Chen, L. Mahfouf, and A. Simmons, “The ECMWF opera- Since September 2006 and 2007, Cucurull, K. F. Dymond, D. Ector, S. B. Healy, S.-P. tional implementation of four dimensional varia- the national weather service of France, Ho, D. C. Hunt, Y.-H. Kuo, H. Liu, K. Manning, C. tional assimilation. Part I: Experimental results Météo-France, has been using data sets McCormick, T. K. Meehan, W. J. Randel, C. Rocken, with simplified physics”, Quart. J. Roy. Meteorol. from European ground-based GPS W. S. Schreiner, S. V. Sokolovskiy, S. Syndergaard, Soc., 126, pp. 1143-1170, 2000 networks and from satellites equipped D. C. Thompson, K. E. Trenberth, T.-K. Wee, N. L.
38 InsideGNSS november/december 2008 www.insidegnss.com Authors out pioneering work at the European Centre for Jean-Luc Issler is in Medium-range Weather Forecasts on four-dimen- charge of the Transmis- Paul Poli has led research sional variational data assimilation. This algorithm, sion Techniques and Sig- at the NASA Goddard still considered state-of-the-art today, is used by nal Processing Depart- Space Flight Center and meteorological agencies worldwide. ment at CNES. He is at Météo-France on the involved in the develop- use of GPS data for Michel Mauprivez is cur- rently deputy head of the ment of several types of numerical weather pre- GNSS, FFRF, and TMTC equipment in Europe, for diction. He finalized this Upper-air Observations Department of Météo- space or ground users. Issler is the French dele- work with the operational application of these data gate to the Galileo Signal Task Force. He received at Météo-France. He received his Ph.D. from the France. He joined Météo- France in 1996 after a in 2004 the Astronautic Prize of the AAAF ( French University of Maryland, Baltimore County, USA, aeronautical and space association), and in 2008 and recently joined the weather reanalysis team career in the French Air Armed Forces where he held responsibilities as a the EADS Prize of science and engineering deliv- of the European Centre for Medium-range Weath- ered by the French Academy of Sciences, mainly er Forecasts. telecommunications and radar Officer. His work at Meteo-France involves improving the observation for his technical work on Galileo signals and Jean Pailleux, Ph.D., is network for real-time water vapor monitoring, spaceborne GNSS equipment. currently deputy director capitalizing on existing GPS networks. Antoine de Latour has o f Météo-France Sylvie Dufour is currently head of the Observations been a navigation engi- research center. He neer in the CNES Trans- helped implement the Prospective, Coordination and Studies Depart- ment of Météo-France. She recently helped forge mission Techniques and early and recent meth- Signal Processing ods of data assimilation an agreement with the Institut Géographique National to secure the provision to meteorology Department since 2003. at the European Centre for Medium-range Weath- He is involved in the er Forecasts and at Météo-France. His work was of GPS data originally intended for geodesy. She joined Météo-France from the public sector and is Galileo Program in which he supports the European instrumental in helping satellite data make the Space Agency and the European Commission. De impact they have today in operational meteorol- interested in maximizing the benefits of public- funded infrastructure. Latour is involved in the design of the Galileo sig- ogy. He regularly serves as an expert for the World nals, in the use of GNSS for space applications, in Meteorological Organization. Michel Grondin has been the GPS/Galileo radio frequency compatibility Véronique Ducrocq, Ph.D., working at CNES since assessment, and in the development of a GNSS RF is currently head of the 1981. He was responsible signal simulator. In particular, he studied in depth research division on mid- for the GPS receiver the CBOC Galileo signal definition and the PRS sig- latitude convection in equipment on the DEME- nal. De Latour proposed new tracking signal tech- Météo-France research TER micro-satellite. niques for generic receivers and spaceborne center. She directs Since July 2005 he has receivers. He was graduated from the Ecole Supéri- research to enhance the been working in the Transmission Techniques and eure d´Electricité (Supélec) in Paris and obtained understanding and predictability of mid-latitude Signal Processing Department. He provided sup- a master’s degree from the University of Stutt- precipitating systems and severe weather events. port to the integration of the GPS receiver of the gart. Her work has promoted the use of GPS observations Alpha Magnetic Spectometer (AMS) experiment to fly on board the International Space Station. He Lionel Ries has been a in meteorology to improve quantitative precipita- navigation expert in the tion forecast for thunderstorms. also tuned CNES GPS L5 software receiver for use in low earth orbit with the aid of a GPS simulator. Transmission Technique Patrick Moll is currently He graduated from the Ecole Superieure and signal processing involved with observa- d’Electricité (Supélec) in Paris. (TT) Department at CNES tions handling in Météo- since June 2000, where France numerical weath- Françoise Lechat-Carvalho he coordinates naviga- er prediction system. He has been a propagation tion technical activities. He is responsible for contributed to the and signal processing research activities on GNSS2 signals, ground and research that lead to the engineer at the Trans- spaceborne receivers, payloads, and systems. He operational assimilation of ground-based GPS mission Technique and contributed to the invention of the CBOC signal. He data over Europe. He also contributed to the Signal Processing provided support to the 2004 US-EU agreement on implementation of modern variational techniques Department of CNES GPS and Galileo. He was responiable for the devel- at Météo-France and the European Centre for since 2006. She studies the propagation of GNSS opment of the L1-L2C signal processing algorithms Medium-range Weather Forecasts. signals through the troposphere and the iono- now implemented in the ASIC and processor of the sphere, the characterization of these atmospher- TOPSTAR 3000 receiver, in the frame of a CNES R&D Florence Rabier, Ph.D., is ic layers with the use of GNSS, and the propagation activity. He graduated from the Ecole Polytech- currently head of the channel for navigation and mobile telecommuni- nique de Bruxelles, at Brussels Free University, and observations team in cations in S-band. She is involved in the Indian- received an M.S. degree from the Ecole Nationale Météo-France’s numeri- French GSAT4/GAGAN Ka-band propagation Superieure de l’Aeronautique et de l’Espace cal weather prediction experiment. She was graduated in electronics and (Supaero ) in Toulouse. research center. Between communication systems engineering from Insti- 1992-1998 she carried tut National des Sciences Appliquées in Rennes. www.insidegnss.com november/december 2008 InsideGNSS 39