Session E1, Paper #1

The CNES Real-time PPP with Undifferenced Integer Ambiguity Resolution Demonstrator
D. Laurichesse, Centre National d´Etudes Spatiales, France

Integer ambiguity resolution on undifferenced GPS phase data has been the object of a lot of attention in recent years. By combining pseudo-range and phase information, it is indeed possible to directly fix integer ambiguities on zero-difference phase measurements, for dual-frequency problems, without any atmospheric models. Phase measurement then become pseudo-range-like measurements with millimeter noise level.
The synchronization of a regional network can then be performed using these unambiguous phase measurements and precise GPS satellites ephemeris (such as IGS products), giving the GPS emitter clocks. These clocks (for which we introduced the name integer phase clocks) have an interesting property : they allow the positioning of independent receivers using a PPP-like method, with integer ambiguities characteristics on the phase observables.

These user phase residuals, which are below one centimeter, are precise enough to fix ambiguities, thus allowing a precise positioning on a global scale using a full ´integer PPP method´. The positioning precision is at the centimeter level.

Our complete method, adapted for dual-frequency problems, was presented for the first time at the ION GNSS 2007 meeting. Methods and results were refined and summarized in the Summer 2009 issue of Navigation. This article contains actual results from ground Precise Point Positioning and Precise Orbit Determination of satellites equipped with dual frequency GPS receivers.

The extension of our method to real-time applications has been developed in papers by the same author presented at the ION NTM 2008, ION ITM 2009 and ION GNSS 2010 meetings ). The core of the real-time implementation is a Kalman Filter working in mixed-mode (with both real- and integer-valued phase ambiguities). The filter produces GPS constellation states (orbits and clocks) with the ´integer´ property. Algorithms for user receivers were also introduced. Their performance in terms of real-time precision and convergence time was compared to standard RTK methods.

To demonstrate that this processing strategy is compatible with the latency constraints imposed by real-time applications and with the level of performance typical of common personal computers, CNES is developing a complete real-time ´integer PPP´ demonstrator. This paper presents the goals and the architecture of this demonstrator, along with some actual results for both the system side and the user side.

The system side of the demonstrator collects network measurements, computes state-space products, and disseminates them over the net, using the tools that CNES uses in the frame of the Real Time IGS Pilot Project.

On the user side, the demonstrator collects measurements from a local receiver, retrieves state-space data from the internet, and performs real-time kinematic ´integer PPP´. Several stations are monitored in real-time using this technique. Monitoring plots errors are generated on a real-time basis.

As part of this demonstrator, CNES proposes a free user test package that provides all the tools for users to perform their own PPP with ambiguity resolution. The test package includes an access to the CNES caster, and ICD to understand the nature of the ambiguity resolution quantities, and a PPP software. This PPP software is freeware and the source code is provided. All this material is available on a dedicated web site.

[Return to Program]

Session E1, Paper #2

Undifferenced Ambiguity Resolution Applied to RTK
S. Carcanague, M3Systems, France; O. Julien, ENAC, France; W. Vigneau, M3System, France; C. Macabiau, ENAC, France

The Phase Lock Loops of a GNSS receiver can provide very precise carrier phase measurements that can potentially be used for positioning. However these measurements are inherently ambiguous since they include an unknown integer number of carrier cycles referred to as the carrier-phase ambiguity. To estimate this ambiguity, important biases have to be removed from carrier phase measurements such as ... . There are two ways to do this: (1) to difference the observables from the user receiver (or rover) with the measurements from a reference receiver that is spatially close in order to remove common biases, (2) to remove the biases directly by either using a linear combination between observables, or estimating them or obtaining their values from an external source. The first technique is the basis for Real-Time Kinematic (RTK) that uses at least 2 receivers to estimate the differenced carrier-phase ambiguities. The second technique is the basis for Precise Point Positioning (PPP) that estimate the receiver coordinates, the zenith tropospheric delay and the carrier-phase ambiguities from a ionosphere-free carrier phase combination using precise ephemeris.

Ambiguities can be estimated either directly as integers if the residual measurement errors are small compared to the carrier wavelength or as floats if this is not the case. Once the ambiguities are estimated correctly, carrier phase measurements can be used as unambiguous measurements and the position can be determined with a very high precision, usually at centimeter-level.
To fasten carrier-phase ambiguity resolution, code measurements are commonly used in RTK to provide an initial float estimation of the ambiguities to the ambiguity resolution algorithm. The more precise the code measurements are, the faster the ambiguity resolution will be [Milbert, 2005]. In some case, such as for low-cost RTK using cheap mono-frequency receivers, the ambiguity is classically kept as a float as in [Realini, 2009]. In this case, the accuracy of the final solution and the convergence time is directly linked to the accuracy of the double-differenced code measurements. Considering that the reference receiver and the rover are close enough, the precision of the double-differenced code measurements depends on the level of noise- and multipath-induced errors on the reference receiver and the rover. It is thus crucial to try to reduce these errors as much as possible to improve RTK performance.
Recent improvements to PPP have allowed the resolution of the carrier-phase ambiguities on L1 and L2 as integers [Laurichesse, et al., 2009], [Collins, 2008], [Ge, et al., 2008], usually taking a fairly long convergence time. The technique uses 2 types of products: satellite widelane biases and precise satellite narrowlane clocks, which are freely available on the CNES-CLS IGS center website. They allow for undifferenced ambiguity resolution.

The methodology developed in this paper targets short-baseline RTK users with a rover in a difficult environment or low-quality rover receiver. To improve RTK in such conditions, the following methodology is applied. First, the ambiguities are solved on the reference receiver using a PPP software and CNES-CLS IGS products. The resulting precise unambiguous carrier phase measurement can then be differenced with the code measurements from the rover. In order to remove the ionospheric delay, the widelane carrier phase measurements on the reference receiver are differenced with the narrowlane code measurements on the rover. Satellite biases have to be removed as well, using the widelane satellite biases provided by CNES-CLS IGS center. An algorithm using this combination and single-differenced widelane carrier phase was first introduced by the authors in [Carcanague, et al., 2011]. Tests were performed on 3 IGS stations and the precision of the float ambiguities was shown to be greatly improved, as the reference and the rover had similar level of noise. This concept was also shown to be very useful when investigating the noise- and multipath-induced errors of the rover measurements. Indeed, if the stations are close enough, the pseudorange error is only due to the widelane carrier phase noise of the reference station and the code noise and multipath of the rover. Multipath error can then be precisely deduced on each pseudorange by knowing the reference trajectory.
In this paper, the same concept, based on the exclusive use of unambiguous carrier phase measurements from the reference receiver will be applied to an urban environment and to low-cost RTK. First the algorithm will be presented and a detailed analysis of the studied urban environment including level of noise on the code and the carrier phase, the availability and the occurrence of cycle slips and loss of locks will be performed. Then, the algorithm will be tested on real data from a dual frequency rover using single-difference and double differenced observables, which will lead to conclusions on the suitability of the method to RTK in the studied urban environment. Different strategies will be investigated to adapt the algorithm to low-cost RTK and consumer-grade mono-frequency receivers. Finally, it will be shown that performing PPP with ambiguity resolution on the reference receiver allows to instantly initialize a PPP filter on the rover in the case of a communication link outage. Tests will demonstrate that a high level of accuracy can be maintained by both single-frequency users and dual-frequency users during long reference receiver data unavailability.

Carcanague, S., Julien, O., Vigneau, W., & Macabiau, C. (2011). A New Algorithm for GNSS Precise Positioning in Constrained Area. Proceedings of ION International Technical Meeting, January 24-26, San Diego, CA.
Collins, P. (2008). Isolating and estimating undifferenced GPS integer ambiguities. ION NTM 2008, 28-30 January 2008, San Diego, CA.
Ge, M., Gendt, G., Rothacher, M., Shi, C., & Liu, J. (2008). Resolution of GPS Carrier-phase Ambiguities in Precise Point Positioning (PPP) with Daily Observations. Journal of Geodesy, Vol. 82, No. 7,pp. 389-399.
Henkel, P., Gomez, V., & Gunther, C. (2009). Modified LAMBDA for absolute carrier phase positioning in the presence of biases. ION GNSS 2009, 22nd International Meeting of the Satellite Division of The Institute of Navigation, Savannah, GA, September 22-25, 2009.
Laurichesse, D., Mercier, F., Berthias, J., Broca, P., & Cerri, L. (2009). Integer Ambiguity Resolution on Undifferenced GPS Phase Measurements and its Application to PPP and Satellite Precise Orbit Determination. Navigation, Journal of the institute of Navigation, Vol. 56, Nø 2, Summer 2009.
Milbert, D. (2005). Influence of Pseudorange Accuracy on Phase Ambiguity Resolution in Various GPS Modernization Scenarios. NAVIGATION, Spring 2005, 52(1), pp.29-38.
Realini, E. (2009). goGPS free and constrained relative kinematic positioning with low cost receivers. PhD thesis, Politecnico Di Milano.

[Return to Program]

Session E1, Paper #3

Issues in Ambiguity Resolution for Precise Point Positioning
P. Collins, York University and Natural Resources Canada, Canada; S. Bisnath, York University, Canada

New techniques are now available for isolating carrier-phase ambiguities as integer parameters in single-station data. These methods have permitted the Ambiguity Resolution (AR) techniques formerly reserved for relative positioning to be applied to Precise Point Positioning (PPP).
Of these methods, the Decoupled Clock Model appears to be the most general, requiring no explicit observation or parameter differencing, making minimal assumptions about the nature of the obscure hardware delays, yet providing an optimal model for such delays within the overall context of oscillator synchronisation.

PPP-AR with the Decoupled Clock Model provides an improvement in solution consistency over the real-valued ambiguity solution, when long fixed-point data sets are processed, or convergence is achieved in kinematic data.

Despite isolating the undifferenced ambiguities as integer-valued parameters, the problem of solution convergence has generally remained. This appears to be due to marginally defined ambiguities, where the early solution errors remain significant compared to the ambiguity wavelength. Provided the satellite correction errors are minimised, these errors are dominated by pseudorange noise and multipath which remain dominant while the phase ambiguities are poorly resolved.

The highest standard of GNSS positioning, in terms of precision and timeliness, is set by the Real-Time-Kinematic (RTK) method, whereby differential processing over short baselines or small networks provides near-instantaneous relative coordinates at the centimetre-level or less. As such, this remains the standard with which to compare PPP-AR.

This paper will show that both RTK and PPP models can be reduced to a simple common observation model. This model has been applied before to baseline processing, but the implications for single-station processing have not been emphasized. With the expanded interpretation, simple covariance analysis shows how RTK works versus PPP.

Briefly, an RTK solution is formulated explicitly in terms of the ionosphere even when the effect is assumed to cancel. If a very tight covariance constraint can be provided on the ionosphere delay, then a strong solution with a small ambiguity search space will result. On the other hand, in PPP the ionosphere-free and M-W combinations both eliminate the ionospheric component but at the price of a weaker model. The ionospheric effect remains in the covariance and cannot be removed. As a result, the ionosphere-free ambiguity search space remains large and significantly elongated even after ambiguity decorrelation has been performed.

As confirmation of this interpretation, previously reported results by other researchers have shown rapid PPP re-convergence using ionospheric predictions. In essence, ionospheric predictions are used as bridging parameters for re-convergence after a solution reset (where all ambiguities have been re-initialised). With the covariance analysis in mind, it is clear that the predictions and corresponding covariance constraints drastically improve the ambiguity search space to a similar level as moderate baseline-length RTK.

This paper will discuss how a comprehensive solution to Ambiguity Resolution for Precise Point Positioning takes into account the local ionosphere in achieving rapid re-convergence of the position solution. The results will be derived from real data, including GPS stations subjected to co-seismic movement.

[Return to Program]

Session E1, Paper #4

Performance Assessment of Long-baseline Integer Ambiguity Resolution with Different Observation Models
X. Yu, X. Zhang, J. Liu, Wuhan University, China; J. Shi, University of Calgary, Canada; C. Cai, Central South University, China; Y. Gao, University of Calgary Canada

For long-baseline relative positioning, the effects of ionosphere and troposphere errors cannot be completely cancelled out by double difference due to the decreasing spatial correlation as baseline separation increases. Generally there are two observation models that can be used to estimate the baseline parameter with integer ambiguity resolution. The first one uses ionosphere-free (IF) linear combination between dual-frequency observables, which can eliminate first-order ionospheric effect (count for 99% of the total effect). Before the IF integer ambiguities can be fixed, Melbourne-W?bbena (MW) wide-lane (WL) ambiguities are first pseudo-fixed by smoothing multi-epoch observations and then narrow-lane ambiguities are fixed based on the fixed WL ambiguities and estimated float IF ambiguities. The second model directly uses original L1 and L2 carrier-phase observations. Instead of removing ionospheric errors by IF combination, this model estimates ionospheric errors by introducing additional unknown parameters.
Since the first model does not estimate the ionosphere parameters, it can decrease the total number of unknown parameters and will also largely reduce the computational load. It has therefore been applied in many high-precision software packages such as GAMIT and BERNESE. The second model makes use of all uncombined L1 and L2 observations to increase observation redundancy, which is often used for long baseline kinematic positioning. Mathematically, it has been proved that the coordinate solutions of both models are identical, but few results about their effects on ambiguity fixing are reported.

In this paper, the efficiency of integer ambiguity resolution using both models has been investigated. The ratio of the second minimum quadratic form of the residuals over the minimum quadratic form and the bootstrap ambiguity success rate are used to analyze the reliability of integer ambiguity resolution. Long distance (100-500km) static and kinematic GPS datasets during different levels of solar activity are then processed using these two different models, respectively. In the first model, the unknown parameters to be estimated include coordinate, troposphere and IF ambiguity parameters. In the second model, the ionosphere parameters are also estimated for each satellite along with L1 and L2 ambiguity, coordinate and troposphere parameters. To improve the ratio value that reflects the reliability of fixed ambiguity, we developed a new ambiguity resolution method in which the original L1 and L2 ambiguities are transformed into wide-lane and narrow-lane ambiguities, and then resolve them sequentially using the LAMBDA method. IGS precise orbit products are applied to reduce the effect of the satellite orbit errors. The ratio value and the bootstrap success ratio are computed on an epoch-by-epoch basis.

The numerical results show that the coordinates estimated by these two models with float ambiguity resolution are almost identical. But the success rate of ambiguity resolution in the second model is significantly higher than the success rate in the first model. Furthermore, the bootstrap success rate of ambiguity resolution using the first model can reach 100% only if 2-hour or longer observation period is used. In addition, the time required to fix the ambiguities using the first model is clearly longer than using the second model when the ambiguities are validated by both the ratio value and the bootstrap success rate. Compared to the traditional strategy that resolves L1 and L2 ambiguity directly in the second model, our new method resolves wide-lane and narrow-lane ambiguities sequentially can greatly improve the ratio value of correct ambiguity fix and also shorten the time to first fixing (TTFF). Our investigation concludes that the second model is preferable for long-baseline relative kinematic positioning as it increases the success rate of ambiguity resolution. Moreover, our proposed method can further improve the second model by enhancing the ratio value of correct ambiguity fix.

[Return to Program]

Session E1, Paper #5

A Novel Device for Autonomous Real-Time Precise Positioning with Global Coverage
D. Calle, P. Navarro, A. Mozo, R. Piriz,D. Rodriguez, G. Tobias, GMV, Spain

Precise Point Positioning (PPP) is a relatively new positioning technique providing centimeter-level error. PPP processes dual-frequency pseudorange and carrier-phase measurements from a single user receiver, using detailed physical models and corrections, and precise GNSS orbit and clock products calculated beforehand (for example products from IGS, the International GNSS Service). PPP is different from other precise-positioning approaches like RTK in that no reference stations are needed in the vicinity of the user receiver. The only observation data that must be processed are measurements from the user receiver. Another advantage of PPP is that since the GNSS orbit and clock products are by nature global, the PPP solutions are also global, i.e., the PPP approach works for a receiver located anywhere on or above the Earth surface, and the resulting position is referred to a well-known terrestrial reference frame (normally ITRF). PPP can be applied at post-processing level and also in real-time applications, provided that real-time input orbits and clocks are available. One disadvantage of standard PPP however is its relatively slow convergence time, which is of the order of an hour for decimetric accuracy, as compared to nearly instantaneous convergence with centimetric accuracy in short-baseline RTK.
In 2008 GMV introduced magicGNSS, a web application for high-accuracy GNSS data processing. magicGNSS is available online at http://magicgnss.gmv.com. A free account can be requested online for a one-month trial period. A PPP software module is available in magicGNSS, allowing the user the post-processing of dual-frequency static and kinematic RINEX measurement files. The PPP module supports GPS and GLONASS data. Satellite orbits and clock products are calculated internally in a transparent way for the user. GMV has developed an infrastructure for the generation of precise GPS and GLONASS orbits and clocks with very low latency in a first step, and in real time in a second step. The products generated this way are contributed to the IGS Real Time Pilot Project, and are also used to feed the PPP service, part of the web application magicGNSS.

The product generation is based on an Orbit Determination and Time Synchronisation (ODTS) process, which runs typically every 15 minutes. This process receives as input dual-frequency code and phase measurements collected in real time from a world-wide network of IGS stations, using the NTRIP protocol. Then, they are pre-processed also in real time by a Pre-Processing and Validation module (PPV) and made available to the different algorithms. In parallel to the ODTS, another process estimates the clocks in real time taking as input the observations and the outputs from the last ODTS execution. There is a small latency in the delivery of the clock estimation, which is associated to the time that the algorithm waits for the arrival of the measurements from the station through the Internet, typically one or two seconds.

The GPS and GLONASS satellites are processed together, in order to ensure a consistent solution. It is necessary to estimate an inter-channel bias when processing GLONASS data. This must be done in order to compensate for the different internal delays in the pseudorange measurements through the GLONASS receiver, associated to the different frequencies used by the different satellites. The real-time orbits and clocks are available as a data stream to real-time processing algorithms (such as real-time PPP), and stored in standard formats (SP3, clock RINEX) for offline use. In the near future the products will contain also additional clock biases to allow the resolution of integer carrier-phase ambiguities in PPP, for improved positioning accuracy and shorter convergence time.

In parallel to the PPP product generation, GMV is developing a novel user terminal prototype for PPP, that can be connected to any GNSS receiver providing real-time raw measurements and navigation messages through a serial or USB port (in RTCM format), and can receive PPP orbit+clock corrections via GSM/GPRS (internet through mobile phone network) and also via Iridium (satellite based, with global coverage). This device is a self-contained, DC-powered case, containing mainly a Single Board Computer (SBC board) running Linux for the PPP client software, an additional board hosting the new Iridium 9602 modem, plus a LCD touch display for terminal control and visualization of results. The major advantage of this design is that the PPP user terminal can be connected to virtually any professional GNSS receiver in the market, since practically all of them are able to generate raw output in RTCM format via a serial or USB port.

For remote locations without GSM/GPRS coverage, satellite communications via Iridium shall be used. Iridium is a constellation of 66 satellites (6 planes of 11) in polar orbit at a height of 780 km, used for worldwide voice and data communication from hand-held satellite phones and other transceiver units. The Iridium network is unique in that it covers the whole Earth, including poles, oceans and airways. Iridium operates in the 1610.0-1626.6 MHz frequency band. We intend to use the Iridium Short Burst Data (SBD) service, which allows two-way communications to a mobile transceiver. The SBD service allows Mobile Originated messages up to 340 bytes, and Mobile Terminated messages up to 270 bytes. The service has a low, uniform global latency reported to be of the order of half a minute. PPP service via Iridium is based on clock corrections sent at a rate of one minute; this is believed to be a reasonable compromise between positioning performances and communications costs.

This paper describes the implementation of GMV´s infrastructure for real-time PPP products and their usage in the new user terminal, with a presentation of preliminary real-time positioning results and an assessment of their accuracy.

[Return to Program]

Session E1, Paper #6

Real-time Combination of GNSS Orbit and Clock Corrections Streams Using a Kalman Filter Approach
L. Mervart, Technical University Prague, Czech Republic; G. Weber, Federal Agency for Cartography and Geodesy, Germany

A number of Real-Time IGS Analysis Centers (ACs) estimates GNSS orbit and clock corrections to broadcast ephemeris. This paper presents a new approach to process several orbit and clock corrections streams in real-time to produce, encode, upload and save a combination of correctors from various providers.
The approach is based on a Kalman Filter. Satellite clocks estimated by individual ACs are used as pseudo observations within the adjustment process. Each observation is modeled as a linear function of three estimated parameters: AC specific offset, satellite specific offset common to all ACs, and the actual satellite clock correction which represents the result of the combination. The three parameter types differ in their statistical properties. The satellite clock offsets are assumed to be static parameters while AC specific and satellite specific offsets are stochastic parameters with appropriate white noise. The solution is regularized by a set of minimal constraints.

In view of IGS real-time products, the combination approach has been integrated in the ´BKG Ntrip Client´ (BNC). We demonstrate that

 the software with its Graphic User Interface and wide range of supported Operation Systems represents a perfect platform to process many broadcast corrections streams in parallel;  outages of single AC product streams can be mitigated through merging several incoming streams into a combined product;  generating a combination product from several AC products allows detecting and rejecting outliers;  a Combination Center (CC) can operate the program to globally disseminate a combination product; the combination stream is encoded in RTCM-SSR messages and uploaded to an NTRIP broadcaster.  an individual AC could prefer to disseminate a stream combined from primary and backup IT resources to reduce outages;  this enables a PPP user to follow his own preference in combining streams from individual ACs for Precise Point Positioning;  it allows an instantaneous quality control of the combination process not only in the time domain but also in the space domain; this can be done through direct application of the combination stream in a PPP solution even without prior stream upload to an NTRIP Broadcaster;  this enables the output SP3 files containing precise orbit and clock information for further offline processing using other tools.

It is shown that a major effect in the combination of GNSS orbit and clock corrections streams is the selection of ACs to include. A combination product can be improved in accuracy by using only the best two or three ACs. However, with only a few ACs to depend on, the reliability of the combination product suffers and the risk of total failures increases. So there is an important tradeoff that must be considered when selecting streams for a combination. The major strength of a combination product is its reliability and stable median performance which can be much better than that of any single AC product.

This mainly applies in situations where we have a limited number of solutions to combine and their quality varies significantly. The situation is different when the total number of ACs is larger and the range of AC variation is smaller. In that case, a standard full combination is the best.

It is so far only the satellite clock corrections which are combined while orbit correctors in the combination product as well as the product update rates are just taken over from one of the incoming corrections streams declared as the ´Master Stream´. So the ´Master Stream´ is responsible for two things: the satellite positions and the combination rate. Combining only clock corrections using a fixed orbit reference has the possibility to introduce some analysis inconsistencies. We may therefore eventually consider improvements on this approach.

The final goal is the provision of a reliable GNSS product stream free from outages and outliers. We show that real-time ´Precise Point Positioning´ using a combination product is possible with an accuracy better one decimeter after about 15 minutes or less convergence time.

[Return to Program]

Session E1, Paper #7

GLONASS Inter-frequency Biases and Their Effects on RTK and PPP Carrier-phase Ambiguity Resolution
N. Reussner, L. Wanninger, TU Dresden, Germany

The frequency division multiplexing (FDMA) of the GLONASS signals causes inter-frequency biases in the receiving equipment. These biases vary considerably for receivers from different manufacturers and thus they are able to complicate or prevent carrier-phase ambiguity fixing. Starting this year new GLONASS signals will be added that will use code division multiplexing like GPS does. But still, for the coming some 10 years, only GLONASS FDMA-signals will be able to provide continuous dual-frequency coverage.
In RTK positioning a complete and reliable ambiguity fixing requires a-priori information of the carrier-phase inter-frequency bias differences of the receivers involved. We estimated GLONASS carrier-phase inter-frequency biases for about 130 individual receivers of 9 manufacturers. In general, receivers of the same type and even receivers of the same manufacturer show similar biases, whereas the differences among the manufacturers can reach up to 0.2 ns (more than 5 cm) for adjacent frequencies and thus up to 24 ns (73 cm) for the complete L1 or L2 frequency bands. A few individual receivers were identified which behave differently as compared to other receivers of the same type or which experience variations in their inter-frequency biases.

The situation is similar for GLONASS PPP ambiguity resolution. Here, fractional cycle biases of the satellite (and receiver) hardware delays have to be estimated at reference stations and then be applied to rover receiver observations in order to be able to resolve the integer ambiguities. If the various reference stations involved use GLONASS receivers of different manufacturers, the estimated fractional cycle biases can not be merged to one data set as easily as this is done with GPS. The GLONASS inter-frequency biases must be modelled in the merging process. The same is true for the ambiguity resolution at the GLONASS rover receiver. Ambiguity resolution can only be performed successfully if a-priori information of the carrier-phase inter-frequency biases exists and if the receiver individual biases are estimated.

We will present our latest results of GLONASS inter-frequency biases estimated from differential carrier-phase positioning (RTK) and as well from Precise Point Positioning (PPP). Furthermore, we will describe our algorithms for GLONASS PPP ambiguity resolution in detail.

[Return to Program]

Session E1, Paper #8

GNSS Positioning of Ocean Buoys in Japan for Disaster Prevention
M. Kanzaki, Y. Matsushita, H. Kakimoto, Hitachi Zosen Corporation, Japan; C. Rocken, T. Iwabuchi, L. Mervart, J. Johnson, Z. Lukes, GPS Solutions, USA

As an island nation, which is hit by frequent earthquakes and severe storms, Japan has a compelling interest to observe the oceans that surround it. In order to achieve observations of waves, tides and tsunamis and also to monitor atmospheric water vapor for weather forecasting an array of ocean buoys is deployed by several government agencies in the seas surrounding Japan.
Hitachi Zosen Corporation (Hitz), headquartered in Osaka, Japan, has developed ocean-monitoring GPS buoys since 1996. The company has built fifteen large buoys, which are operated off the east coast of Japan´s main islands by the Ministry of Land, Infrastructure, Transport, and Tourism (MLIT). These buoys have a diameter of 5 meters, are 18 meters high and weigh 47.5 tons (not including the mooring chains and anchors). They have been designed to withstand wave heights of up to 30 meters and are equipped with a dual frequency GPS receiver that is powered with battery-charging solar panels and has real-time radio communications to shore. The buoys are moored, and the present design allows for deployment in water depths up to 300 meters using a mooring chain of 540 meters in length.

The large monitoring buoys off Japan´s coast are presently computing real-time kinematic (RTK) positions relative to a fixed reference station on land. These reverse RTK positions are radioed back to shore where they are used for routine monitoring of open water wave heights and to alert first responders to the arrival of tsunami waves. The buoys have captured several tsunamis in the past. This includes the tsunamis caused by the 2001 Peru earthquake (first detection of tsunami by a GPS buoy), the 2003 Tokachi-Oki earthquake, and the 2004 Tokaido earthquake.

While the reverse RTK buoys perform well and are very reliable, their range of operation is limited to 20 km from the nearest land based reference site. In the case of an approaching tsunami this gives only a very short warning period for residents on shore. To extend the warning period and to give people more time to escape to safety it is desirable to place buoys further offshore. This poses several challenges. Mooring and station keeping of the buoys in deep water is difficult and expensive. Buoys far from shore, in remote ocean regions, are often vandalized and equipment such as solar panels is an attractive target for thieves. Also buoys that are far offshore require different means of communications for real-time data transmission. But most importantly for the study shown in this presentation, accurate positioning of buoys far from shore poses a challenge because the reverse RTK technique that is currently employed will not work any longer.

Tsunami waves in the open ocean travel at fast speeds and have long wavelengths and small vertical amplitudes. Therefore the ocean buoys require about 5-cm vertical positioning capability in order to detect even a small tsunami in open water.

In order to achieve 5-cm vertical positioning of a buoy in the open ocean far from the nearest reference site, we have developed, together with GPS Solutions of Boulder, CO, the GNSS data processing method of precise point positioning with ambiguity resolution (PPP_AR). The method of PPP-AR has been described previously by Mervat et al. (ION-GNSS, Savannah GA, 2008). Since then the processing technique has been refined, made more reliable and modified to work in true real time.

PPP_AR requires operation of a dual frequency reference network that is used to estimate the GPS clocks and the non-integer satellite specific biases that need to be removed in order to resolve the integer L1 and L2 carrier phase ambiguities on the between-satellite single difference level. In Japan we can use data from the 1300-station GEONET network, which is operated in real-time with 1-Hz sampling by the Geographical Survey Institute (GSI). It is not necessary to process all data from Geonet and for this study we process a sub-network of only ten stations to generate the PPP_AR clocks and bias corrections. These 10 stations are spaced on average by 250 km and span Japan from north to south. Data from this network are processed in real-time and it is important that as many carrier phase ambiguities are resolved over these 250 km baselines as possible in order to obtain good PPP_AR corrections. These PPP_AR corrections are then used by a client user site for precise point positioning with ambiguity resolution.

To demonstrate that it is indeed possible to achieve 5 cm vertical positioning in PPP mode at large distances from the reference network we have conducted tests with static clients in known locations that we position in kinematic mode, just as we would do with a buoy. The results from these tests indicate that 1-2 cm horizontal and 2-4 cm vertical kinematic PPP_AR positioning is feasible up to 1000 km from the reference network.

For additional tests we also have access to GPS data from a buoy, which is operated close to shore by the University of Tokyo. This buoy is located next to a sonic wave-monitoring tower. We are thus able to compare tides estimated in PPP_AR mode and in standard RTK mode to the observations from the sonic tower.

This presentation will provide an overview of Japan´s ocean buoy program with emphasis on the Hitz built buoys. We will present results from recent developments to position buoys with PPP_AR in real time and discuss if the achievable accuracy of this positioning mode is sufficient for a tsunami warning buoy system.

[Return to Program]