Proceedings of the ION GNSS 2004, Long Beach, Session D4, 22-24 Sept. 2004

Monitoring the Neapolitan Volcanic Area Using an Advanced Multiple Reference Station RTK

DGPS Technique

Giovanni Pugliano, Parthenope University of Naples Franco Obrizzo, Folco Pingue, Vincenzo Sepe, Osservatorio Vesuviano-INGV, Naples

Paul Alves, Gérard Lachapelle, The University of Calgary BIOGRAPHY Giovanni Pugliano received his Ph.D. in Geodetical and Topographical Science from the Parthenope University of Naples. He is currently employed at the Department of Geomatics, Parthenope University of Naples, where he works in the field of precise GPS applications. Franco Obrizzo is a geophysical researcher at the Osservatorio Vesuviano- INGV. His area of expertise is the volcano monitoring, especially ground deformation analysis and refinement of models for dynamical behavior of many volcanoes in the Southern of Italy. Folco Pingue is a geophysical researcher at the Osservatorio Vesuviano- INGV. His research interests are in volcanic and tectonic areas monitoring, especially in ground deformation analysis with several methods (GPS, Leveling and others). His research interests involve development of theoretical models for volcanic and tectonics areas modeling and for computing static stress changes. Vincenzo Sepe is a geophysical researcher at the Osservatorio Vesuviano- INGV. His main application fields are in volcano monitoring, especially in ground deformation analysis by comparative studies of datasets from different techniques (GPS, Leveling and others) for understanding the behavior of many volcanoes in the neapolitan volcanic area. Paul Alves is a graduate student at the Department of Geomatics Engineering of the University of Calgary. He received a B.Sc. in Geomatics Engineering in May, 2000, and is continuing his studies towards a Ph.D. in Geomatics Engineering in the field of positioning and navigation at the University of Calgary. Professor Gérard Lachapelle holds a CRC/iCORE Chair in Wireless Location in the Department of Geomatics Engineering. He has been involved with GPS developments and applications since 1980 and has authored/co-authored numerous related publications and

software. More information is available on http://plan.geomatics.ucalgary.ca/ ABSTRACT The Neapolitan Volcanic area, located in the southern sector of the Campania Plain (Italy), includes the Somma-Vesuvius volcano, the Campi Flegrei area and the islands of Ischia and Procida. The presence of the active volcanoes in a very dense area needs continuous monitoring of the dynamics to study the pre-eruptive processes. Ground deformation represents an important precursor because it is linked to magma overpressure and migration through the rehological parameters characterizing the volcanic rocks. Thanks to a greater rapidity, real-time kinematic (RTK) positioning can be useful for periodic surveys and for quickly solutions to field problems in periods of crisis. In particular, it is of great interest to effectively use the high density of permanent stations of the Osservatorio Vesuviano-INGV GPS surveillance network, located in the Campi Flegrei volcanic district. This paper presents the results of a test based on the use of an advanced post-mission multiple reference station RTK DGPS positioning method, namely MultiRefPM™. In general, this multiple reference station approach can provide a position accuracy considerably better than the standard single reference station approach. The resulting position accuracy is smaller than the usual volcanic deformation and can therefore be used as a cost-effective monitoring technique. INTRODUCTION The Neapolitan Volcanic area includes the Somma-Vesuvius volcano, the Campi Flegrei area and the islands of Ischia and Procida. In particular Campi Flegrei is a caldera complex located to west of the city of Naples. The dynamics of this volcanic field was characterized by slow and continuous vertical movements as well known as Bradyseism. The secular trend of ground movement at Campi Flegrei is subsidence of the caldera. Superimposed on this long-term trend, some fast and intense episodes of ground uplift occurred (up to 2 m vertical in two years). Starting from 1969, Campi Flegrei was interested by two

Proceedings of the ION GNSS 2004, Long Beach, Session D4, 22-24 Sept. 2004

main uplift phases (1969-1972 and 1982-1984) and others minor with a maximum vertical displacement of about 3.5m and with more than 15,000 earthquakes with magnitudes in the range 0.5-4.0. The presence of the active volcanoes in a very dense area needs continuous monitoring of the dynamics related to the pre-eruptive processes. Ground deformation represents an important precursor because it is linked to magma overpressure and migration through the rehological parameters characterizing the volcanic rocks. At the Osservatorio Vesuviano-INGV the geodetic monitoring system is mainly based on GPS and precise leveling techniques: the Geodesy department installed and operates a GPS permanent network and spirit leveling networks. Thanks to a greater rapidity, the real-time kinematic (RTK) positioning can be useful for periodic surveys and for quickly solving field problems in period of crisis. In particular, in regard to the Campi Flegrei volcanic district, it is of great interest to focus the analysis on the possibility to take advantage of the high density permanent network of the Osservatorio Vesuviano-INGV GPS surveillance network to operate with a single receiver. Several methods based on the use of GPS reference networks for real-time kinematic positioning have been proposed and tested in recent years. The use of such methods is advantageous to overcome some of the limitations of the standard single reference station differential carrier phase positioning method. Accuracies at the sub-decimetre level are possible under ideal conditions. Based on the one-way transmission of corrections from some reference station to the user, Network RTK systems use network information to assist the rover to calculate accurate and reliable navigation and positioning solutions. An example of such a real-time method is MultiRef™, developed by the Department of Geomatics Engineering, University of Calgary, during the past several years [e.g., Lachapelle et al. 2000; Cannon et al. 2001] and tested in Italy in 2002 [Pugliano 2003]. An enhanced method, namely MultiRefPM™, whereby the user receiver becomes part of the network of reference stations to more effectively deal with error modeling and carrier phase ambiguity resolution is presented herein. In real-time, two-way communication between the user and reference stations is required. There are no drawbacks in post-mission mode and the method is thus ideally suited for the stringent availability and reliability requirements for precision surveying of active volcanic areas. In the sequel of this paper the basic algorithms are described and the results of a test carried out in the Campi Flegrei area are reported. NETWORK RTK GPS POSITIONING The original MultiRef™ approach uses least-squares collocation to predict the differential correlated errors at any location within the network coverage area [Raquet 1998]. The following least-squares collocation equations are used to estimate the corrections for the network reference stations and the rover locations, respectively:

)(B)B(BCBCd

)(B)B(BCBCdT

lT

,llr

Tl

Tl

rNl

Nl

∇∆−Φ=

∇∆−Φ=−

−

λ

λ1

1

ˆ

ˆ (1)

where l

)δ is a vetor of corrections,

l,dlrC δ

is the covariance

matrix between the corrected observations and the network observations, B is the double differencing matrix,

dlC is the variance-covariance matrix of the network

observations, Φ is a vector of the network observations in meters, and N∇∆λ is a vector of the double difference ambiguities in meters. Equation (1) shows that the corrections are a function of the mathematical relationship between the baselines (B), the stochastic relationship between the baselines ( dlC ) and the stochastic relationship

between the baselines and the rover (l,dlr

C δ). When the

corrections are applied, this information is combined with the stochastic relationship between the rover and a single reference station as well as the mathematical relationship due to the double differencing of the measurements. Network RTK procedures estimate the error at the rover by a weighted-average of the surrounding reference station’s measured error. Some systems use a plane to determine the weights of the surrounding stations [Wanninger et al 1999, Vollath et al 2000; Wübbena et al 2001], while others use least-squares to determine the weights [Raquet 1998, Landau et al 2002]. In either case, the errors measured at each reference station are transferred to the rover through mathematical and stochastic modeling. In planar interpolation, each reference station is weighted as a linear function of the distance to the rover only. Using least squares collocation allows for more complicated weighting schemes given by the covariance function. This could weight not only the influence of each reference station but also each satellite error at each of the reference stations. The covariance function could be a function of receiver separation, angular separation between the satellites, measurement noise, ionosphere pierce point distance, or elevation and azimuth. For actual real-time applications, the data from each reference station is transmitted to a central processing centre where the carrier phase corrections are generated. The corrections are then applied to the observations of one of the reference stations. The rover can then use these corrected reference station observations with standard RTK software. MULTIREFPM™ Network RTK implementation consists of three main steps [Lachapelle and Alves 2002]. In the first step, the errors at the reference stations are estimated using carrier-phase observations. The second step interpolates these errors to the rover receiver location whereas the third step is to transmit the corrections to the rover. This is usually

Proceedings of the ION GNSS 2004, Long Beach, Session D4, 22-24 Sept. 2004

carried out by generating virtual reference station data that the rover can accept, using a single reference station data format. Real-time network kinematic positioning is limited by many factors, one of which is the communication network used between the network control centre and the rovers. Due to bandwidth limitations with multiple rovers and an attempt to allow for user privacy, real-time network kinematic positioning methods have attempted so far to operate a broadcast-only system (one-way communication), whereby the network corrections are broadcast to all rovers and there is no information communicated from the rover back to the network. If two-way communication is used, not only can the network stations assist the rover but the rover can also assist the network with additional information. In this case the rover actually becomes part of the network and the reduced inter-receiver distances and additional ambiguity constraints provided by the rover improve the overall ambiguity resolution process very significantly using the now established receiver multiplicity concept initially proposed by Lachapelle et al. [1993] and further tested by Luo and Lachapelle [2003]. This enhanced procedure is also ideal for post-mission applications that are numerous for verification of hydrographic surveys, airborne surveys and land surveying. Network RTK systems use reference stations to precisely measure the correlated errors affecting the region. These errors can only be measured when all other parameters are precisely determined, namely the station position and carrier phase ambiguities. With this in mind, the better a station’s position and ambiguities are known, the more accurately one can separate measurement errors and systematic biases. Reference stations are an obvious choice because their positions are known, but any receiver can be used to estimate measurement errors. For example, a static or kinematic rover can be treated as a reference station. In terms of error modelling, multiple rovers in an area can each give an indication of the local environmental error conditions. Combining all of this information into a coherent model allows for new network rovers, with less defined position and velocity estimates, to benefit from decreased measurement error. The assistance of the rover to the network can be seen in the baseline configurations for the network. Ambiguity resolution performance is a function of inter-receiver distance separation because the correlated errors increase in magnitude as the separation increases. In a broadcast-only Network RTK system, baselines are formed between the various reference stations. Rovers within the network will, by definition, be between two or more reference stations. Therefore connecting baselines to the rovers as well as the reference stations will shorten the overall network inter-receiver separations within the network, thus giving a higher likelihood of resolving the carrier phase ambiguities.

Instead of applying a weighted average (prediction) approach, the rover’s data and estimated states are added to the network filter. The network filter is used solely to estimate and resolve the network ambiguities in the real-time approach. The addition of the rover’s information into the network filter maintains all the information used in the least-squares collocation approach (Equation 1) and adds the rover information. The difference is that the network not only assists the rover but the rover also assists the network. The design matrix of the above integrated approach is

∂Φ∇∆∂

∂Φ∇∆∂

∂Φ∇∆∂

∂Φ∇∆∂

∂Φ∇∆∂

∂Φ∇∆∂

=

O

O

λλ

λ

λ

000000000000

000

000

222

111

zyx

zyx

A

(2) where the first n rows corresponds to the double difference observations between the rover and one of the reference stations and the second set of m rows corresponds to double difference observations between the fixed reference stations with known coordinates, n being the number of double difference observations between the rover and the reference station(s) and m is the number of double difference observations between reference stations. The first three columns correspond to the rover’s position estimates and the following n + m columns correspond to the ambiguities of all of the double difference observations. No partial derivatives with respect to the coordinates exist between reference stations because they do not directly observe the rover’s position. The design matrix can be extended to accept any number of reference stations and rovers. The processing results shown include the code and carrier phase observations processed in a single Kalman filter. This model can be expanded to incorporate any observation (system) model (estimating troposphere delay, ionosphere error, or the rover’s velocity estimates, for example). The selection of the double difference observables is based on the shortest inter-receiver separations, with the conditions of linear independence and connectivity being preserved. Thus a rover may be connected to one or several reference stations, depending on the reference station-rover receiver configuration. Short distances are selected to minimize the magnitude of the differential errors. As an example, in the case of four reference stations and one rover, the double differences over the shortest four linearly independent receiver separations would be used. The rover may be involved in one to four sets of double differences.

Proceedings of the ION GNSS 2004, Long Beach, Session D4, 22-24 Sept. 2004

In order to maintain the information from the correction-based approach, mathematical and stochastic information must be preserved in the integrated approach. The mathematical correlation is due to inter-receiver separations that share a common reference station (or rover station) which use the same observations in their double difference measurements. Stochastic correlation is defined by the covariance function. The covariance function states the likelihood of two values being the same based on a physical process. For example, it is known that the ionosphere is a spatially correlated error, therefore two stations close to each other are likely to have a similar ionospheric error. This likelihood is represented in the stochastic correlation. POSITION AND VELOCITY ESTIMATION In general, rover receivers are moving. In the test results presented, static receivers are used because their positions are known and provide a truth with which to compare with the MultiRefPM™ method. However, kinematic receiver estimation is implemented whereby both the rover position and velocity are determined at each epoch. Velocities are estimated using a first order Gauss-Markov process. In the case of the MultiRefPM™ approach, all of the network ambiguities are estimated in the same Kalman filter as the position and velocity estimates. When the ambiguities are fixed, the float solution is adjusted to take into account the new, fixed ambiguity information. This is done using the following conditional decorrelation [e.g., Teunissen and Kleusberg 1998, Odijk 2002]:

floatfloatfloatfloatfloatfloatfixed

floatfloatfloat

baaabbb

fixedfloataabfloatfixed

CCCCC

aaCCbb

,1

,

1, )(

−

−

−=

−−= (3)

where fixedb is a vector containing the estimates of all the

floating parameters (position, velocity, ionosphere and unfixed ambiguities) adjusted given the fixed ambiguities,

floatb is a vector of the floating parameters before

adjusting for the fixed ambiguities, floatfloat abC , is the

covariance matrix between the floating parameters and the float ambiguities that have been fixed, 1−

floataC is the

variance-covariance matrix of the float ambiguities that have been fixed, floata is a vector of the float ambiguities

that have been fixed, and fixeda is a vector of the fixed

ambiguities. Note that not only the values of the estimated parameters change. The variance-covariance matrix of the floating parameters is also adjusted to account for the fixed ambiguity information. THE NEAPOLITAN VOLCANIC AREA The Neapolitan Volcanic area is a densely populated area (about 1,500,000 people) with the presence of several dangerous volcanoes characterized by explosive eruptive mechanisms which make the monitoring of eruption precursors for correct hazard mitigation very important. The study of the deformation sources in the volcanic apparatus, and the need for reliable evaluations and forecast of pre-eruptive phenomena, stimulated the development, optimisation and the continuous technological evolution of geodetic monitoring systems, both for continuous acquisition and for periodic surveys [Beauducel et al. 2004].

Figure 1: The Neapolitan Volcanic Area, in Italy

Proceedings of the ION GNSS 2004, Long Beach, Session D4, 22-24 Sept. 2004

The area includes Vesuvius to east of Naples, the Campi Flegrei area and the islands of Ischia and Procida located to west of the Bay of Naples. The last eruption of Vesuvius occurred in 1944. It is currently in a quiescence phase characterized by a low deformations and seismic activity. In 1999, there was an increase in seismic activity in the area. The activity culminated with an earthquake at 07:41 GMT on October 9, (Md = 3.6, Lat. 40°49.01’ Lon. 14°25.67’, depth ≈ 4.0 Km) and was not accompanied by significant ground deformation. The Campi Flegrei area is characterized from a caldera 35,000 years old (eruption Campanian Ignimbrite) by which numerous volcanic monogenic apparatus developed inside. The last eruption of this caldera rose again in 1538 and carried to the origin of Mount Nuovo. From the geological point of view, the caldera is mainly formed by volcanic rocks and subordinately by clastic sea sediments; from the structural point of view, the configuration of Campi Flegrei is the result of deformations related to the regional and volcano-tectonic events. The regional tectonic is the cause of direct faults in the NE-SW and NW-SE direction and subordinately in the NS direction. The magmatic chamber is located at low depth (about 4-5 km). The dynamics of this volcanic field was characterized by slow and continuous vertical movements as well known as Bradyseism. During 1969-1972 (maximum uplift of 170 cm) and 1982-1984 (maximum uplift of 184 cm) this area has experienced two intense episodes of strong uplift of the ground and moderate seismic energy activity. Even though the two main uplift crises are not culminated into an eruption, these events caused significant damages to the buildings and to the economy of the Campi Flegrei, which has 250,000 inhabitants. Both of the episodes were followed by a phase of subsidence. In particular, since late 1984 the caldera has undergone a subsidence phase, with an average rate of about 5 cm/year, which was interrupted by brief inflation in 1989, 1994 and more recently during the

period March-September 2000. High precision spirit leveling data from four different campaign of measurements on September 1999, May, June and July 2000, in keeping with GPS static data, pointed out an inflation period with a maximum uplift of about 4 cm and a horizontal displacement of about 2 cm. Figure (2) illustrates the vertical displacement at benchmark 25, located close to maximum uplift area, during the period January 1985 - December 2002.

Figure 2: Vertical displacement at benchmark 25, starting from 1985 Ischia Island was disrupted by the last eruption in 1302 and by two strong earthquakes (about 3000 victims) respectively in 1881 and 1883. At present, the island is characterized by a subsidence phenomenon in the South and Northwest areas and an almost total absence of seismicity. Geodetic monitoring by the Osservatorio Vesuviano in the Neapolitan Volcanic area is carried out by continuous and periodic networks [Pingue et al. 2003]. In figure (3) the geodetic monitoring system is shown. It is mainly based on GPS and precise leveling networks.

Figure 3: GPS and leveling networks operating on the Neapolitan Volcanic Area

Proceedings of the ION GNSS 2004, Long Beach, Session D4, 22-24 Sept. 2004

The leveling networks consist of: the Vesuvian Leveling Network formed by about 300 benchmarks distributed on a distance of about 225 km, with a mean distance of about 700 m, on sixteen loops, the Campi Flegrei Leveling Network formed by about 300 benchmarks distributed on a distance of 120 km, with a mean distance of about 400 m, on eleven loops, and the Ischia Island Leveling Network formed by 200 benchmarks distributed on a distance of about 90 km, with a mean distance of about 300 m, on seven loops. The GPS Network covers the whole neapolitan volcanic area (Vesuvius, Campi Flegrei and Ischia). Its goal is to define the global dynamic of the region, binding itself to campanian apennine massifs. The CGPS Network consists of 20 Permanent GPS Stations, covering the three volcanic zones. This network is particularly useful for detecting phases of rapid deformations. In particular, regarding the Campi Flegrei area, the GPS network was established in 1997. It consists of 35, 3D vertices, eight of which are permanent stations. The geometry of Campi Flegrei network was studied in order to monitor the principal geological structures allowing it to estimate the deformation field acting in the area with high accuracy. The network configuration defines three main N-S lines and two main E-W lines. TEST NETWORK DESCRIPTION The MultiRefPM™ method is employed for a test in the Neapolitan Volcanic area. This test was conducted to investigate the available precision and reliability of RTK positioning for tasks such as ground movement studies in volcanic areas. The Campi Flegrei area was selected as a test zone (figure 4)

Figure 4: The Campi Flegrei Area

The network consists of five GPS reference stations extending throughout the area. Four of the receivers (BAIA, IPPO, MORU and RITE) were part of the existing CGPS Network. These were all Trimble 4000SSI dual frequency receivers using Dorne-Margolin ground plane antennas. The other receiver (NISI) was a dual frequency Leica SR530 which was temporarily set up for this test. The RITE receiver, located very close to the maximum deformation area, was used as a rover. The campaign of measurements was carried out during the month of March, 2004. Data was collected at a 1 Hz rate over a 48-hour period starting at 0:00 UTC (1:00 local) on March 10th. To test the possibility of utilizing RTK DGPS positioning for the monitoring of the active structures inside the area, four additional Leica SR 530 dual frequency receivers were added. These receivers were on kinematic stations with known coordinates and raw measurements at 1 Hz rate were collected. There were two days of testing. The data were collected for six hours on the first day and for seven hours on the second. During this period, the receivers remained stationary for 30 to 40 minutes at each point. The route covers the whole area of maximum deformation along the main E-W direction, namely the coast line, and the main N-S direction. The placement of each of the kinematic stations was carried out close to known leveling benchmarks, mainly optimized for studying surface deformation problems. Some kinematic stations were poor site because of obstructions to the signals of the receivers. As a result, data from a total of 30 points were used for the analysis presented in this paper. Figure (5) illustrates the final layout of the GPS static and kinematic stations on the Campi Flegrei area.

Proceedings of the ION GNSS 2004, Long Beach, Session D4, 22-24 Sept. 2004

Figure 5: Relative locations of the GPS static and kinematic stations in the Campi Flegrei area

TEST METHODOLOGY The MultiRefPM™ method, like other multiple reference station RTK DGPS positioning methods, is based upon the fundamental premise that the coordinates of the reference stations are known and that they are precise. In this study the coordinates of each of the receivers were calculated by the Bernese software version 4.2 [Hugentobler et al., 2001] using data thinned to 30 second intervals with an elevation mask of 15°; IGS final-type precise ephemerides were used in order to minimize orbital error effects. After having generated the session solutions for each day of measurement, the final network coordinates were obtained by carrying out an overall adjustment of the two daily solutions. In a comparison of the different session’s solutions and the final solution, the results reveal a high agreement at the millimetre level. In addition to providing a proper geometry for monitoring purposes, the kinematic stations allow for a great variety of scenarios. Two study was conducted analyzing the position domain. On the RITE kinematic station MultiRefPM™ processing was conducted on 24 hours of data to test the method in terms of the temporal position accuracy. The other test conducted uses the 30 kinematic stations, which provide different spatial conditions. The test objective of this investigation is to evaluate the precision and reliability of the estimated coordinates, and the ability and time to fix carrier phase ambiguities. The networked (MultiRefPM™) solutions are compared to the raw (single reference station RTK) solutions obtained using the nearest reference station.

TEST RESULTS FOR 30 KINEMATIC STATIONS This analysis is conducted in the position domain. The objective is to analyze the effects of the MultiRefPM™ approach on the accuracy of the rover position. More specifically, the evaluation is based on the difference between the raw and MultiRefPM™ solutions, epoch-by-epoch, and the known coordinates obtained using a Bernese batch method. The use of an epoch-by-epoch technique in this case is important for a meaningful sensitivity analysis. In operational mode, the positions would be filtered and the short-term noise would be further reduced. In reference to the 30 kinematic stations, Figure (6) shows the 3D position errors for all thirty cases. The comparison of the solutions clearly indicates the benefits of the MultiRefPM™ method.

Figure 6: 3D position error for 30 kinematic stations for the raw and MultiRefPM™ solutions

Proceedings of the ION GNSS 2004, Long Beach, Session D4, 22-24 Sept. 2004

The position errors in latitude, longitude and height were also calculated; and Figure (7) shows the results for all thirty cases obtained at each epoch using uncorrected (single reference station RTK) as well as networked (MultiRefPM™) measurements.

Figure 7: Position errors for 30 kinematic stations using raw and MultiRefPM™ methods In both cases, the best available ambiguities are used begin either float and fixed. When using the single reference station RTK mode the ambiguities are often unable to be resolved. With the MultiRefPM™ approach, ambiguities are also sometimes unresolved but, thanks to the superior modelling of the measurement errors, the position errors remain below 1 dm. This result is truly

remarkable as the approach advocated herein allows the user to continuously obtain a sub-dm accuracy, regardless of whether the carrier phase ambiguities are resolved as integer values or not. The MutliRefPM™ method shows an improvement in accuracy in all three position components. Comparing the solutions, improvements of up to 97%, 99% and 96% in terms of RMS (figure 8) were observed. Table (1) summarizes the results of the comparisons for all thirty cases.

Figure 8: Comparison of RMS position errors for 30 kinematic stations

Table 1: Raw and MultiRefPM™ 3D RMS position error values for 30 kinematic stations Kin. Stations Lat. (m) Lon. (m) H (m)

Raw Network Raw Network Raw Network 1 2 3 4 5 6 7 8 9

10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

0.01 0.02 0.02 0.09 0.01 0.26 0.01 0.01 0.12 0.01 0.36 0.10 0.01 0.07 0.01 0.12 0.12 0.01 0.03 0.34 0.02 0.01 0.02 0.15 0.02 0.01 0.01 0.01 0.12 0.08

0.01 0.01 0.01 0.02 0.01 0.06 0.01 0.01 0.01 0.01 0.01 0.01

< 0.01 0.01 0.01 0.01 0.01 0.01 0.02 0.02 0.01 0.01 0.01

< 0.01 0.01 0.01 0.01 0.01 0.01 0.04

0.02 0.02 0.02 0.08 0.01 0.75 0.01 0.01 0.08 0.01 1.03 0.05 0.01 0.02 0.01 0.02 0.23 0.01 0.10 0.22 0.01 0.01 0.02 0.06 0.01 0.01 0.02 0.02 0.02 0.02

0.01 0.01 0.02 0.01 0.01 0.02

< 0.01 < 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.02 0.01 0.01 0.01 0.04 0.01

< 0.01 0.01

< 0.01 0.01 0.01 0.02 0.02 0.02 0.02

0.02 0.08 0.07 0.17 0.08 0.17 0.08 0.05 0.25 0.03 0.46 0.09 0.05 0.04 0.02 0.17 0.21 0.05 0.38 0.38 0.09 0.06 0.11 0.41 0.06 0.03 0.06 0.03 0.12 0.15

0.02 0.07 0.06 0.07 0.07 0.03 0.06 0.05 0.05 0.04 0.02 0.03 0.05 0.02 0.03 0.05 0.02 0.06 0.05 0.10 0.03 0.07 0.04 0.03 0.06 0.03 0.06 0.03 0.06 0.08

Proceedings of the ION GNSS 2004, Long Beach, Session D4, 22-24 Sept. 2004

TEST RESULTS FOR A 24-HOUR PERIOD The MultiRefPM™ processing was conducted for a 24-hour period on a RITE kinematic station, located very close to the maximum deformation area. The initial investigation was conducted in the position domain. Similar to the previous section, the position errors were examined to asses the benefits of the MultiRefPM™ method. Figure (9) illustrates the position errors in terms of latitude, longitude and height related for the RITE station.

Figure 9: Position errors for RITE stations using raw and MultiRefPM™ methods Table (2) shows the results for the raw and network cases. The network solution maintains a higher accuracy level than that of the raw case. In addition to confirming the

improvements ensuing from the utilization of the MultiRefPM™ method, this type of analysis highlights the achievable accuracies (at the cm level or better) which make this method applicable to monitoring horizontal and vertical displacements in the deformation area. Table 2: Raw and MultiRefPM™ RMS and Max position error values for RITE stations

Coord. Component RMS (cm) Max (cm)

Raw Network Raw Network

Latitude 2.4 0.4 48.9 2.3 Longitude 5.4 0.4 30.8 2.3 Height 5.9 1.1 131.6 7.8

Having verified the contribution of the MultiRefPM™ method in the position domain, it is nonetheless necessary to analyze the effects on carrier-phase ambiguity resolution. Figure 10 outlines the improvements related to performance in general in the ambiguity domain. The benefits resulting from MultiRefPM™ are evident in terms of the percentage of fixed ambiguities, which increased from 84% to 94% for the raw and MultiRefPM™ methods, respectively.

Figure 10: Ambiguity resolution for RITE stations for a 24-hour period

Proceedings of the ION GNSS 2004, Long Beach, Session D4, 22-24 Sept. 2004

CONCLUSIONS The multi-reference station DGPS RTK technique constitutes an important advancement to the field of precision surveying. The comparative analysis between the standard single reference station differential carrier phase method and MultiRefPM™ clearly shows the benefits resulting from this method. A most remarkable result obtained during the field test was the ability of the user to maintain sub-decimetre accuracy when the carrier phase ambiguities were not resolved as integer values. The simultaneous use of multiple reference stations has resulted in high reliability accuracies of 1 dm or better in the Campi Flegrei area, making the multi-reference station DGPS RTK technique applicable to the study of Campi Flegrei, as well as for other similar area. This accuracy level is of course available in three-dimension, where using only short observation time spans reduces the observation time currently required by classical leveling technique, while increasing periodic surface deformation measurements and improving safety. The method offers obvious advantages in post-mission mode. Given the test results presented herein, the method appears to be sufficiently promising for testing under actual real-time conditions. REFERENCES Beauducel, F., G. De Natale, F. Obrizzo and F. Pingue

(2004). 3-D Modelling of Campi Flegrei Ground Deformations: Role of Caldera Boundary Discontinuities, Pure and Applied Geophysics, 161, 2004, 1329-1344.

Cannon, M.E., G. Lachapelle, L.P. Fortes, P. Alves, B.

Townsend. (2001). The Use of Multiple Reference Station VRS for Precise Kinematic Positioning, Proceedings of the Japan Institute of Navigation, GPS Symposium 2001 (Tokyo), 29-37.

Lachapelle, G., C. Liu, and G. Lu (1993). Quadruple

Single Frequency Receiver System for Ambiguity Resolution on the Fly. Proceedings of the 6th International Technical Meeting of the Satellite Division of the U.S. Institute of Navigation (Salt Lake City), 1167-72.

Luo, N., and G. Lachapelle (2003). Relative Positioning of

Multiple Moving Platforms using GPS. IEEE Transactions on Aerospace and Electronic Systems, 39, 3, 936-948.

Lachapelle, G., and P. Alves (2002) Multiple Reference

Station Approach: Overview and Current Research. Invited Contribution, Expert Forum on VRS, Journal of Global Positioning Systems, 1, 2, 133 – 136.

Lachapelle G., P. Alves, L.P. Fortes, M.E. Cannon, B.

Townsend. (2000). DGPS RTK Positioning Using a Reference Network, Proceedings of the 13th

International Technical Meeting of the Satellite Division of the U.S. Institute of Navigation (Salt Lake City),. 1165-1171.

Landau, H., U. Vollath, and X. Chen (2002) Virtual

Reference Station Systems, Expert Forum on VRS, Journal of Global Positioning Systems, Vol. 1, No. 2, 137 – 143.

Odijk, D. (2002). Fast Precise GPS Positioning in the

Presence of Ionospheric Delays. Ph.D. Thesis, Optima Grafische Communicatie, the Netherlands.

Pingue, F., G. De Natale, C. Del Gaudio, F. Obrizzo, V.

Sepe, G. Cecere, P. De Martino, S. Malaspina, C. Serio, V. Siniscalchi, U. Tammaro. (2003). Monitoring and deformation analysis in Neapolitan volcanic area, Proceedings of 11th FIG Symposium on Deformation Measurements (Santorini).

Pugliano, G. (2003). Testing of a Multi-Reference Station

Network for RTK DGPS Positioning in Italy, Proceedings of the 16th International Technical Meeting of the Satellite Division of the Institute of Navigation (Portland), 1406-1416.

Raquet, J. (1998). Development of a Method for Kinematic

GPS Carrier-Phase Ambiguity Resolution Using Multiple Reference Receivers, PhD Thesis, UCGE Report Number 20116, Dept of Geomatics Engineering, University of Calgary. (http://PLAN.geomatics.ucalgary.ca)

Teunissen, P.J.G. and A. Kleusberg. (1998). GPS For

Geodesy, Springer, 2nd Edition. Vollath, U., A. Beucherl, H. Landau, C. Pagels, and B.

Wagner (2000) Multi-Base RTK Positioning Using Virtual Reference Stations. Proceedings of the 13th International Technical Meeting of the Satellite Division of the Institute of Navigation (Salt Lake City), 123 - 131.

Wanninger, L. (1999) The Performance of Virtual

Reference Stations in Active Geodetic GPS-netowrks under Solar Maximum Conditions. Proceedings of the 12th International Technical Meeting of the Satellite Division of the Institute of Navigation (Nashville), 1419 - 1427.

Wübbena, G., A. Bagge, and M. Schmitz (2001) RTK

Networks based on Geo++ GNSMART – Concepts, Implementation, Results. Proceedings of the 14th International Technical Meeting of the Satellite Division of the Institute of Navigation (Salt Lake City), 368 - 378.