Place and date: University of Nottingham, Nottingham, UK, Thursday 27^th and Friday 28^th February 2003

Halfdan Kierulf (ESEAS-RI P1, NMA, Norway)

Hans-Peter Plag (ESEAS-RI P1, NMA, Norway)

Simon Williams (ESEAS-RI P3, NERC-POL, UK)

Richard Bingley (ESEAS-RI P4, UNOTT, UK)

Norman Teferle (ESEAS-RI P4, UNOTT, UK)

Jorge Garate (ESEAS-RI P7, ROA, Spain)

Coskun Demir (ESEAS-RI P9, GCM, Turkey)

Hrvoje Mihanovic (ESEAS-RI P12, HHI, Croatia)

Juan Jose Martinez-Benjamin (ESEAS-RI P16, UPC, Spain)

Leszek Jaworski (ESEAS-RI P20, SRC, Poland)

Ryszard Zdunek (ESEAS-RI P20, SRC, Poland)

1. Opening of the meeting and adoption of the agenda

2. Overview of ESEAS-RI WP2 and tasks T2.1 to 2.8

3. Introduction to task T2.4: Assemble information on physical parameters that may affect GPS coordinate time series

4. Detailed discussion of tasks T2.1 and T2.3: ESEAS GPS data analysis

5. Detailed discussion of task T2.2 and T2.5: ESEAS GPS coordinate time series analysis

6. Introduction to task T2.6: Carry out absolute gravity measurements at selected ESEAS Observing Sites

7. Introduction to task T2.7: Assemble information on (historical and current) precise levelling at all ESEAS Observing Sites that are co-located with GPS@TG stations

8. Agreement on work programme and activities of all partners for next 6 months

9. Any other business

10. Date and venue for next meeting of ESEAS-RI WP2 and ESEAS WG2

Richard Bingley opened the meeting. Alan Dodson, the Director of the IESSG at the University of Nottingham, welcomed all participants and wished them a successful meeting.

The agenda was adopted without changes and it was agreed that all of the presentations made at the meeting would be made available as pdf file attachments to the minutes.

An overview of ESEAS-RI WP2 and tasks T2.1 to 2.8 was given by the workpackage leader, Richard Bingley (see the pdf file of the presentation for the full details).

A general discussion on the specifications for any new GPS equipment to be installed at tide gauges took place. It was agreed that the ESEAS CB should make some recommendations that the receiver and antennas should be consistent with those used by the IGS (eg AOA Benchmark, Ashtech micro-Z/iCGRS or Trimble 4000SSI receivers equipped with Dorne Margolin choke ring antennas, set to record at 30 second epoch intervals, with a 5 degree elevation angle cut-off) and that the software used to convert the raw binary data to RINEX format should be consistent with that used by the IGS (eg tecq, with default settings and the preservation of the 30 second epoch interval).

A general discussion on the need to establish the ESEAS GPS data centre and archive took place. Richard apologised for not inviting Guy Woppelmann to this meeting and asked if a revised French proposal had been submitted to the ESEAS Central Bureau following the ESEAS-RI Kick-Off Meeting in Istanbul. Hans-Peter Plag reported that a revised proposal had not been received and that the issue relating to the identification of ESEAS GPS data in a combined IGS TIGA-PP/ESEAS GPS data archive still had to be resolved. It was agreed that, as Director of the ESEAS CB and Coordinator of the ESEAS-RI project, Hans-Peter would contact Guy regarding the status of the revised French proposal and the above issue. Furthermore, it was agreed that if a solution was not imminent then it may be necessary to establish a temporary data centre to enable GPS data analysis to start. It was noted that NMA, Norway could take on the responsibility of being the temporary data centre, if necessary.

An overview of task T2.4 was given by the task leader, Hans-Peter Plag (see the pdf file of the presentation for the full details).

Hans-Peter presented the background information relevant to the task and showed how the information available from task T2.4 may affect tasks T2.1, T2.2, T2.3 and T2.5.

Hans-Peter discussed the effects of solid Earth tides, ocean tide loading and atmospheric loading (which are included in the IERS conventions) and the effects of non-tidal ocean loading and terrestrial hydrological loading (which are not included in the IERS conventions). It was noted that, in principle, GPS data analysis should take account of those processes that cause significant station motion over the period of the solution, eg daily, and that any other processes could be considered when carrying out GPS coordinate time series analysis.

It was agreed that station motion models for solid Earth tides and ocean tide loading should be applied during the GPS data analysis, although the model (eg FES99), components (ie only vertical or horizontal and vertical) and constituents (eg M2 and S2 or all) to be used for ocean tide loading would have to be discussed further.

It was agreed that station motion models for non-tidal ocean loading and terrestrial hydrological loading should not be applied during the CGPS data analysis, but should be considered during the GPS coordinate time series analysis. At this stage it was noted that task T2.4 would be greatly assisted by the involvement of Hans-Peter in the IERS Special Bureau for Loading (IERS SBL), which will be creating research products including time series, and possibly near real-time predictions, of station motions due to different surface load processes. It was noted that the SBL will be producing these for the ITRF stations, but that this could be extended to include any ESEAS Observing Sites.

The main point for consideration at this stage was identified as being at what stage should station motion models for atmospheric loading be used? Hans-Peter stated that this was a current topic of research being carried out by the IERS SBL, as part of the preparation of the IERS conventions (2000) and that he would soon be able to report on whether station motion models for atmospheric loading should be used during the GPS data analysis or as part of the GPS coordinate time series analysis.

An overview of tasks T2.1 and T2.3 was given by the leader of both tasks, Halfdan Kierulf (see the pdf file of the presentation for the full details).

Halfdan explained that for GPS data analysis it was necessary to consider issues relating to reference frame, analysis strategy, physical models and inputs. For reference frame, the main question was whether to carry out the analysis as a separate or as a combined approach to the estimation of global parameters (ie satellite orbits, satellite clocks and Earth rotation) and local parameters (station coordinates, tropospheric ZTD, etc). For the analysis strategy, consideration needed to be given to whether to use precise point positioning solution based on undifferenced data (PPP) or a network solution based on double-differenced data (DD), and what softwares were available. For physical models, Halfdan explained that the issues were partly related the surface loading processes (and hence to task T2.4), but also included information relating to the satellites, ambiguity resolution, the atmosphere and clocks. Lastly, consideration should also be given to the inputs relating to different softwares, including weighting, iterations and cut-off elevations.

The overview given by Halfdan was followed by a series of brief presentations from each of the partners that will be involved in task T2.3, on their experience in static GPS data processing and their proposed contribution to the task (full details of the presentations given by each of the partners can be found in the respective pdf files).

- Halfdan Kierulf gave a presentation on behalf of NMA, Norway (see pdf file

- Norman Teferle gave a presentation on behalf of UNOTT, UK (see pdf file)

- Jorge Garate gave a presentation on behalf of ROA, Spain (see pdf file)

- Coskun Demir gave a presentation on behalf of CM, Turkey (see pdf file)

- Juan Jose Martinez-Benjamin gave a presentation on behalf of UPC, Spain (see pdf file)

- Ryszard Zdunek gave a presentation on behalf of SRC, Poland (see pdf file)

At this point it was noted that, following the response to the ESEAS 2^nd call for participation, other potential contributors to task T2.3 were the Swedish Geodetic Survey/Onsala Space Observatory and EUREF.

Halfdan Kierulf and Hans-Peter Plag stated that they intended to prepare the deliverable for task T2.1 as a document that would provide a detailed description of the ESEAS GPS data analysis strategy, including details of software specific options to be used. With this in mind, a discussion took place relating to each of the issues introduced by Halfdan.

In terms of reference frame, it was agreed that ESEAS should use a separate approach to the estimation of global parameters (ie satellite orbits, satellite clocks and Earth rotation) and local parameters (station coordinates, tropospheric ZTD, etc). In relation to this, it was agreed that for a stable reference frame the global parameters should be solved for by using a set of stations that are consistent over time and that, in order to align with ITRF2000, these stations should be at sites that are co-located (ie with more than three space geodetic techniques). With this in mind, it was noted that there are already initiatives in place, through SOPAC and IGS TIGA-PP, to re-compute global parameters for the last 10 years or so, using such an approach. Whether ESEAS needs to carry out its’ own re-computation of global parameters will depend on whether the global parameters produced by SOPAC or IGS TIGA-PP meet the requirements of ESEAS, both in terms of the network of stations used but also in terms of whether the station motion models used by these organisations are consistent with the station motion models we would like to use for ESEAS.

In terms of analysis strategy, it was agreed that data from each ESEAS Observing Site should be processed using the same global parameters but with three or more different softwares/strategies. In relation to this, the partners present at the meeting agreed to contribute to task T2.3 using the following softwares and processing strategies (ie PPP or DD):

- NMA, Norway, GIPSY/OASIS II, PPP

- ROA, Spain, GIPSY/OASIS II, PPP

- UNOTT, UK, Bernese 4.2, PPP

- SRC, Poland, Bernese 4.2, DD

- UPC, Spain, Bernese 4.2, DD

- GCM, Turkey, GAMIT, DD

In terms of physical models, it was agreed that station motion models for solid Earth tides and ocean tide loading (and possibly atmospheric loading, depending on the outcome of Action Item WP2/WG2-1-4) should be applied during the GPS data analysis. It was also agreed that all of the partners listed above, except for UNOTT, UK using Bernese and PPP, should attempt to resolve the integer ambiguities and should produce both a ‘float’ and a ‘fixed’ solution. The partners that will be involved in the processing were asked to provide input to the document being prepared for task T2.1 in terms of the options available within the software they propose to use. Specifically:

- What information relating to the satellites is included in the software?

- Is the computation of station displacement due to solid Earth tides consistent with IERS conventions?

- Is the computation of station displacement due to ocean tide loading consistent with IERS conventions?

- Is it possible to include coefficients for ocean tide loading based on the FES99 model?

- What is recommended/default in terms of the components (ie only vertical or horizontal and vertical) and constituents (eg M2 and S2 or all) to be used for ocean tide loading?

- Is the computation of station displacement due to atmospheric loading consistent with IERS conventions?

- What strategy is applied when attempting to resolve the integer ambiguities and produce a ‘fixed’ solution?

- What is recommended/default in terms of the hydrostatic delay, the wet delay and their respective mapping functions for the troposphere (eg hydrostatic delay modelled using Saastamoinen with the Neill dry mapping function and wet delay estimated as an unknown, with the Neill wet mapping function)?

In terms of the inputs relating to the different softwares (including weighting, iterations and cut-off elevations) the partners that will be involved in task T2.3 were asked to provide information on these, as input to the document being prepared for task T2.1.

One issue that was not discussed at the meeting was the modelling of antenna phase centre variations, and whether we should use relative values (ie with Dorne Margolin choke ring antennas assumed to have zero variations with elevation angle) or absolute values. The answer to this may partly depend on what is being used in the re-computation of the global parameters by SOPAC and IGS TIGA-PP. This should be reported as part of Action Item WP2/WG2-1-5 and then considered as part of Action Item WP2/WG2-1-6.

A discussion on the treatment of dual-CGPS stations during the GPS data analysis then took place, as there has always been a suggestion from EUREF that in the EPN, they avoid mixing baselines of a few kilometres with baselines of several hundred kilometres when computing a network solution based on the use of double-differenced data. It was agreed that the partners proposing to use Bernese or GAMIT should contact the software suppliers for clarification and advice on the implications of including baselines of a few kilometres with baselines of several hundred kilometres when carrying out a network solution based on double-differenced data. It was also agreed that Simon Williams would contact SCIGN for advice on the same matter. It was noted that this was not an issue when computing a precise point positioning solution using undifferenced data as each station is treated separately.

It was agreed that if the advice was to avoid mixing baselines, then one of the ways in which the problem could be avoided would be to separate the dual-CGPS stations so that they were part of two sub-networks. Considering the agreement that each ESEAS Observing Site should be processed using the same global parameters but with three or more different softwares/strategies and the discussion on the analysis strategies to be used by the partners, the contributions to task T2.3 were revised to reflect the agreement that the partners would use the following softwares (ie GIPSY/OASIS II, Bernese 4.2 or GAMIT) and processing strategies (ie PPP or DD) on the following sets of stations:

- NMA, Norway, GIPSY/OASIS II, PPP, on all ESEAS Observing Sites (including both dual-CGPS stations, where applicable)

- ROA, Spain, GIPSY/OASIS II, PPP, on all ESEAS Observing Sites (including both dual-CGPS stations, where applicable)

- UNOTT, UK, Bernese 4.2, PPP, on all ESEAS Observing Sites (including both dual-CGPS stations, where applicable)

- SRC, Poland, Bernese 4.2, DD, on all ESEAS Observing Sites (except for one of the dual-CGPS stations, if necessary and where applicable)

- UPC, Spain, Bernese 4.2, DD, on at least all of the excepted dual-CGPS stations not processed by SRC, Poland

- GCM, Turkey, GAMIT, DD, on all ESEAS Observing Sites (except for one of the dual-CGPS stations, if necessary and where applicable)

Here it was suggested that the GS/OSO, Sweden response to the ESEAS 2^nd call for participation could be used as a complement to the GCM, Turkey, if they would be prepared to use GAMIT, at least on the excepted dual-CGPS stations.

The next part of the discussion related to the flow of data in and out of the GPS data analysis.

It was agreed that the ESEAS GPS data centre should play a more active role than simply archiving the data. In this respect, it was suggested that the ESEAS GPS data centre should run tecq on the incoming RINEX data in order to produce an s file for each station, from which certain values could be extracted for quality control purposes. It was agreed that Norman Teferle would investigate what quality parameters can be extracted from tecq (eg MP1 and MP2) that would be useful as part of a ‘data time series file’, for monitoring the quality of the data in the archive prior to processing.

It was agreed that each ESEAS GPS data analysis centre should produce daily, loosely constrained solutions in SINEX format and that these should be archived. It was also agreed that each ESEAS GPS data analysis centre should produce a set of ITRF2000 station coordinates for each day and that these should be used as the basic input to a ‘solution time series file’ for each station.

A discussion then took place regarding the format and content of the time series files. It was agreed that the time series files would be created in XML format (as an international standard) and that a file would consist of a header, followed by a single line of information for each day. Hans-Peter Plag showed an example of an ASCII solution time series file that is being used by NMA, Norway, in which the header included nominal station coordinates and the daily information included date, epoch, daily coordinate estimates, as differences from the nominal station coordinates (in E, N and V), standard errors of the daily coordinate estimates (in E, N and V), mean ZTD and ZWD estimates for the day, standard error of the ZD estimates, and some quality parameters that were extracted from GIPSY/OASIS II relating to the a-posteriori unit variance and number of observations accepted/rejected. It was agreed that each of the partners involved in task T2.3 would investigate what quality parameters can be extracted from their respective softwares that would be useful as part of a solution time series file.

An overview of tasks T2.2 and T2.5 was given by the respective task leaders, Simon Williams (see the pdf file of the presentation for the full details) and Norman Teferle (see the pdf file of the presentation for the full details).

Simon explained that the main aim was to use the outputs from the GPS data analyses, ie the solution time series file, in order to produce vertical (and horizontal) station velocities in the ITRF2000 along with an estimation of uncertainties in the rates. Simon went on to explain that the production of velocities includes the detection and removal of outliers, the detection, estimation and removal of coordinate offsets and the detection, estimation and removal of periodic signals. Furthermore, it was noted that the estimation of uncertainties in the rates should take account of both white noise and coloured noise. Norman then presented a list of issues that needed to be considered for the two tasks, namely; solution time series files (including quality parameters), outliers, coordinate offsets, periodic signals, common mode biases, loading and methods/softwares to be used.

The overview given by Simon and Norman was followed by a series of brief presentations from each of the partners that will be involved in task T2.5, on their experience in time series analysis and their proposed contribution to the task (full details of the presentations given by each of the partners can be found in the respective pdf files).

- Simon Williams gave a presentation on behalf of NERC-POL, UK (see pdf file)

- Halfdan Kierulf and Hans-Peter Plag gave a presentation on behalf of NMA, Norway (see pdf file)

- Norman Teferle gave a presentation on behalf of UNOTT, UK (see pdf file)

Simon Williams and Norman Teferle stated that they intended to prepare the deliverable for task T2.2 as a document that would provide a detailed description of the ESEAS coordinate time series analysis strategy. With this in mind, a discussion took place relating to each of the issues introduced by Norman.

In terms of analysis strategy, it was agreed that solution time series files from each ESEAS GPS data analysis centre should be processed with three different softwares/strategies. In relation to this, the partners present at the meeting agreed to contribute to task T2.5 using the following softwares and strategies:

- NMA, Norway will use in-house software with a deterministic approach to signal analysis based on EOFs to model linear and non-linear components.

- NERC-POL, UK will use in-house software with a stochastic approach to noise analysis based on an MLE, with the combined modelling of coordinate offsets and periodic signals, and compute uncertainties based on power law noise.

- UNOTT, UK will use in-house software with a stochastic approach to noise analysis, with the separate modelling of coordinate offsets and periodic signals, and the computation of uncertainties based on empirical estimates of white noise and coloured noise.

In all three cases, it was stated that tools would be available to enable the analysis of individual coordinate time series, the analysis of common mode biases and the analysis of correlations between time series. It was noted that the latter may include comparisons of solutions from different GPS data analysis centres, the comparison of coordinate time series with time series of physical parameters not incorporated in the GPS data analysis (eg atmospheric loading, hydrological loading), and the comparison of coordinate time series with time series of quality parameters (ie those in the solution time series files).

For the common mode bias, it was noted that the nature of the bias depends on the processing strategy applied during the GPS data analysis (ie PPP or DD). In the case of the PPP we need to know the global common mode bias whereas in the case of the DD we need to know the regional common mode bias. For the global common mode bias it was noted that this could be computed by analysing the coordinate time series of a number of global sites. At this stage, there was a discussion on whether these global sites should include or exclude the ones that were used to compute the global parameters. It was agreed that NMA, Norway would carry out a series of tests based on an EOF analysis of (a) the coordinate time series for global sites used in a computation of global parameters and (b) the coordinate time series of global sites positioned using PPP and these global parameters, in order to decide which approach should be used to estimate the global common mode bias as part of the ESEAS coordinate time series analysis strategy.

Depending on the outcome of Action Item WP2/WG2-1-5, the ESEAS GPS data analysis centres would either be using global parameters re-computed by SOPAC or IGS TIGA-PP, or global parameters computed by ESEAS. In either case, it was noted that approach (a) does not restrict the selection of global sites to use in the estimation of the global parameters whereas approach (b) does, as certain global sites would have to be left out from the estimation of the global parameters in order to be included in the estimation of the global common mode bias. Furthermore, it was noted that if the global parameters re-computed by SOPAC and IGS TIGA-PP can fulfil the requirements of ESEAS (see Action Item WP2/WG2-1-5) then approach (a) would not require ESEAS GPS data analyses using PPP to include any global sites but approach (b) would.

An overview of task T2.6 was given by the task leader, Simon Williams (see the pdf file of the presentation for the full details).

In terms of task T2.6, Simon reported that NERC-POL, UK planned to make absolute gravity measurements at least once per year at Aberdeen and Lerwick, and possibly twice per year at Newlyn, over the next three years. Ryszard Zdunek reported that SRC, Poland will use FGI to make absolute gravity measurements twice over the next three years and Coskun Demir reported that GCM, Turkey are seeking a collaboration with a Turkish organisation that has recently purchased an FG5.

Simon explained the difficulties in finding a good site for absolute gravity measurements at tide gauges, and that these difficulties would most likely force NERC-POL to have to move the absg station at Aberdeen. He also went on to explain that due to the fact that the absg stations are several kilometres from the tide gauge at the three UK sites, precise levelling links had not been established yet between the absg station and the TGBM.

In terms of task T2.8, Simon explained that the aim is for absg stations with reliable, repeated measurements to be used to obtain estimates of vertical land movements at these sites that are independent of many of the biases that affect CGPS coordinate time series, and which should enable an indication of how absolute our emerging CGPS station vertical velocity estimates are.

An overview of task T2.7 was given by the task leader, Coskun Demir (see the pdf file of the presentation for the full details).

Coskun outlined that the aim of task T2.7 was to assemble information on any precise levelling that has been carried out within a few kilometres of the tide gauge. Coskun then showed that the GCM, Turkey has a software package that ultimately, can be used to analyse the precise levelling information and assess whether any deformations may have occurred within a few kilometres of the tide gauge.

Station Comments:

=========================== FOLLOWING PSMSL DOCUMENTATION ADDED 11-JUN-91 :

Newlyn 170/161 RLR(1964) is 11.7m below TGBM SW4676 2855

=========================== BENCHMARK INFORMATION UPDATE ADDED 12-JUL-93 :

Newlyn 170/161 RLR(1964) is 11.7m below TGBM SW4676 2855

For Newlyn, the EOSS station handbook contains the following information on benchmarks:

Tide gauge benchmark description OSBM bolt insode hut adjacent to

Frequency of levelling

Tide gauge zero relative to EUVN

Tide gauge zero relative to TGZ = admiralty chart datum = 3.05m below ODN

national levelling datum

Tide gauge zero relative to TGZ = 7.8012m below TGBM

tide gauge benchmark

Auxiliary benchmark description Aux 1 - Flush bracket 1565 on wall S pier NW