The IGS is working with the oceanographic community to begin implementation of the long-considered step of adding continuous GPS (CGPS) stations to tens of tide gauges around the world so as to reference sealevel at this set of points to the ITRF. A technical committee formed under the auspices of the IGS and the PSMSL was established at the 1997 JPL Workshop in order to write a set of standards for groups constructing these CGPS stations. During its discussions the committee has come to realize that the oceanographic community is pursuing several different agendas, and that these efforts have distinct accuracy requirements and are impacted in different ways by various geotechnical considerations and local environmental factors. There are two main agendas:
(i) The 'centimeter' agenda. The idea is to establish the absolute vertical position of the tide gauge and its benchmarks (and ultimately the local sealevel history) with an accuracy of 2 - 3 cm, for the purpose of calibrating and/or validating modern satellite altimeter measurements (e.g. TOPEX). The oceanographers engaged in this activity typically place little importance on providing tectonic corrections for historical sea level data obtained at their calibration stations.
(ii) The 'millimeter' agenda. Tide gauge measurements reflect vertical motions of both the land (to which the gauge is attached) and the sea surface. The intra-annual, annual and even decadal fluctuations recorded by a tide gauge are usually totally dominated by absolute sealevel changes. But when averaged over much longer periods of time (~100 years) the rates of vertical motion of the land and the sea surface are often similar in magnitude (~ 1 mm/yr). In order to obtain an absolute sealevel history from a long historical record of relative sealevel change, it is necessary to estimate the rate of vertical motion of the land or structure supporting the tide gauge. The goal is to use CGPS to measure the vertical velocities of tide gauge stations with an accuracy of better than 1 mm/year within ten years, and with considerably better velocity accuracy over longer periods of time. This more ambitious agenda was the main focus of the well-known Carter reports.
The second agenda is much more challenging and will lead to quite severe requirements for the selection of 'suitable' tide gauge sites, the CGPS installations and data analysis. Tide gauges that can be instrumented with CGPS for the purposes of the centimeter agenda may often be unsuitable for pursuing the millimeter agenda. As such the technical committee is now working on separate sets of recommendations for these distinct 'end case' applications. A key realization is that these recommendations need to be considered as tide gauges are being selected, and not just during the implementation stage.
The Impact of Geodetic Requirements on Site Selection
Some requirements are common to any positioning agenda and would apply to any tide gauge. For example, one should always 'tie', by precise leveling, the GPS antenna to the system of tide gauge benchmarks (TGBMs) as well as to a reference point on the tide gauge itself. This is especially important when one is working at a tide gauge with a long history and this history is a major focus of the effort. This is because over many decades the tide gauge is likely to have been modified, moved and even replaced on several and perhaps many occasions. It is the TGBMs and the associated program of precise leveling measurements that provide the temporal continuity of the vertical datum at such stations. In this sense one could argue that when the focus is on a very long sealevel time series, the system of TGBMs is even more important than the specific tide gauge that is operating today. Nevertheless, the need to tie the GPS antenna to the TGBMs as well as to the tide gauge is clear no matter what application we are pursuing or where we are working. Not all requirements are this straightforward, however.
The technical requirements for geodetic positioning of tide gauges need to be considered during the selection of gauges for GPS augmentation, and not just subsequently. Only a small fraction of all available tide gauges will be retrofitted with a CGPS station. It would be short-sighted to base the selection of these tide gauges purely on oceanographic criteria, without regard to the relative geodetic suitability of the candidate tide gauges. The ease with which a given tide gauge can be positioned geodetically depends both on the positioning accuracy required and the local environmental conditions. Consider the standard issue of sky view, for example. It is unlikely that one could ever obtain subcentimeter vertical accuracy by making GPS measurements at a tide gauge located at the base of a 200 meter cliff. This raises the question of how far the CGPS station can be removed from the tide gauge. In many cases the existence of walls or buildings besides the tide gauge will require the GPS antenna to be moved at least several meters from the gauge itself. This will rarely matter as long as the GPS antenna mount is connected to the same structure as the gauge, making significant relative motion of the tide gauge and GPS antenna impossible. But could it ever be acceptable to install a CGPS station hundreds of meters or even kilometers away from the tide gauge in order to optimize conditions for the GPS measurements (e.g. an improved sky view) or for other purposes such as security of the GPS equipment?
There are a number of factors that must be addressed before one can answer this question. What positioning agenda are we pursuing? How often would the GPS antenna be referenced to the tide gauge using precise leveling? What is the nature of the ground beneath and between the tide gauge and the GPS antenna? I do not have space to discuss here all the relevant issues in a comprehensive manner, but the following points may illustrate the nature of the problems and trade-offs involved.
Some tide gauges are attached to seawalls or piers that were constructed on sediments or, worse still, engineering fill. The well-known tide gauge at Valparaiso (Chile) is an example of the latter situation. These structures typically subside as the underlying material compacts. Oceanographers have long been aware that tide gauges may be subject to purely local vertical motions of this kind, and this is one of the reasons why they have referenced their tide gauges to TGBMs. Some oceanographers have concluded that a CGPS station serves a similar role to a TGBM, and so there is no particular need to place the CGPS station very close to the tide gauge. However, a CGPS station is being positioned in a global reference frame on a daily basis. If a somewhat remote CGPS station is moving by an unknown amount relative to its tide gauge, then the absolute position of the tide gauge is being determined only as frequently as the levelling tie is being made. This may be just once a year. This represents a massive dilution of the positioning power of CGPS.
The negative impact of losing the temporal continuity of CGPS measurements is further illustrated by the following example. Some tide gauges are tied to large standing structures such as piers that may be undergoing thermoelastic displacements with amplitudes of millimeters and sometimes even centimeters. Suppose one such tide gauge is being tied by precise leveling to a CGPS station built on a rock outcrop 1 km away. If the leveling survey happens roughly annually and always takes place around the same time of year, even large annual thermoelastic motions of the gauge might never be observed at all!
If a tide gauge and a non-colocated CGPS station are not constructed on a coherent outcrop of solid rock, the possibility of subsurface displacements changing the relative level of the tide gauge and the GPS antenna is present even when near-surface 'engineering' instability of a pier or a harbor wall can be ruled out. An analysis of all releveling data obtained in the USA prior to 1980 showed that most of the largest vertical displacements occurred in and near cities built on sediments such as those underlying the coastal planes of the Atlantic and Gulf Coasts. Most of these signals manifest differential subsidence associated with water withdrawal. Many tide gauges are based in this kind of setting.
One also needs to keep in mind that in many parts of the world precise leveling surveys do not take place at reasonable intervals (except on paper). In this case there will be no strong constraints on possible relative motions of the tide gauge and the GPS antenna.
If one is pursuing the centimeter agenda it will usually be acceptable to offset the GPS station from the tide gauge by distances as great as 1 km, provided that frequent first-order leveling ties are performed. In some settings the leveling ties should be performed several times during the first year to ensure that annual cycles of deformation affecting any large structures supporting the gauge (or the GPS antenna) are not going undetected or being misunderstood due to aliasing. Additionally these surveys should be performed throughout the lifetime of the observation program to ensure that slower (but not necessarily steady) secular movements will be adequately characterized. The larger the separation of the GPS station and the tide gauge, the more difficult and expensive the leveling program will be, and the more likely it will not be executed properly throughout the lifetime of the program. The inconvenience of frequent leveling should be contemplated very carefully before one abandons the idea of putting the GPS station very close to the tide gauge.
It will rarely be acceptable to separate the tide gauge and the GPS antenna by more than a few meters when one is pursuing the millimeter agenda. The only exception to this would be when both the tide gauge and the GPS station were being installed in a single massive outcrop of competent and locally rigid rock. Under these circumstances one could separate the tide gauge and the GPS antenna by a few tens of meters.
Suppose the oceanographic agenda in a given region could be satisfied by adding a CGPS station either to tide gauge A or tide gauge B. Suppose gauge A is built on a solid rock outcrop in an area with minimal human impact, and it will be possible to build a monument that couples the GPS antenna to the rock substrate just a few meters from the tide gauge. Suppose that tide gauge B is located on a harbor wall at the base of a large building. The nearest place with a good sky view is over a kilometer away and this place, the tide gauge, and the road between them overlie a sedimentary basin, which includes a heavily pumped aquifer. It would be unreasonable to select B over A even if the location of B made it slightly preferable for purely oceanographic reasons.
The selection of tide gauges for GPS augmentation must involve some sort of compromise between oceanographic and geodetic requirements. The higher the positioning accuracy required, the more weight must be placed on the environmental and geodetic suitability of the site. We must be very careful when we determine whether we are pursuing the centimeter or the millimeter agenda at a given site. If the geodesists hear one or two oceanographers say that their primary interest is the centimeter agenda then it is naturally tempting to settle on that goal, since it makes it much more likely that the proposed site can be considered suitable, and because the standards for installation will be more lax and therefore easier to achieve. However, in many cases, stations built to this standard will not be suitable for pursuing the millimeter agenda. So if the oceanographers change their minds about the precision levels they seek to attain, then much of early work on CGPS positioning of tide gauges will subsequently be viewed as substandard. This prospect is particularly worrying given the history of upwards creeping accuracy requirements in almost every application of GPS geodesy.
The Writing of Standards for CGPS Installations at Tide Gauges
The above discussion should make clear that the writing of standards for CGPS installations at tide gauges is more involved than most members of our technical committee first realized. Nevertheless we are working on a document which is based both on multidisciplinary discussions, and on experience gained during past installations.
The University of Hawaii is one of several groups working in this area. Our effort is a collaboration between the Pacific GPS Facility led by the author, and by the UH Sea Level Center led by Mark Merrifield, another member of the technical committee. Our first project, began last year, was installation of a CGPS station at the Honolulu Tide Gauge. This dataset is freely available on a daily basis. We are presently engaged in selecting another six tide gauges for GPS augmentation. Early site selection studies indicate that almost every station will be a special case! We will be documenting each project as it occurs, and will make this information available on the web. We hope that other groups will be doing this too. Hopefully we will be able to refer to these websites in the IGS/PSMSL standards document, so that specific cases can be made to serve as illustrations of the general requirements associated with CGPS positioning of tide gauges.
Data Processing and the Role of the IGS
By the beginning of the year 2000 we expect that 15 - 25 tide gauges around the world will have been retrofitted with CGPS stations. These data will be made available to the IGS via one or more data centers. The IGS needs to consider now what mechanisms need to be established to ensure that these data are processed in an optimum manner. We must keep in mind that, initially at least, most of these stations will not be reporting within a few days of real-time. Thus the processing streams associated with IGS orbit production are not relevant to this agenda. Additionally, because many tide gauges are built on relatively unstable structures, these new CGPS stations may not be of much interest to geodetic groups already engaged in ITRF densification efforts if their primary motivation is crustal motion research. Who is going to be responsible for this new processing effort? And who is going to pay for it?
Getting Involved
Anyone wishing to join the IGS/PSMSL technical committee on CGPS positioning of tide gagues, and/or participate in our email discussions, should send an email message to bevis@soest.hawaii.edu.