Technical issues and recommendations related to the installation of continuous GPS stations at tide gauges


Michael Bevis, Wolfgang Scherer  and Mark Merrifield



1. Introduction


Geodesists and oceanographers have begun to implement the long considered step of building continuously operating GPS  (i.e. CGPS) stations at or near dozens of  tide gauges around the world in order to position them, and their local histories of sea level, in a well-defined and global spatial reference system (Carter et al., 1989; Carter, 1994; Neilan et al., 1998).  This ambitious and exacting undertaking has a variety of motivations. Some scientific or technical agendas require exact knowledge only of the spatial position of the tide gauge. Other agendas, such as deriving changes in ‘absolute’ sea level over ~ 100 year time scales, involve estimating the secular vertical velocity of the tide gauge. Positioning a tide gauge with a vertical accuracy (not just precision) of 1 cm with respect to a well-defined global datum is a difficult task, but given at least several years of CGPS data it is certainly a viable one. Determining the vertical velocity of a tide gauge or the underlying bedrock with an accuracy better than 1 mm/yr, even with a decade of CGPS observations, remains a very challenging problem.


Given the inherently global nature of these tide gauge problems, and the unprecedented measurement accuracies being sought, it is essential that very high standards of instrumentation, installation, operational procedure and data analysis are achieved.  There is clearly a need to develop at least minimal standards to guide the individual agencies and research groups working on this problem.  On the other hand, scientists and engineers engaged in activities as diverse as satellite calibration, hydrography, global geodynamics and climatology often have differing requirements,  and sometimes this will quite logically lead to somewhat divergent and at least mildly incompatible approaches. When we add to this diversity of intent the often wildly different constraints imposed by local geology, topography, sky view, site stability and history, major differences in the density and quality of the regional geodetic infrastructure, and significantly differing levels of available funding, any narrowly drawn and rigid set of guidelines will soon prove naïve or impractical.


Nevertheless there are some general rules which should be followed in almost every case. And while individual teams must choose between alternative approaches based on their particular application, their resources and the local context, it may still prove useful - to new practitioners in particular -  to have access  to a set of documents that describe some of the tradeoffs that are commonly encountered. This website does not constitute a standards document in the usual sense, but it is hoped that it will prove useful to groups building CGPS stations at tide gauges, and help individual stations contribute as much as possible to the overall international effort. We hope to collect case histories from around the world, and add new material to this website as it becomes available. (Please notify Michael Bevis at should you wish to contribute).


2. Absolute versus relative sea level


Oceanographers use tide gauges to record the history of sea level with respect to the underlying solid earth (which is usually land, or its nearshore underwater extension, but can also be ice).  The position of the sea surface with respect to a global datum such as the ITRF ellipsoid is known as absolute sea level (ASL), and temporal variability of this sea surface height is known as absolute sea level change. The vertical motion of the land or sea floor, or more generally of the lithosphere, with respect to the ITRF ellipsoid is called absolute vertical crustal motion. The difference between these two motions, which constitutes the motion of the sea surface with respect to the adjacent land, or underlying seafloor, is referred to as relative sea level (RSL) change.


Even when tidal and atmospheric fluctuations are removed from consideration, sea level can change by hundreds of millimeters over time scales of order one year or less.  Since the typical rate of vertical crustal motion is of order one millimeter per year, for short periods of time the rates of absolute and relative sea level rise or fall are almost identical over most of the worlds oceans and coastlines. This may not be true in areas of great tectonic activity (e.g. seismically active subduction zones or active volcanoes) or in areas with serious subsidence problems associated with the pumping of fluids. But over most of the globe we expect short-term changes in absolute sea level to be nearly identical to changes in relative sea level. However, when we consider secular rates of sea level rise over time periods of order 50 – 100 years, then typical rates of absolute sea level rise (1 – 2 mm/yr) are of the same order of magnitude as typical rates of absolute vertical crustal motion over much of the world, and so absolute and relative sea level changes are quite distinct. Long term rates of inundation in most coastal zones depend roughly equally on the shifting levels of both the land and the sea. (Shifting ice levels, past and present, also play an important role). In tectonically active areas, or areas with major subsidence problems, relative sea level rise over long periods of time may be dominated by motion of the land surface.


In principle, satellite altimeters measure absolute sea level, tide gauges record relative sea level, and if CGPS (or another space geodetic technique) is used to measure vertical crustal velocity at some point on a coastline, this geodetic measurement can be used to transform relative sea level rise into absolute sea level rise, or vice versa.  In practice these transformations must be implemented with very great care. It is very desirable that all three measurements are made at a limited number of fiducial sites around the world so that rate closure or consistency can be checked, thereby validating the individual measurement systems.


3. Tide gauges and their local vertical reference systems


Tide gauges record sea level with respect to a local vertical datum realized using the gauge itself, a network of  benchmarks, and (usually) a tide staff (IOC, 2000). The benchmarks are survey markers which are tied to an internal or external reference mark or ‘contact point’ on the tide gauge by precise spirit leveling.  While some groups use the term tide gauge benchmark (TGBM) to refer only to the most important (and usually closest stable) benchmark, we use the term tide gauge benchmarks (TGBMs) to refer to all leveling survey markers associated with the tide gauge, and refer to the standard or canonical benchmark as the primary tide gauge benchmark (PTGBM). Ideally some TGBMs are located close to the tide gauge while others are located one or several kilometers away. Ideally there should be at least ten TGBMs, though in many cases there are less than this. It is potentially problematic when there are less than five TGBMs, particularly when none of them are set into bedrock.


The purpose of the network of TGBMs is twofold: (i) to refer sea level to ‘stable ground’ and (ii) to register sea level to a reference system external to the tide gauge itself,  which might be destroyed in a storm or harbor accident, or be replaced by another tide gauge to accommodate (for example) development of the harbor. Very commonly tide gauges are built on wharves or piers which are known to be sinking into the ground. It is not desirable to reference sea level to unstable structures nor to unstable ground such as a nearby land surfaces (e.g. parking lots or wharfside loading areas) that overlie engineering fill.  Unstable ground is a common problem for tide gauges which are constructed in areas devoid of surface bedrock.  This is why the network of TGBMs typically has an aperture of one to three kilometers or more. These TGBMs are tied not just to the tide gauge (and the tide staff when there is one) but they are tied by repeated leveling to each other, ideally at intervals of one or two years (but unfortunately less frequently in many locations).  By examining the relative height history of all the TGBMs in the network, it is usually possible to identify a subset of TGBMs that have not undergone any significant relative subsidence, and can be viewed as ‘stable,’ meaning that they are not in motion with respect to the underlying solid crust of the earth. Typically the stable TGBM closest to the tide gauge is chosen to be the PTGBM.  Occasionally this designation changes over the history of the network as a benchmark once thought to be stable is later realized to be moving. Sea level recorded by the tide gauge proper is adjusted to account for relative motion between the gauge (and staff when present) and the PTGBM.  It is very important for geodesists to realize that tide gauge records which extend back 50 years or more usually were not produced by the present day tide gauge. Rather there has usually been a series of tide gauges in the general area, all of which were tied to the TGBM network.  It is this network which provides long term continuity to the sea level time series.  Accordingly, positioning the TGBM network is as important, and arguably even more important, than positioning the tide gauge of the present day.


Most tide gauges around the world were originally constructed to track tidal cycles and to assist with navigation into and out of harbors, and not to monitor sea level changes per se. Many of these ‘tide gauges of opportunity’ were not originally equipped with a network of TGBMs, and in some cases they have not been retrofitted with an adequate leveling infrastructure even after being adopted for long-term sea level monitoring. Even when we restrict our attention to those tide gauges with a TGBM network and a well established program of leveling, in many cases the integrity of the local height system realized using the TGBM network has been slowly degrading.  Leveling surveys are being repeated less frequently and the technical quality of these surveys is degrading as agencies retreat from expensive and time consuming (but exquisitely accurate) first order leveling standards. TGBMs that are destroyed, typically by human activity, are not always being replaced. And groups setting up new tide gauges sometimes fail to build networks of sufficient aperture to ensure that several of these marks are located in truly stable ground (often because they cannot or do not wish to level over long distances). Thus the GPS geodesist must consider not only the geometry of the network of TGBMs, and the history of precise leveling, but also assess the future viability of the leveling program when choosing where to install a new CGPS station. In some parts of the world it would be very dangerous to install a CGPS station several kilometers or more away from a tide gauge and simply assume that some agency will repeatedly tie the CGPS station to the tide gauge and the TGBMs by means of first order leveling for many years to come.  If the CGPS station is far removed from the tide gauge and is not connected to the leveling network, then it will not be possible to constrain relative motion of the tide gauge and the CGPS station, thereby rendering the latter almost useless.


4. The siting of CGPS stations at tide gauges


The number of CGPS stations deployed at each tide gauge


While financial constraints usually limit geodesists to deploying a single geodetic GPS station at or near a given tide gauge, it would obviously be attractive to deploy two or more. Occasionally a group deploying a CGPS station at a tide gauge site is lucky enough to gain access to data from another pre-existing CGPS station just tens of kilometers away. With very great luck this pre-existing CGPS station would be part of the global tracking network of the International GPS Service (IGS). Unfortunately these happy accidents rarely occur except in regions with highly developed GPS infrastructure, like Europe, Japan and the Western USA. In the next section we discuss issues related to deployment of a single CGPS station. Later we will briefly consider the deployment of multiple CGPS stations.


Choosing the exact location of the CGPS station


To obtain the best possible vertical accuracies from a CGPS station it is crucial that, among other things, the GPS antenna has a clear view of the sky in all directions for elevation angles above 15 degrees, and, ideally, a clear view down to 10 degrees elevation. This is an exacting requirement.  One cannot build a high performance CGPS station at the base of a cliff, nor by the side of a large building, since these obstacles will block the signals emanating from GPS satellites located in large areas of the sky. Unfortunately a significant fraction of tide gauges and TGBMs are located in places with poor sky view. There are other requirements too, such as the security of the GPS equipment, the availability of electrical power and telephone lines, etc.  It is the authors’ experience that there are rarely many and occasionally no really good locations for a CGPS station at or near a targeted tide gauge. In these cases arguments about ‘the best’ siting strategy often turn out to be moot. Nevertheless there has been quite a lot of debate about whether or not it is better in principle to build a CGPS station immediately adjacent to a tide gauge, or close to one or more TGBMs which are known to be in ‘stable’ ground (assuming that there will be only one CGPS station).


The advantage of locating the CGPS antenna immediately adjacent to the tide gauge (assuming it has good sky view) is that with reasonable care one can almost ensure that there will be no subsequent relative motion of the antenna and the gauge. Accordingly the vertical offset between antenna and gauge can be determined at the time of construction, and, in the event that frequent leveling does not take place in later years, then the vertical offset can be assumed to be constant.  Another advantage of collocating the CGPS antenna and the gauge is that it makes the leveling tie extremely easy to perform, even for the most reluctant practitioners of precise leveling. Leveling error tends to increase with distance traversed, and so collocation promotes high measurement accuracy. (This being especially important when first order leveling procedures are not followed). Of course,  if leveling crews do visit the tide gauge regularly, they could and should tie the CGPS antenna to the entire TGBM network.


The main disadvantage of locating the CGPS station immediately adjacent to the tide gauge is that in most cases the ground near the gauge is unstable, and so vertical velocity determined at this point probably manifests instability of the antenna monument or the underlying shallow subsurface material as well as (or instead of)  vertical motion of the rigid lithosphere. Thus vertical motion of the CGPS station would not manifest vertical crustal motion of the whole region surrounding the gauge (say the entire island on which a gauge is situated). Consider the case of a tide gauge station with a 100 year history, involving several distinct tide gauge instruments all of which were tied to a network of TGBMs, with the latest tide gauge in the series residing at the end of a visibly deformed and rickety pier. Should it be clear that all the TGBMs located more than 1 km inland from the gauge have proven stable over long periods of time, then it would clearly be preferable to build the CGPS stations near one or more of the stable TGBMs located closest to the gauge, rather than at the gauge itself.  (Keep in mind, however, that this may not be possible. Perhaps the entire area determined to be stable ground is heavily forested and has terrible sky view).


Now consider a counter example.  Suppose that a CGPS station is being retrofitted to a tide gauge purely to support calibration of a satellite altimeter, and that determining the absolute position (not the vertical motion) of the gauge  is the highest priority. Suppose also that this tide gauge is only 10 years old, and so cannot be used to estimate long-term relative sea level changes. Suppose further that a review of the leveling program suggests that perhaps none of the TGBMs are truly stable (perhaps by virtue of being too close to an unstable coastline), and that the leveling program is now in decline. Then it would be more reasonable in these circumstances to build the CGPS station immediately adjacent to the tide gauge, and for the GPS crew to perform a quick  but accurate tie between gauge and CGPS antenna during their installation visit.


To the extent that there is a regular program of leveling surveys at a tide gauge station, one could argue that it should not matter greatly where the CGPS station is located provided that the CGPS station is tied to both the tide gauge and the network of TGBMs.  After all, if the gauge has always been tied to the stable TGBMs by precise leveling over fairly long distances, then what is wrong with tying either the gauge or the TGBMs  to the CGPS stations by a similar program of leveling? Indeed, under some circumstances it might be desirable to install the CGPS station neither near the gauge nor the stable subset of TGBMs. This might happen if neither the tide gauge nor the TGBMs were set in rock, but solid rock outcrop does exist either close to the gauge or the stable TGBMs. It might seem strange that TGBMs were not installed in bedrock if it is present, but this does happen – sometimes inexplicably, sometimes because there were TGBMs in rock but these marks have been destroyed, and sometimes because the bedrock has only recently been uncovered.


Installing multiple CGPS stations


Several groups , including the European Seal-Level Observing System (EOSS) consortium (Plag et al., 2000), advocate building a pair of CGPS stations in association with each tide gauge, one directly adjacent or physically coupled to the tide gauge, and another further away on stable ground. The goal is to measure both vertical motion of the tide gauge (which may consist of a mixture of crustal motion, local subsurface instability and monument instability) and crustal vertical motion (which is usually all that affects a well constructed CGPS station built on ‘stable ground’, at least over significant periods of time). Certainly the multiple CGPS station approach is attractive if it is affordable – particularly if the GPS data are analyzed using kinematic as well as static (long baseline) GPS processing techniques, which would allow recognition of local relative motions developing over periods as short as one hour (e.g. thermoelastic motions). These short-period signals will not be detected in the standard batch processing techniques that will be used to position one or both CGPS stations in the global reference frame.  But knowledge of their existence could prove useful.


The dual (or multiple) CGPS station approach will rarely if ever eliminate the need for repeated leveling surveys.  Just as many TGBMs originally thought (or hoped) to be set in stable ground later turn out to be unstable,  a CGPS station set into apparently stable ground, but not in bedrock, may also turn out to be unstable.  Indeed, in the absence of bedrock, the past history of the TGBM leveling program is probably the best basis for choosing a stable site. Unless a remote CGPS station can be demonstrated not to be unstable in a local sense, it is dangerous to assume that any measured vertical motion manifests only crustal motion, and is highly representative of the motion of the surrounding area.


Probably the largest problem facing implementers of the multiple CGPS station approach is that it may not be possible to find two good CGPS sites close to a tide gauge.  It is often difficult enough finding one!  If a second ‘stable’ site cannot be found within several kilometers of the tide gauge, but has to be located at a considerable distance from the tide gauge and the existing TGBMs, then unless this station is set into solid rock, it will be necessary to greatly extend the original leveling network, or build a second leveling net around the remote CGPS station, in order to assess the local stability of the remote station.


Some groups are considering installation of one dual frequency geodetic GPS system, and a second cheaper (survey grade) single frequency GPS system. The geodetic GPS station would be positioned over long baselines relative to other CGPS stations used to realize the global reference frame, whereas the single frequency system would be used only to determine a local GPS tie and so address local relative motions.  This raises the question of where to put the better receiver.  We strongly advocate putting the geodetic grade receiver at the same site one would choose if only one CGPS station were going to be installed.  In particular, it makes little sense to us to install a geodetic receiver at a place with poor sky view and a single frequency receiver at a place with good sky view.  Indeed, we are convinced that if one of the sites has really poor sky view, it  is probably a waste of resources to install any CGPS receiver there at all.


The best scenario, if two CGPS stations are used, is that they both use dual frequency receivers. Then both stations could be tied directly to the global reference frame, which would then reduce the impact of either receiver going down for significant periods of time.  We expect the cost differential between survey grade single frequency systems and full-blown dual frequency systems to collapse during the next few years (we write this in the year 2000). Even prior to this reduction in the differential cost of single and dual frequency  receivers, it is worth reflecting on the fact that after a CGPS site is built and maintained for several years, receiver hardware costs often comprise only half or less of the total expenditures associated with the overall effort. Furthermore, some of us believe that single frequency receivers are destined to disappear from the survey market altogether, which makes their use in a long-term monitoring effort even less attractive.


On the problems associated with subsidence caused by fluid withdrawal.


An analysis of apparent surface tilting deduced by comparing repeated measurements made over 70,000 km of survey lines in the U.S. national leveling network demonstrates that a large fraction of the largest tilts are associated with  lines passing through cities on the Atlantic Coastal Plain and the Gulf Coast (Reilinger et al, 1984). There can be little doubt that all or nearly all of these signals manifest near-surface subsidence associated with water withdrawal (Poland and Davis, 1969; Chi and Reilinger, 1984).  Similar subsidence patterns have been observed in Europe (Wood, 1990; Bingley et al., 1999).  When surface subsidence associated with the collapse of an aquifer has spatial scales measured in tens of kilometers, small networks of TGBMs (not set in low permeability bedrock) may not be able to distinguish between these broad subsidence patterns and crustal motion organized at the scale of the lithosphere. Just because a group of TGBMs have no relative vertical motion does not rigorously imply they are set in stable ground, if ‘stable’ is meant to imply no motion with respect to the underlying elastic lithosphere. Vertical surface velocities above aquifers often vary rapidly in space (e.g. Reilinger, 1980), and so a measurement made at one point may not adequately characterize surface motion even in the immediately surrounding area.


Those CGPS efforts intended to relate absolute and relative sea level histories for tide gauges established many decades ago are particularly threatened by broad patterns of subsidence associated with water withdrawal. This is because the rate of water withdrawal and therefore of surface subsidence has almost certainly changed over the years. It is probably unreasonable to assume that the vertical velocity of a CGPS station measured in the next decade could be used to correct tide gauge records collected over the last 50 or 100 years.  Nor can the leveling program be used to resolve this problem if the leveling network is entirely contained within the area undergoing subsidence. Given that many of our oldest tide gauges are located in large coastal cities, it is essential to address this problem whenever measuring vertical velocity is a major focus of the project. Subsurface geology should be considered with care.  Urban tide gauges founded on hard bedrock should be favored, whenever possible, over those established over soft sediments (especially aquifers) when selecting tide gauges for retrofitting with CGPS.



5. Monumentation and instrumentation


When establishing a new CGPS station at a tide gauge, we strongly recommend following the standards of the International GPS Service (IGS) as closely as possible.  These guidelines and standards are available at the IGS website ( These guidelines include the use of dual frequency code and phase measuring GPS receivers., and  IGS-standard antennas such as the Dorne-Margolin choke ring antenna.  Our specific recommendations, below, are provided to emphasize those considerations that we feel are especially important in the context of positioning tide gauges, or those which are frequently neglected.


The antenna monument


It is necessary to employ a monument that holds the GPS antenna above any nearby objects so that it has a clear view of the sky at all elevation angles above 15 degrees  (or even closer to the horizon, if possible) in all directions.  This monument needs to be as rigid and as stable as possible. In the ideal case it is anchored in solid rock, but this will not often be possible.  It is very important that the GPS antenna is not mounted above a flat metal plate, as this is known to cause positioning errors (Elosegui et al., 1995). We have seen several cases of groups installing antennas above metal plates or curved metal shells, despite the fact that this is known to produce systematic errors in the vertical coordinate of the station.  It is now widely recognized that the characteristics of a GPS antenna, including such phenomenon as its phase center variation, may be affected by objects, such as the antenna monument, that lie within the near field of the antenna.  In our own field installations we attempt to minimize the effects of the antenna monument by making it as narrow as possible, subject to its need to support the antenna even during major storms. We also favor a monument which is axisymmetric in its shape (with a vertical axis of symmetry passing though the center of the antenna), so that it is less likely to influence the antenna in ways which vary systematically with azimuth.



We recommend using an antenna monument which is cylindrical, or nearly so, with a flat circular mounting surface which matches the diameter of the mounting surface of the antenna. If an exact match is not possible we recommend that the support’s mounting

surface be smaller than the base of the antenna rather than larger.  We use monuments that can be taken apart to provide access to the bolt that holds the antenna to the mounting surface. Two typical designs are illustrated schematically in Figure 1.  When a flange is used to remove the top of the monument for access to the antenna bolt, we insist that this flange is located at least 30 cm below the mounting surface, and we keep the flange as narrow as possible. Furthermore the holes in the flange are arranged such that the monument fits together only in one way.  Recently we have taken to constructing our antenna supports out of variable radius aluminum poles, since these are remarkably rigid while remaining very light. This allows us to ship the monuments to the site rather than fabricating a monument locally.


We emphasize that it is nearly always a bad idea to install a GPS antenna on a roof.  This is because (i) many buildings are unstable, and are capable of changing in shape as well as moving with respect to the deep subsurface, and (ii) it is usually difficult to level to the antenna if it is located high above the ground.  An exception can be made to this prohibition when (i) it is absolutely necessary, (ii) the building is known to be stable (e.g. it is a small, steel-reinforced block house built directly onto a coherent concrete pier), and (iii) it is possible to level to the antenna despite its position (most likely using the ‘inverted leveling’ configuration). Installing a GPS antenna on the roof of a building should always be a last and rather desperate resort - never undertaken for mere convenience.


Since CGPS positioning of tide gauges is an exacting and long-term undertaking, it is essential to build monuments that will maintain their mechanical integrity over many years of exposure to the elements. Some care must be taken when bringing dissimilar metals in contact as this may induce electrochemical reactions that lead to high rates of corrosion. When different metals are in use, it is often a good idea to use special coatings, thin Mylar washers, etc., to suppress electrochemical effects. These measures should not compromise the rigidity or stability of the monument, however.  Another consideration in choosing a metal is its general resistance to corrosion. For example, aluminum monuments tend to resist corrosion when exposed to sun, rain, and even warm salt-laden air (when was the last time you saw a rusty airplane?). But aluminum corrodes extremely rapidly in many underground environments. Accordingly, when we bolt an aluminum antenna monument onto bedrock or a concrete substrate, using steel bolts (at least 1 foot in length) for anchors, we prevent direct contact between these bolts and the base of the aluminum monument by using plastic sleeves and washers.  Stainless steel monuments, though expensive and heavy, resist corrosion both underground and above ground. But care must be taken in coupling an aluminum antenna to a steel mounting surface.


Vertical ties between the GPS antenna, the tide gauge and the TGBMs


When the GPS antenna is tied by precise leveling to the tide gauge, the TGBMs, and the tide staff if one exists, it is the level of the Antenna Reference Point (ARP) that should be used to formally characterize the height of the antenna (Fig. 1).  The ARP is typically defined to lie in the plane of the antenna’s basal mounting surface so that the normal to this surface passing through the ARP passes also through the geometrical center of the antenna. This typically means the ARP lies in the center of the threaded hole used to hold the 5/8” antenna bolt.  Once the antenna is coupled to the mounting surface of the monument, the ARP is not visible, and it is not accessible to the contact point of a leveling rod. Accordingly, the rod must be inverted and put into contact with the lower surface of the ground plane of the antenna (Fig. 1), ideally as closely as possible to the center of the antenna.  Then the known vertical offset between the lower surface of the ground plane and the ARP must be applied as a correction to the leveling measurements, so that the ARP is the actual vertical reference  point recorded in the leveling report.  This correction should be explained and recorded in the leveling records.  Note that it is essential that the leveling crews have a copy of the CGPS station site log which includes a description of the geometry of the antenna, including the position of the ARP and its vertical offset relative to the lower surface of the ground plane.  It is also necessary that the leveling crew records the serial number of the antenna, if it is visible, to ensure that the site log is still current.   The leveling crew should measure more that one point on the ground plane to ensure that it is really level.  If it is not, this fact should be recorded and measurements should be made at two points either side of, but equidistant from the ARP, so that the height of the ARP can be inferred.  Since antenna monuments are always constructed so that the mounted antenna is level to within better than 1 degree, should it subsequently be found to be tilted then the station is unstable.




Most GPS antennas need to be protected from the elements, birds, insects, etc., by radomes – which are antenna covers that are transparent at GPS frequencies.  It is now well established that raydomes affect the GPS signal recorded by the receiver, and these influences need to be modelled if they are not to produce biases in vertical coordinate estimates (and delay estimates).  Accordingly it is essential to record the type of raydome in use at a given station in as much detail as possible, and to document any changes made over time. Radomes come in different shapes, e.g. hemispherical and conical, and can be made using different materials. Merely describing a dome’s shape and thickness is inadequate – the physical material must be known too, as it controls the refractivity of the dome.  It is certainly a good idea to state any manufacturer’s part number, though we are worried that some vendors may be changing the way in which domes are fabricated without changing the part or serial number associated with their product. We prefer hemispherical domes made of low-refractivity, uniform-density, air-filled foams, but recognize that these are not suitable in some settings due concerns such as ice build-up  and resting birds.


Met packs


A CGPS station can be used to infer the total water vapor content of the overlying atmosphere (e.g. Duan et al., 1996) provided that very accurate surface pressure measurements, and fairly accurate surface temperature measurements are available at or near the GPS antenna.  Many modern GPS receivers can log specially designed meteorological sensor packages, known as ‘met packs’, so that this surface meteorological data can be downloaded along with the GPS data.  Adding a met pack to a CGPS station enables it for ‘GPS meteorology’.  These water vapor measurements are useful both in the analysis of weather systems (Businger et al., 1996) and climate cycles or climate change (Foster et al., 2000).  Since sea level change is strongly related to air-sea interaction, using the CGPS station to measure water vapor, as well as the position (and possible motion) of the tide gauge, promotes a natural scientific synergy.  Building multiple-use CGPS stations is also advantageous in that it makes it more likely that leading GPS processing groups, especially those participating in the IGS, will incorporate these CGPS stations into their geodetic analyses.  CGPS stations equipped with suitable met packs, in particular, are highly prized by many GPS processing groups.


When a met pack is installed at a CGPS station, or when a previously established barometer from a nearby site it used instead, it is necessary to know roughly (say to within one meter) any vertical offset between the barometer and the GPS antenna. This should be brought to the attention of the leveling crews, if necessary.


Data flow


Many CGPS stations around the world are equipped with data communication channels, such as modems/phone lines, or, increasingly, internet links, which allow the GPS data to be downloaded on a daily basis (or even more often) into a regional or global data archive.  This arrangement allows the IGS to include these stations in the operational data processing its analysis centers carry out for producing precise orbital solutions (ephemerides) for the GPS satellites.  See the IGS website at for details. CGPS stations of this kind are referred to as ‘on-line’. In some parts of the world suitable data communication links are unavailable, or prohibitively expensive, and data are sent to an associated processing center on zip disks, optical disks or other inexpensive media, often by mail.  Sometimes these data are available to the processing group only after delays of weeks or months have been incurred. 


It is important to realize that on-line stations reporting rapidly to an IGS data center are far more likely to be routinely analyzed by IGS analysis centers than are off-line stations.  Therefore, groups wishing to take advantage of the world-class IGS analysis centers should build on-line stations whenever this is practical, and seek formal incorporation of these stations into the IGS global tracking network.


Data Quality Monitoring


The group in charge of a specific CGPS station at a tide gauge should monitor the signal-to-noise ratio (SNR) information output by the receiver to ensure that system performance does not degrade over time due to receiver hardware problems, exposure-related problems with antenna cables or connectors, new or worsening sources of radio frequency interference at the site, etc. Ideally the GPS data should be converted into RINEX format using a translator that preserves the SNR parameters reported by the GPS receiver. Typically these SNRs should be plotted against satellite elevation angles in order to establish the SNR signature of a given station, allowing it to be compared with similar stations, and monitored over time.


The field team should demand continuous feedback from any internal or external data processing groups using the GPS observations, so as to see if the processing software is detecting unusual numbers of cycles slips or rejecting unusual amounts of the data during geodetic analysis. Over the long term it is important to determine if the station’s position time series has unusual levels of scatter, etc., and, if so, to resolve if this manifests problems in the data being collected, or in the processing strategy being employed.



6. Geodetic data processing


Accurate vertical positioning in a well-defined global reference frame remains a state-of-the-art problem in space geodesy.  The good news is that processing techniques and reference frames continue to improve, and data collected now can be reprocessed in the future, which virtually ensures that we will achieve our measurement goals sooner or later. While this document focuses largely on the practical aspects of siting CGPS stations at tide gauges, and on the details of installation, we fully recognize that processing these data in an optimal manner poses another set of major challenges. The International Association of Geodesy (IAG), the International Association for the Physical Sciences of the Ocean (IAPSO) are working with the International GPS Service (IGS) to develop optimal processing strategies for the tide gauge problem. Discussion of these strategies is beyond the scope of this document.  Please feel free to contact the first author, or the Central Bureau of the IGS via its website, for further details concerning the data processing plans of the IGS.


We would like to conclude by noting that our immediate task is to build the CGPS stations.  This first step is perhaps the most crucial in that a seriously flawed installation will limit a station’s geodetic performance forever after, no matter how clever our processing strategies become as they evolve over the years. Site selection and installation constitute the very foundation of a positioning effort, and subsequent geodetic analysis of the data rests and builds upon this foundation. While the requirements and desiderata associated with CGPS positioning of tide gauges, and their networks of TGBMs, seem so numerous that it will rarely if ever be possible to resolve them all to our total satisfaction, we do need to build the best CGPS stations that we possibly can, and to maintain each station and its local vertical ties over long periods of time.  The instrumental challenges are at least as severe as those associated with data analysis.



Acknowledgements.  We thank Philip Woodworth, Dana Caccamise, David Phillips, Thomas Herring, C.K Shum, John Beavan, Herb Dragert and Hans Peter Plag for their comments and suggestions.  This paper and website was produced under the auspices of the Joint Working Group of IAG (SC8), IAPSO (CMSLT), IGS, PSMSL and GLOSS.




Businger, S., et al., 1996, The promise of GPS in atmospheric monitoring, Bull. Amer. Met. Soc.,  77, 5-18.

Bingley, R., V. Ashkenazi, N. Penna, S. Booth, R. Ellison and A.Morigi, 1999, Monitoring Changes in Regional Ground Level, Using High Precision GPS, Environment Agency R&D Technical Report W210, ISBN 1-85705-121-1.

Carter, W.E., et al., 1989, Geodetic fixing of tide gauge bench marks, Woods Hole Oceanographic Institute Technical Report, WHOI-89-31, 44 pp.

Carter , W.E. (Ed.), 1994, Report of the Surrey Workshop of the IAPS Tide gauge Bench Mark Fixing Committee held 13-15 December 1993 at the Institute of Oceanographic Sciences Deacon Laboratory, Wormley, UK. NOAA Technical Report NOSOES0006.  81pp.

Chi, C. and R. Reilinger, 1984, Geodetic evidence for subsidence due to groundwater withdrawal in many parts of the U.S., J. Hydrology, 67, 155 – 182.

Duan, J. , et al., 1996, GPS Meteorology: Direct Estimation of the Absolute Value of Precipitable Water, J. Appl. Met., 35, 830-838.

Elosegui et al., 1995, Geodesy using the Global Positioning System: The effects of signal scattering  on estimates of site position, J. Geophys.Res., 100, 9921—9934.

IOC, 2000, Manual on sea-level measurement and interpretation. Volume 3 - Reappraisals and recommendations as of the year 2000. Intergovernmental Oceanographic Commission Manuals and Guides No. 14. IOC, Paris. (ed. P.L.Woodworth)

Plag, H.-P., P. Axe, P. Knudsen, B. Richter, and J. Verstraeten (Eds), 2000,  European Sea-Level Observing System (EOSS): Status and future developments. COST Report.

Foster, J., et al., 2000,  El Niño, Water Vapor and the Global Positioning System, Geophys. Res. Letts., 27, 2697 - 2700.

Wood, R.M., 1990, London: not waving but drowning,Terra Nova, 2 (3), 284-291.

Neilan, R., Van Scoy, P.A. and Woodworth, P.L. (eds), 1998, Proceedings of the workshop on methods for monitoring sea level:  GPS and tide gauge benchmark monitoring and GPS altimeter calibration. Workshop organised by the IGS and PSMSL, Jet Propulsion Laboratory, 17-18 March 1997.

Poland, J., and  G. Davis, 1969, Land subsidence due to withdrawal of fluids, Rev. Eng. Geol., 2, 187 – 269.

Reilinger, R., M. Bevis and G. Jurkowski, 1984, Tilt from releveling: An overview of the U.S. data base, Tectonophys., 107, 315 – 330.

Reilinger, R.E., 1980.  Elevation changes near the San Gabriel fault, Southern California, Geophys. Res. Lett., 7, 1017-1019.



Last Revised:  12 October 2000