24 March 1995
The last few weeks have witnessed a lot of Internet traffic within the TOPEX/POSEIDON (T/P project) which started with Gary Mitchum's request for support for WOCE 'fast centre' tide gauge activities in his battles with his funding agency. Thanks to Bill Carter, the subject somehow metamorphosed into a general one of, in short, 'Are global tide gauge networks required now we've got an accurate space based system (i.e. T/P)?'
Many tide gauge people who are not involved with T/P will also be interested in this discussion, so we have included some of the more coherent email contributions below. It is a good example of the sort of on-going debate which should be taking place within GLOSS, WOCE, IPCC etc.
As I'm the editor of this Bulletin, I'll also inject my 10 pence worth, although some of the people below also make the same points. It seems to me, as primarily a tide gauge person, that even if we can assume that T/P quality missions will be ongoing (another debating point), then a well distributed global gauge network will be required to calibrate the altimeters. No-one wants just single calibration sites like Harvest, Lampedusa or Venice. Fans of T/P (and I am one of them) will recall that, while it has an apparent superb resolution of about 2-3 cm, its bias means it has a relative accuracy compared to 'the truth', or to other missions, of about 15 cm, and that this value is uncertain to about 2-3 cm. Any one gauge will have better stability than that, and the global ensemble of gauges should, when corrected for land movements with GPS (and thanks to the ITRF again), provide extremely precise datum control to the altimeters.
Scientifically, even if ongoing T/P quality missions could be guaranteed, then the arguments that they are 'better' than a tide gauge network:
- gauges don't measure global sea level, only coastal level, and
- gauges measure only relative sea level, not absolute level
can be turned to advantage by the coastal, rather than deep sea, community as follows:
- people live at the coast not in the deep ocean (usually) and nations will always want to know what is happening on their own beaches not only what the T/P project tells them, and
- people only care about relative levels flooding their houses.
Finally, what interests me most, there is the historical record to consider. There is no historical altimetry worth talking about, but there is about 100 years or more of tide gauge data. The replacement of one technology by another, especially technologies which are supposed to monitor long term changes, should be accompanied by a long period of parallel operation to understand the relative systematic errors. The tide gauge community itself has made this mistake so often in many countries, replacing stilling wells with bubblers then with acoustic systems, that the best datasets of time series are often the ones that stuck with the older technologies. Let's not make an even bigger mistake now with altimetry, let's make all the measurements we can.
In his March 6 message, Dave Enfield asked, "How many people out there (raise your hands) think that altimetry is so good and permanent that we don't need in-situ sea level height any more?" For the particular problem of global sea level rise, my hand is up. I would also like to address the critical role of the terrestrial reference frame in the Topex/Poseidon results.
Based on recent e-mail messages prompted by Gary Mitchum's funding dilemma, I can only conclude that some people have been so busy defending their research budgets that they have missed the revolution. We are not *attempting* to develop a global absolute sea level monitoring system -- we achieved that goal when T/P became fully operational in 1992. It is here, whether the C&GC panel wants it or not.
Perhaps this achievement has been missed by some because it is based on a new paradigm. A global, absolute sea level monitoring system requires only two major components:
(1) A high-quality satellite altimeter, with a reliable internal calibration.
(2) Accurate satellite orbits derived from a tracking network connected to, and working within, a stable terrestrial reference frame, i.e. the International Earth Rotation Service (IERS) International Terrestrial Reference Frame (ITRF).
The ITRF was developed by the international geodetic community over the past decade primarily from traditional geodetic funding sources, but new funding from sources such as the NOAA Climate and Global Change Program (C&GC) have been essential to achieving the global coverage required for satellite altimetry. For example, the C&GC funds have partially supported the development and operation of three VLBI stations that are the heart of the southern hemisphere ITRF network.
Current budget questions should not be posed in terms of the "potential" geodetic contribution to the monitoring of global sea level, but rather to the ongoing geodetic contribution to the existing, functioning system. The ITRF is an essential component of T/P, and it must be maintained and continued over the life of current and future altimeter missions, or the sea level monitoring capability will be lost.
Continued operation of tide gauges may be justified for certain local applications, but the pretense that they are essential to current or future monitoring of global absolute sea level should be dropped.
In response to Bill Carter's comment that "Continued operation of tide gauges may be justified for certain local applications, but the pretense that they are essential to current or future monitoring of global absolute sea level should be dropped", I would like to make a couple of points. First, in the absence of a decade long altimeter time series necessary for detecting global climate change signals, satellite altimetry cannot yet be viewed as a replacement for the tide gauges for the global mean sea level problem, especially given the likely gap between TOPEX/POSEIDON and a future mission of similar accuracy. Second, although the tide gauges cannot compete with the spatial resolution provided by the T/P data, they are important for monitoring the calibration of the T/P altimeter, which is the largest source of uncertainty for measurements of global mean sea level. Although the altimeter calibration is measured internally, this only checks the altimeter electronics, and not the other components of the measurement such as the ionosphere correction, troposphere correction, etc. The internal calibration (computed at Wallops) currently gives a drift of 2.7 mm/year. Comparisons of the T/P data to the WOCE tide gauge data give a value of 3 +/- 2 mm/year, which strengthens the confidence in the T/P mean sea level measurements considerably. As the error is reduced with a longer time series, this result will become quite important. While I agree that the future is certainly going to be driven more by satellite altimetry than by the tide gauges, tide gauges will continue to play an important role in the mean sea level problem for the forseeable future.
The hidden problem is that continued cuts in budgets are turning researcher against researcher - everyone suggesting that others take the cuts. The geodetic part of the NOAA program has been cut from $675K two years ago to $275K this year. If the global tracking network dies, satellite altimetry will quickly degrade and we will be waiting another 50 years to see how sea level is doing using tide gauges.
I would like to respond to Bill Carter's comment on "the use of tide gauge data for future monitoring of global absolute sea level variation should not be a requirement" and his other opinions on the subject.
I absolutely agree with him that the establishment and continued maintenance (operation) of an accurate ITRF by NASA, NOAA, NRL, and others are one of the critical requirements for the potential use of satellite radar altimetry systems for the monitoring of global sea level variations. I agree that VLBI stations are part of the important maintenance of the ITRF, but would like to add that SLR is one technique which has been demonstrated to be able to measure motions of geocenter variations with an accuracy of a few mm (including the vertical component). Even with the present funding environment, it would be extremely unwise to discontinue vital support for the maintainence of the ITRF (which includes VLBI and SLR) for research topics which have decadal or longer time scales (e.g., sea level).
I disagree that global tide gauge should not be treated as one of the requirement for global sea level variation monitoring. It is my opinion that T/P altimeter system has demonstrated an unprecedented accuracy to have the potential to monitor mean sea level variation with an accuracy of 1 mm/yr. However, I do not agree that we have (1) enough data span and (2) adequate knowledge of errors (drift, seasonal signals, geocenter, media correction, etc.), to claim that T/P has provided a definitive observation that the mean sea level is raising at the 3 mm/yr level during 1992-1994. As Bruce Douglas pointed out earlier that the critical ionosphere instrument correction error has not yet been fully understood. Steve Nerem and others also pointed out, in general, that the error characteristics of an altimeter system have not been completely understood at the mm/yr level. It makes the continuation of some of the more accurate tide gauges (G. Mitchum's UH/WOCE and P. Woodworth's PSMSL gauges) even more important for validation purposes (with altimeter, as Steve Nerem already pointed out), and equally important, to continue the 100-year+ tide gauge sea level time series.
I urge those who has an relevant interest in sea level to strongly support continuing both the tide gauge data correction and the operation of the SLR/VLBI stations to maintain the ITRF.
As a one who helped plan the Topex/Poseidon mission at the systems level, (I was the T/P project scientist before Lee Fu), as one who first proposed, along with Byron Tapley and George Born, that altimetry can be used for monitoring global mean sea level, and as an experimentalist who has operated many different instruments at sea, Bill Carter's blanket statements make me very uneasy. Yes, there has been a revolution in geodetic accuracy over the past two decades, it was this development that led me to altimetry in the first place. Can altimetry replace tide gauges? I am not yet convinced that they can.
Cosider the following. Seasat carried a very well calibrated altimeter, yet the altimeter appeared to have an error of 50 - 70 cm. Topex/Poseidon carries two very well calibrated altimeters, yet they differ by around 16 cm. These figures give system accuracy. The few mm per year discussed by Carter and others is the precision of an instrument. Before altimeters can supplant tide gauges, Carter and others must demonstrate that system accuracy, not instrument precision, approaches 1 cm.
First. What can go wrong? Many things. For example, can we accurately calculate center of mass of a spacecraft as the fuel on the craft is used? In principle yes. In practice there can be problems. Suppose for the sake of argument that the bladder inside the fuel tank goes up rather than down due to an error in documenting the system. The error could produce a few mm per year change in apparent sea level, but the error would be hard or perhaps impossible to detect because it is not possible to go up and check the configuration of the satellite. Now consider the hundred other subsystems that also influence altimeter system accuracy. Are all these parts accurate and stable? How do we know? Internal calibration of the altimeter does not contribute to a detection of these errors.
Second. What does altimetry measure? The few mm/year applies only to the global volume of the sea. Regional and basin-wide variability of sea level can be due to oceanic processes or to errors in the altimetric system. The dominant error may be the geographically correlated error in the satellite's ephemeris. How will these errors be detected?
In my experience, system errors dominate instrument error, and the only way to check system accuracy in altimetry is through a global network of tide gauges. We need to carefully separate system accuracy, system precision, altimeter instrument accuracy, and altimeter instrument precision and discuss each separately. A "A high-quality satellite altimeter, with a reliable internal calibration" is necessary but not sufficient to detect changes in sea level.
The same applies to a need for an "Accurate satellite orbits derived from a tracking network connected to, and working within, a stable terrestrial reference frame, i.e. the International Earth Rotation Service (IERS) International Terrestrial Reference Frame (ITRF)". How do we determine the accuracy of an orbit? I have listened for 15 years to debates by those who calculate orbits. I have heard many statements that it is very difficult to determine independently orbit accuracy. At best, we can claim that various ephemerides are consistent, but do we have an independent system able to determine satellite position with an accuracy of 1 cm? T/P comes close with three almost independent tracking systems providing data to four almost independent teams who compute orbits. I say "almost" because the techniques and teams have some common elements, including models for Earth's gravity field, computer subroutines, big G, etc.
Remember, 1 cm accuracy in global mean sea level, requires a measurement system with an accuracy of one part in a billion. Few systems in physics have this accuracy.
Regards, Bob Stewart
The problem that I have with the responses that I have seen so far to my message concerning the value of tide gauge observations for global sea level monitoring is that people ascribe too much capability to the tide gauges. If we have 25 to 50 years to wait we may extract significant additional information on global sea level change from the tide gauge network - otherwise they provide local information of relative sea level but little more than the long term average information on sea level already available from the past 100 years or so of operations. A regional network of ide gauges such as that in the west Pacific may very well be corrupted by vertical crustal motions, from glacial rebound, core mantle pressure, tectonics, etc., at significantly more than a few tenths of a mm per year. Unless the stations are carefully monitored for vertical crustal motion, for example by continuous GPS observations tied to the ITRF, it would be very dangerous to favor tide gauge comparisons over the internal calibration. Even if there is a break of a a few years in satellite altimetry, the tide gauge data collected during that interval would not provide a reliable figure for any change in global sea level during that period. Calibration of the satellite altimeter by a few special stations, such as the platform,or by use of tranponders is a much more reliable and feasible approach. For those familiar with polar motion and the history of monitoring of polar motion I would recall the great outcry to save the International Latitude Stations and to continue their operation for at least another 18.6 years to compare the new VLBI-SLR-LLR results over a full nutation period. Had we done so we would have wasted at least $50M by now. The ILS observations were simply not accurate enough to make comparison with the new results useful. I argue that the same is true with the traditional tide gauge observations.
I've been reading these messages with considerable interest and agree with much that has been said on all sides. But Bill Carter's thoughts raise very awkward questions. Scientifically and technically I believe he is completely correct - the global tide gauge network is inadequate to define global mean sealevel with very much accuracy (but it should be possible to define that accuracy and to my knowledge it has never been done). The problem is not scientific or technical - it is almost purely political: is there enough political support to sustain T/P quality altimetric measurements indefinitely? If the answer is "yes", then we need only a supporting calibration network on the surface, and should be designing it to minimize bias and drift questions. But if the answer is "no", or "probably no", then the question arises as to the scientific return of the tide gauges on their own.
It is probably also true that a gap of a "few years" in altimetry is tolerable. But how long a gap is that really? Suppose it were 30 years? (I think a 30 year gap is unlikely, but would not want to bet much of my own money that it will be "only" five years.)
I think this problem is one of probabilities and insurance, and not strictly of science, hence the great difficulty. The existing tide gauge network is a form of insurance against the collapse of the altimetric system, but it is not a substitute for it (although if one's castle burns down, adequate money to live in a house trailer might be very attractive.)
What is this insurance worth? My instinct is that it is worth quite a lot (suppose one said 1% of the cost of the altimetric missions, would that be regarded as a reasonable insurance premium?)
The discussion revolving about Bill Carter's comments have had two themes: value to the T/P mission and value to the detection of long-term change. Bill is a geodesist and his perspective is naturally the latter. However, the users of altimeter data comprise a much wider community for which long-term change is only one interest and not the most important one. When you move the discussion from the geodetic arena to the oceanographic and climate arenas, and from the secular time scale to the seasonal-to-interannual time scale, many of Bill's arguments are irrelevant or work the other way. I won't belabor the point by listing them, most of us are perfectly aware of this. The point is, seasonal-to-interannual climate prediction is a priority less vulnerable to politics and the vagaries of funding decisions. This is because its consequences hit J.Q. Public where it hurts every few years and it's easier to demonstrate the connections of those consequences to the ocean-atmosphere interactions we're trying to understand than it is to prove that the secular changes we see are due to global warming or something else. As a result, OGP's priorities are in the area of seasonal-to-interanual climate. If the ITRF is critical to making altimetry a tool for short-term climate, then it's important to OGP under the current funding priorities. But don't try to tell me the tide gauge data aren't important to the short term climate problem, because I know that altimetry will not answer all the questions about sea surface height in that arena and I know I can't depend on keeping good altimetry on line continuously over the next 15-20 years (minimum) for which we will need it. Moreover, Carl Wunsch is absolutely correct about the backup issue and if his arguments are extended to short-term climate they become even more convincing. And besides, what kind of $$ are we talking about? The costs of maintaining the in-situ sea level data are piddling in comparison to the altimetry, and if Bill Carter could get all the funding he wants to do space geogesy, he would outspend Gary by orders of magnitude. In performance-to-cost terms, Gary's operation is a good deal. You can probably finance Gary just by keeping the funding for space geodesy comensurate with our national funding priorites (ITRF/Poseidon, etc.).
David B. Enfield
In all the correspondence I've seen so far, there has been no mention of another capability of tide gauges which altimetry will never be able to replace, that is the monitoring of short period signals. We are reaching the point with TOPEX/POSEIDON where much of the "noise" is actually unresolved oceanography, and we need to understand this noise if we are to know the limitations on altimetry. This is also relevant for tides. A tide gauge can produce high accuracy measurements of tidal components much more quickly than altimetry, and recent tests suggest that the tides are only stationary to about 2 cm. If we are to understand the factors which modulate the tides, we will need the "quick" capabilities of tide gauges.
It is in the nature of the systems that they perform complementary roles: Altimetry provides relatively high spatial resolution at low temporal resolution, tide gauges provide high temporal resolution at low spatial resolution. It seems to me to be foolish to throw away the tide gauges until we understand the full impact of the short period signals they are uniquely placed to measure.