Condensed from a paper published in the Marine Technology Society Journal, Vol. 25, No. 4, 1992.

Sea Level As an Indicator of Climate and Global Change

Bruce B. Parker

ABSTRACT

This paper discusses the difficulties in predicting future global sea level rise. It also examines the problems involved in determining a reliable global sea level trend over the past century from historical water level records and, specifically, whether the apparent recent rise can be attributed to global warming due to the accumulation of greenhouse gases. The paper also discusses the importance of interannual-to-decadal sea level variation and its role in helping to understand climate variations such as El Nino Southern Oscillation and the accompanying global effects via teleconnection.

INTRODUCTION

Progress in the study of climate and global change depends heavily on the creative use of long historical geophysical data sets. Some of the most reliable and useful of available long data sets are the sea level records taken for a century or more at numerous tide gauges around the world.

But sea level is not only a parameter with a valuable historical record, it is a parameter of direct concern to those who fear the possible consequences of an earth warmed as a result of the accumulation of greenhouse gases. If indeed the worst projections of sea level rise came to pass, the result would be flooded coasts, eroded beaches, loss of wetlands, threats to aquifers, and other causes of significant economic loss. The larger scientific question is: What use can be made of sea level data in order to better understand climate and global change? However, the part of the question of most concern to many will always be: Is sea level rising, and, if so, will it rise fast enough to pose a threat to mankind?

The concern over sea level rise has led to a number of recent reviews and assessments of the scientific evidence, many sponsored by national and international agencies. The present review is not meant to be comprehensive. What is intended is to provide some idea of the complexity of the problems involved in predicting future sea level rise, and even in determining recent sea level rise and its causes.

Important questions to ask when looking a sea level rise research are the following. Can a reliable global sea level trend over the past century be determined from the available historical water level records? Is the apparent sea level rise seen in the historical data really due to the greenhouse effect? Or, could sea level simply still be rising since the last ice age, or could this apparent sea level rise actually be due to the land subsiding at most tide gauge locations? Can an increase in the rate of sea level rise be detected in these data, which would indicate that there is a greenhouse warming effect? Are these results of any use in projecting future global sea level rise, and if not what can we use?

DETERMINING RECENT SEA LEVEL RISE FROM HISTORICAL DATA

The Historical Sea Level Records

The historical sea level data records are from several hundred tide stations around the world, some of which have been in operation for up to a century or more. These stations were not originally installed to study global change, as this was certainly not a concern in the mid-1800s or early 1900s. They were installed primarily for navigational purposes, and the data were used to make tide predictions. Automated measurements were taken hourly using a float inside a protective stilling well. These measurements were related to permanent benchmarks on land and were quite accurate in comparison to other geophysical data. Comparison observations were made at a tide staff in order to tie the float gauge to the benchmarks, which were leveled directly to the tide staff (that is, the relative elevation of the benchmarks and the zero point on the tide staff were determined by surveying techniques). This process, including the leveling, was very important in some countries, such as the United States, because marine boundaries were determined by the mean low water datum (separation between state and federal jurisdiction) and mean high water datum (separation between state and private ownership). When offshore oil was discovered, these marine boundaries became even more critical. All this was to the benefit of those reaserchers who would use the data from these gauges for sea level rise studies. The most important requirement for these data is datum continuity, that is, the maintenance of a direct relationship of the measurement to the benchmarks over the years. There can be a variety of errors in the water level measurement, but they are generally random and tend to average out over a monthly or yearly average. But when trying to determine trends on the order of 1 or 2 mm/yr, it is vital that accurate datum continuity be maintained.

The large numbers of tide stations in operation in the United States and some other countries have also been of great benefit. Although the low-frequency variations of sea level have large spatial scales and one would assume that only a few gauges would be required, the tide station redundancy proved to be quite beneficial from a quality control point of view. Much depended on the tide observer, who visually made the daily simultaneous observation at the tide staff. These observations, when compared with the observations made by the float gauge, allowed the float gauge record to be related to the benchmarks (which were leveled to the staff not the gauge). The tide observer was a human intervention that was a potential source of error, but nearby tide gauges provided another comparison for quality control.

Effects on Water Level and Relative Sea Level

A clarification of terms is necessary. Water level is the measurement made by a tide gauge of the distance of the water’s surface above some reference point, or datum. The device that carries out the measurement has been called a tide gauge primarily because at most locations the astronomical tide is the largest part of the water level variation and the resulting data were usually obtained in order to make tide predictions. (The word tide has been used by some in the same fashion as we use the phrase water level, with the distinction then made between the astronomical tide and what was referred to as the meteorological tide.) The phrase sea level or mean sea level (MSL) is used to mean that the water level observations have been averaged over some period (usually at least a month), so that the shorter period variations have been averaged out and in fact can be viewed as oscillating about this mean sea level. Sea level is only a mean for a particular time period, and it varies over longer time periods, on a monthly, interannual, and longer basis. This variation in sea level is measured relative to the land (to which the benchmarks are permanently attached). If the land sinks, it will appear that sea level is rising, and likewise if the land rises it will look like sea level is falling. Thus we refer to this as relative sea level.

It is important to distinguish between (1) the various oceanographic and meteorological causes of water level variation in general, at all time and space scales; (2) the causes of long-term (eustatic) sea level rise; and (3) the possible effects seen in a finite sea level record that may look like sea level rise.

The water level measured at a tide gauge is affected by a number of oceanographic and meteorological phenomena, including the astronomical tide, changes in atmospheric pressure, wind, river discharge, ocean circulation, changes in water density, and added water volume due to the melting of ice. The astronomical tide is caused by the gravitational effects of the moon and the sun, creating very long waves in the ocean, which propagate over the continental shelf and into shallow bays, where amplifications and nonlinear distortions take place. Increased atmospheric pressure decreases water level and vice versa. In shallow areas the onshore-offshore wind component can directly push water toward the shore (wind setup) or away from it (setdown). The usually more dominant effect is caused further offshore by the longshore wind component, which can raise or lower the water level because the Coriolis force causes transport to the right of the wind direction. Along certain coasts wind also causes upwelling, which affects the temperature and density of the water column. Steric sea level changes, that is, water level changes due to density changes, are caused by temperature changes in the water column and the resulting thermal expansion or contraction. Changes in ocean circulation, especially at the western boundaries of oceans, affect sea level through changes in density and through geostrophic adjustments (i.e., through maintenance of a balance between Coriolis and the cross-stream pressure gradient). River discharge can raise water level at a station in the river due to a frictional effect, or to a lesser extent at nearby stations by the addition of less dense fresh water. Some of the river runoff may have been in the form of snow or ice for months of each year or for several years. Additional freshwater volume (that has been in the form of ice for centuries) can be added by the melting of glaciers or the ice sheets on Greenland and Antarctica. All of these phenomena produce signals much larger than the estimated 1 to 2 mm/yr sea level rise signal, and thus complicate the task of calculating such a trend.

A rise in global sea level due to the warming caused by the greenhouse effect could probably be caused by only two effects: (1) thermal expansion of the water column (due to a decrease in water density caused by its warming); and (2) the addition of water volume due to the melting of glaciers and ice sheets (assuming the warming did not also increase snow precipitation over the ice sheets). Water is also stored in groundwater, lakes, and as moisture in the atmosphere, but this potential effect on sea level (positive or negative) is probably minor unless there is a significant warming of the atmosphere and effect on climate. Regional changes in sea level effects could also result from climate change due to the warming, for example, a change in the transport of the Gulf Stream would affect the sea level along the coast of the southeast United States.

The various effect on water level and sea level cover a range of time scales. The largest signal at most gauges is the astronomical tide, whose range is usually on the order of meters (and on the order fo 10 m at a few locations). The tide, being a well defined periodic signal (with most of the energy at approximately twelve-hour and twenty-four-hour periods), can be easily taken out of the data record by filtering or harmonic analysis. In the monthly and yearly means used for sea level rise studies, the averaging of the hourly data eliminates the tide. Wind waves can be quite large (on the order of centimeters to meters) with periods on the order of seconds. Their effects are supposed to be eliminated right a the tide gauge, either damped out by the stilling well that surrounds the float or averaged out by rapid sampling; if not taken out completely their effects should be averaged out over long time periods because of their random nature. However, there has been concern about a nonlinear response in the stilling well leading to a small wave setup that could lead to a slightly higher water level measurement when the winds are stronger or from a direction most conducive to a higher sea state. During years with more storms and more waves, there could be a small increase in the calculated mean sea level. (This is an example of the kinds of possible problems that one must worry about when dealing with a small signal like sea level rise.) Storm surges (on the order of centimeters to meters and with periods on the order of minutes to days), caused by changes in atmospheric pressure and wind, also tend to be averaged out in monthly or yearly means. But, for gauges in shallow water, nonlinear interaction between storm surge and the tide modifies both, and the result could have an asymmetric aspect that might not average out and might be the cause of subtle variations from month to month and perhaps year to year.

Seasonal variations in sea level (on the order of 40 cm, or higher in special locations) can be caused by seasonal changes in temperature, seasonal changes in river runoff, or seasonal changes in wind. These variations can be calculated and removed from the record, or filtered out when using yearly means. But these variations must be considered in conjunction with the interannual sea level variations discussed next, since the key factors affecting the interannual signal often have a preferential season for their greatest effect.

The most difficult signal to remove is the interannual variation in sea level (10 to 20 cm), which itself plays an important role in climate and global change research (see end of this paper). In the tropical Pacific the most important interannual sea level variations are a result of El Nino Southern Oscillation (ENSO). But interannual sea level variations at other locations around the globe also reflect the climatic interaction between ocean and atmosphere, which must be understood in order to remove these variations from the mean sea level record. At islands the interannual sea level variations tends to correlate well with changes in water density; along continental coasts it tends to correlate well with changes in longshore wind stress and changes in ocean currents (especially along continental east coasts).

The removal of the interannual signal is probably the most important thing that can be done to improve the calculation of a sea level trend at a particular station. Whether the water level data record begins (or ends) during the crest or trough of a particular interannual “event” (e.g., and El Nino) can affect the resulting calculated trend, even for the longest records. Removal of the interannual signal has not generally been done, but many authors have recognized the sensitivity of the calculated trend to the length and time of the record and have avoided using shorter records, where the effect is more pronounced.

What Does the Trend Really Represent?

Once a trend has been calculated at a station, there is the crucial question of what that trend really represents. One very important factor is land movement. As mentioned above, a tide gauge actually measures relative sea level, and land movement will therefore be included in the trend calculated from the sea level data. There are a variety of causes of land movement including tectonic movement at convergent plate boundaries, subsidence due to sediment compaction and the extraction of water or oil from the ground, and glacial rebound. Glacial rebound causes uplift in areas that were covered by ice during the last ice age, subsidence in regions that were close to but not covered by the ice, and generally affects vertical motions over the entire globe. At some locations the large land movement is very obvious, for example, the glacial rebound occurring in Scandinavia and Alaska, and the tectonic activity in Japan; data from these locations have generally not been used to estimate trends in sea level change. (And yet a reliable trend from such an area might be needed to balance out dynamic effects found in estimates from other regions; see below.)

An important problem is how to remove the land movement from the trend calculation. There are modern geodetic techniques (discussed below) that in the near future should be able to determine the land movement to the millimeter accuracy required. In the mean time, many researchers either pick tide stations where land movement is thought to be small or use some global analysis scheme to try to average out the problem. A few have used post glacial rebound models to take out that type of land movement. Some researchers have tried comparing Holocene sea level rise rates (determined using carbon 14 dating techniques on coral-reef terrace sequences or fossil strand-line deposits) with rates calculated from the tide gauge data as a way to remove land movement. Others carry out local geologic studies on nontectonic movement, such as sediment compaction.

A second factor is the effect of regional oceanographic or meteorological phenomena that may have very low-frequency components that could affect the calculated trend. Low-frequency changes in wind speed or direction, atmospheric pressure, or ocean circulation might cause changes in sea level at specific locations, which could affect the trends calculated from data at those locations. Slowly changing wind speed or direction over several decades could have a low-frequency effect on sea level along coasts where the longshore component has the dominant effect on the interannual sea level. Or it could modify the ocean circulation, which in turn would affect the water density, which would affect the sea level at certain islands. Over the entire globe, regional low-frequency changes in sea level might balance each other out, but with insufficient global coverage the data may not be available to carry out the necessary averaging scheme.

The trend we see in historical sea level records may simply be part of a very low-frequency, global-scale natural variability and not really a trend at all–only perceived as a trend because our data records are too short. Variations on the order of a century or longer may even be possible. To prove this, we must find similar trends in other measurable parameters that could help explain such sea level variations. Unfortunately, other data sets going back a century or more may not have the necessary quality. Climate events such as El Ninos vary in size and global effects from decade to decade. If we look at the chaotic nature of ocean-atmosphere interactions that are beginning to be understood from studies of ENSO, its teleconnections(i.e., its remote consequences), and other climate phenomena, we can envisage a slowly evolving sequence of climatic “events” over the years, whose effects, when averaged, could have very long periods of variation and a slowly changing effect on sea level.

The problem of regional effects affecting a calculated trend can be worse if the tide gauge used is located inside an estuary or shallow bay. Slowly varying annual river discharges due to slowly changing annual precipitation patterns could provide a low-frequency sea level signal that might affect the calculated sea level trend. In shallow bays and estuaries nonlinear effects are important. The tide itself has a nonlinear effect on mean sea level, which will remain constant and not be a problem as long as the tide regime stays the same. However, in a relatively small shallow bay dredging can increase the tide range and thus affect the mean sea level, as can the tide range reduction caused by the slow filling in of a bay with sediment.

Long-term variations can conceivably result from any asymmetric nonlinear bias of shorter term processes that does not average out, for example, the small setup in a stilling well due to its nonlinear response to waves, mentioned above, which might lead to higher calculated sea levels during years with more storms and more waves. In such cases there might be differences between nearby gauges on the open coast and in more protected locations that could be studies to eliminate this potential problem.

Obtaining a Global Average

In trying to produce a meaningful global value for sea level rise one has many problems to overcome. One must try to avoid using stations where land movement is a problem. One must deal with biases that will enter due to the uneven distribution of tide gauges around the world, there being many more in the Northern Hemisphere than in the Southern Hemisphere, and perhaps no gauges in certain dynamic regions that might balance the results from other dynamic regions.

It might be necessary to study carefully the lowest frequency signals found a each station to see if a particular combination of stations can be found that causes these low-frequency variations to balance out globally. Then the trends from these stations would be used to produce an estimate of the global sea level trend. It may be necessary to include regions along both west and east boundaries of each ocean for each wind regime, north and south of the equator. Or we may have to wait for satellite altimetry results to tell us more about sea level variability on a global scale in order to know how to produce the best global average from the tide gauge records.

Recent Results

In recent years there have been numerous studies to calculate global sea level rise from the historical water level record. Most studies have found a rise in global sea level on the order of 1 to 2 mm/year over the last century and no strong evidence for an increase in the rate (i.e., an acceleration) of this rise in recent decades. As discussed above, there have been many problems to overcome in trying to determine such a global sea level rise value, and these have been approached in a variety of ways by various researchers.

IS IT THE GREENHOUSE EFFECT?

Even if we have confidence in the accuracy of a particular calculated sea level trend (and have somehow removed all land movement) and believe that sea level has risen by that amount over the last century, how can we know that this sea level rise was caused by global warming due to the accumulation of greenhouse gases produced by mankind?

During the last ice age, 18,000 years ago, sea level was approximately 100 m lower than it is today. It has been rising ever since, and, during the time the glaciers retreated from covering Canada and the northern United States, sea level rose at approximately five times the present rate. Could not the present slower rise simply be part of the same process?

In fact, the same question has been asked of global warming itself. Globally averaged air temperature data (with some of its own problems of uneven distribution and urban effects) has shown a warming trend since 1880. This increase began presumably before human-caused greenhouse warming could have had a significant effect. The rate of warming between the mid-1960s and the present is higher than that which occurred in the previous period of warming between the 1880s and 1940, but relative to the variability seen in the global temperature record it is unclear whether this is significant.

The globally averaged temperature record shows an increase from 1880 to 1940, then a decrease until the early 1960s, at which time a steeper increase begins that continues to the present. The sea surface temperature (SST) record from ships also shows an upward trend but with a period of decrease beginning around 1960. The global sea level curve is similar to the SST curve and also appears to have a decrease beginning around 1960. (SST measurements also have their own problems and much of the 0.5°C rise in SST between 1900 and 1970 may be the result of the changing measurement technique, from bucket reading to injection temperature.)

One might expect to see an acceleration in the rise of sea level if indeed the effect of the accumulation of greenhouse gases during the last century, and especially during the last half of the century, has had an effect. No researcher to date has convincingly demonstrated such an acceleration. The interannual signal that makes trend determination difficult creates even more of a problem in trying to find an acceleration. The same problem exists with the global air temperature data and trying to conclusively demonstrate a recent intensification in that upward trend.

It is possible that the greenhouse effect has not shown up yet because the ocean has absorbed the heat and transported it in some fashion to its depths. There has been much discussion about the possible influence of the ocean’s thermohaline circulation in delaying the onset of greenhouse-induced climatic change, and whether the greenhouse signal could be “hiding” in the deep ocean. Various researchers have used upwelling-diffusion climate models to show that if, for example, basin-wide upwelling was reduced there would follow a cooling of the sea surface.

So it is possible that it is too soon to see a significant greenhouse signal in air temperature or in sea level. But if the deeper waters are warming, should not that signal show up as a steric increase in sea level? There are several reasons why it might not. The coefficient of thermal expansion decreases with decreasing temperature (and increases to a lesser extent with increasing pressure). If deeper (cooler) waters warm and upper (warmer) layers cool, then the expansion in the deeper waters could be less than the contraction in the upper layers, and we might not see the effect, or, at the very least, the results might be very moderate and easily hidden by natural variability.

The present temperature and salinity profile data base appears to be inadequate to produce reliable, long series of steric heights to look at this question. In attempting to construct long data series at specific locations in parts of the ocean, one finds insufficient data when using observations taken over the years by both research ships and voluntary observing ships (sailing along trade routes). Attempts are being made at national oceanographic data centers to find missing historical data that could fill in crucial gaps.

There is also the possibility that a sea level signal due to warming of the deeper water layers might be confined to the “conveyor belt” area of the Atlantic or other areas where no tide gauges exist to record it.

If the ocean is absorbing heat, the atmosphere may not warm enough in high latitudes to have an effect (positive or negative) on the Greenland ice sheet. There is, in fact, a major question about whether global warming would lead to ice wastage or to ice accumulation on the Greenland and Antarctic ice sheets. A small increase in temperature may lead to an increase in snow precipitation over the ice sheets with an accumulation that could counteract the possible increase in melting or iceberg calving at the edges. There has been very limited data on accumulation and ablation. With our present (insufficient) knowledge, mountain glaciers appear to be a more likely candidate for contribution to sea level rise.

PROJECTIONS FOR FUTURE SEA LEVEL RISE

The most important question is of course: How much will sea level rise in the future? The problem is how to come up with a reasonable estimate? The simplest approach would be merely to extrapolate the sea level trend calculated from the historical sea level records, but we really have no proof that this represents a real global sea level rise, and we do not know what kind of lag there might be between the warming of the atmosphere and the warming of the ocean and the increase in sea level. A small extension of this, with the same problems, would be to correlate the recent historical sea level rise with the air temperature rise, and modify the sea level rise prediction for the predicted rise in air temperature.

Another approach is to model the phenomena that appear to affect eustatic sea level rise, for example, thermal expansion and ice melting from glaciers and the ice sheets, and use that to make predictions based on estimated increases in air temperature (whose prediction may in turn depend on knowledge of what the ocean is doing).

The projections all depend on the use of models dealing with the thermohaline circulation of the ocean (to estimate thermal expansion) and the addition or subtraction of water to Greenland and Antarctica. Such models are based on necessary simplifications of the physics involved, which itself may not be totally understood. It is probably not erroneous to say that we really do not know how much sea level will rise over the next century. If the deep ocean is indeed absorbing the additional heat due to greenhouse warming (and we cannot see it in sea level, either for lack of data or because there is compensating contraction in the upper layers) that could be either a great benefit (if it delays significant sea level rise for centuries) or a big problem (if it delays sea level rise only long enough so that by the time we are sure it is happening, it will definitely be too late to stop it).

All of the projections for sea level rise in the next century are less than the earlier larger estimates that caused great concern, but they are not insignificant and, if they occurred, could have serious consequences. It is important for sea level research to continue, because there are so many unanswered questions, and the problem is so complicated.

Long-term sea level changes are important whether due to global climate or local land movement. Extreme water levels (brought on by a combination of long-term sea level rise, interannual climate-caused events, seasonal effects, storm surges, and tides) produce the same impacts as future sea level rise; and there still is much work to be done in that area of research. As discussed below, there is also much to be learned about interannual-to-decadal sea level variation. Taking such variation out of the data record may improve our trend estimates and even eventually allow detection of an acceleration in sea level rise. And it may also help us understand some of the processes affecting the thermohaline circulation that may be postponing the expected sea level rise.

THE IMPORTANCE OF INTERANNUAL-TO-DECADAL SEA LEVEL VARIATIONS

Although most attention has been paid to the question of sea level rise, there is another aspect of climate and global change of equal importance in which sea level studies can play an important role. As mentioned above, the interannual-to-decadal sea level signal is much larger than the sea level trend. Interannual sea level correlates well with winds (along coasts), water density changes (at islands), and variations in ocean currents, all of which play an important role in the ocean-atmosphere coupled climate system.

There are immediate benefits to understanding interannual variations in climate. For example, there will be a large economic benefit if we can learn how to predict each El Nino, including its size and its global effects (such as floods, droughts, increase or decrease in hurricanes, effect on fish catches, etc.), more than nine months before it occurs.

Sea level research played an important part in understanding El Nino in the Tropical Pacific. It will also play a role in understanding climatic phenomena elsewhere on the globe. For example, one sees maximums and minimums in sea level records along the Atlantic Coast of the United States that coincide with El Nino Southern Oscillation events in the Pacific; these peaks are most pronounced in the winter. They appear to be a result of atmospheric teleconnection related to the westerlies; the interannual longshore wind component along the U.S. Atlantic Coast matches well with the interannual sea level and has similar maximums and minimums coinciding with El Ninos.

Sea level can thus be an important indicator of global teleconnections, which atmospheric researchers have been studying in pressure, temperature, precipitation, and other atmospheric data sets. Research making use of sea level records from all over the world should help improve our understanding of global teleconnection mechanisms, which are so important to climate. The long sea level data records will be very useful, since the chaotic nature of the climate system produces many different situations, and a long data record becomes necessary to study as many of these simulations as possible. No two El Nino events have been exactly alike; they vary in size and global effects from event to event.

Sea level from satellite altimetry has already begun to provide the synoptic pictures that have led to further understanding of El Nino, the Gulf Stream, and other climatic phenomena. It provides the potential for observing the ocean circulation and its variability on space and time scales inaccessible by any other means. Ocean circulation is a crucial part of understanding the coupled ocean-atmosphere system, playing a part in the two primary interactions between the ocean and the atmosphere, the exchange of momentum and the exchange of heat. In the former the winds of the atmosphere act on the ocean to produce ocean currents, and in the latter the ocean acts on the atmosphere and modifies air temperatures. The analogy frequently used is that the ocean is the (thermal and inertial) flywheel in the climate system. Sea level research can play a role in understanding this system.

FUTURE ADVANCES IN GLOBAL SEA LEVEL RESEARCH

Satellite altimetry data provide synoptic pictures of sea surface topography, and with these data one should be able to produce true global averages. Of course, we will have to accumulate altimetry data for many years in order to be able to determine a trend. We can in the meantime, however, learn much about global sea level variability and whether regional variations tend to balance each other globally. This may help us select the best tide gauges (with their longer, more accurate data and higher temporal sampling rate) to use to produce global sea level estimates. It is possible that some of the regional variability that made an accurate trend difficult to determine, may disappear in the global averages produced from altimetery. Altimetry will also play an important role in monitoring the size of the ice sheets on Greenland and Antarctica.

Other exciting recent developments deal with geodetic techniques for measuring land movement, which should eventually allow us to take that signal out of the sea level trends obtained from tide gauges. Very long baseline interferometry (VLBI) uses a network of radio telescopes thousands of kilometers apart to simultaneously track extragalactic radio sources and, after analysis of the differences in arrival times and their variation in time, to establish and maintain a terrestrial reference frame. Global positioning system (GPS) is a navigation system of satellites that provides relative position through the simultaneous view of more than one satellite from more than one land position. Satellite laser ranging (SLR) measures the round trip travel time of a pulse of laser-emitted light from a ground station to a target satellite and back. Absolute gravity measurements can also be used to measure vertical position. Uplift of a tide gauge would be accompanied by a decrease in gravity, as long as the movement did not include the introduction of additional mass near the observation point. Thus one must not locate a station near an aquifer (where the gravitation attraction of the changing groundwater would present a problem) or ocean surf, and corrections must be made for ocean tidal loading. VLBI, GPS, SLR and absolute gravity measurements should, in a few years, be able to determine vertical land movement to the millimeter level required. Further improvements to the glacial rebound models and techniques for determining paleo-sea level changes will also be valuable.

Although land movement due to glacial rebound occurs on a long enough time scale to be considered linear over the period of the historical sea level record from tide gauges, and it should not make a difference in the search for an acceleration in sea level rise in this data record, the accurate determination of land movement is still very important. If there is indeed an ocean-caused lag in seeing a sea level rise due to the greenhouse effect, and we therefore have to rely on models (instead of data analysis) for future projections of sea level rise, it is important that we have accurate trends of past sea level rise (free of land movement) for use in verifying (or calibrating) these models when using them in a hindcast mode. In addition, sea level rise, whether caused by global warming of local land movement is still important to coastal communities, and in many cases the land movement is the largest signal and thus of most concern from a practical or policy making perspective.

Future advances in sea level research will also require long series of temperature and salinity profile data, near key sea level stations, and perhaps in the North Atlantic where there are no sea level stations. More work will be done on improving the thermohaline circulation models. Likewise, new measurement techniques and improved ice models will be developed to look at the important question of the mass balance of the ice sheets on Greenland and Antarctica.