While European tide-gauge records can be found as early as the 1670s [Woppelmann et al., 2006], long-term records began in earnest only in the late 18th century, and not until the late 19th century in the southern hemisphere. Large gaps in time as well as geographical spread, however, complicate efforts to make tide-gauge-based determinations of global sea level rise for the 19th and even for the 20th century [Mitchum et al. 2010].
Still, researchers have devised a variety of methods to overcome these gaps and identify trends in global mean sea-level rise during the 20th century. The result: substantial agreement on the long-term trend but differences in assessment of variations over years and decades [Rhein et al., 2013].
Perhaps the best-known tide gauge reconstruction, and the first of its kind, combined modern satellite altimetry with historical tide-gauge measurements to fill in the gaps for the 1950 to 2000 period [Church et al. 2004]. This led to a globally averaged sea level rise of 1.8 millimeters per year, plus or minus 0.3. A recent extension of this method [Watson et al., 2015] compensated for vertical land motion using GPS estimates and found an increased rate of sea level rise over the previous 20 years—contrasting with previous work suggesting a slowing [Cazenave et al. 2014].
Another approach limits analysis to very long-term records with few discontinuities. Holgate [Holgate, 2007], for example, chose nine long and largely uninterrupted records from the Permanent Service for Mean Sea Level to create a global mosaic: New York from 1856 to 2003, Key West 1913-2003, San Diego 1906-2003, Balboa 1908-1996, Honolulu 1905-2003, Cascais 1882-1993, Newlyn 1915-2004, Trieste 1905-2004, and Auckland 1903-2000. Each record has at least 11 months of data per year and fewer than 15 days missing.
A third relies upon many shorter-term records that are then filtered, removing short-term variations to reveal longer-term trends. One research effort [Jevrejeva et al., 2006] also drew from the PSMSL to examine 12 large-scale ocean regions, removing biennial to decadal oscillations to derive a century-long global sea level trend. Their trend was comparable, for the 1993 to 2000 timeframe, to that yielded by TOPEX/Poseidon altimeter measurements.
A recent reconstruction averaged regional measurements across latitudinal bands to produce a global trend [Merrifield et al., 2009]. And a more recent effort [Thompson and Merrifield, 2014] employed satellite altimeter and tide-gauge data to reveal a faster rate of sea level rise in the southern hemisphere than in the northern.
One novel approach employed a “neural network,” a type of computer model that simulates the action of neurons arranged in interconnected layers. The model’s output again extrapolates tide gauge data to connect coastal to regional and global-mean sea-level readings, thus eliminating potential problems with vertical land motion, and to fill gaps in tide-gauge records [Wenzel and Schroter, 2010].
Radar altimetry from orbit revolutionized global sea-level observations, capturing variations across most of the planet’s oceans every 10 days, as well as at other time intervals. The first measurements precise enough to track changes in global mean sea level began with the launch of the NASA-CNES TOPEX/Poseidon satellite in 1992 (CNES is the French space agency). The spacecraft was equipped with two altimeters and a microwave radiometer, which corrected for the effects of water vapor on radar signal transmission; other instruments ensured precision tracking of the satellite’s orbital position [Mitchum et al. 2010].
TOPEX Poseidon’s mission ended in 2006, but the launch of Jason-1 in 2001 and Jason-2 in 2008 allowed long-term sea-level measurement to continue. Transmitter failure brought the Jason-1 mission, a collaboration of NASA and France’s Centre National d’Etudes Spatiales, to an end in 2013; a second collaboration, Jason-2, remains in orbit, and a Jason-3 mission has been launched. The data from the TOPEX-Jason series are nicely complemented by data from the European Space Agency’s (ESA) ERS-1, ERS-2, ENVISAT and Cryosat altimetric satellites, as well as the more recent SARAL/AltiKa, a collaboration between ISRO, the Indian space agency, and CNES.
The orbital view brought seasonal and yearly variations into sharper focus—for example, a significant rise in global mean sea level during the powerful 1997-98 El Niño event [Nerem et al., 1999], and, as we’ve seen, a sea-level drop in 2011 associated with La Niña [Boening et al., 2012].
Keeping watch on the global sea-level signal requires validation of satellite measurements against tide gauge data [Mitchum, 2000], as well as a modest upward adjustment in sea-level height because of the shifting shape of ocean basins due to glacial isostatic adjustment, occurring within the satellite coverage area [Douglas and Peltier, 2002]. Careful calibrations were needed to avoid potential data bias, for example, after a change was made to the TOPEX altimeter electronics in 1999 and after a time-series shift from TOPEX to Jason-1 in 2002 [Mitchum et al., 2010].
The continuous 22-years-plus record of altimetry measurements helped identify a likely rise in the rate of globally averaged sea level between 1993 and 2010 that is comparable to a similar rise between 1920 and 1950 [Rhein et al., 2013]. A recent study [Watson et al., 2015] used tide gauges to correct for a drift in the earlier TOPEX/POSEIDON record; the combined dataset is more powerful than either one separately. These authors also found an acceleration in the rate of sea level rise over the past 22-plus years.
The advent of gravimetric measurements with the twin GRACE satellites in 2002, along with more recent deployment of floating Argo sensors, opened the way to “closure” of the sea level budget—that is, when the sum of observed ocean mass and density changes equals total sea level change [Leuliette and Willis, 2011].
GRACE measures changes in water mass, including terrestrial storage in the form of groundwater, rivers, snow and ice, and mass changes within the ocean itself, as well as the movement of water between land and ocean.
Early attempts did not achieve closure of the sea level budget for four-year trend lines [Willis et al., 2008, Chang et al., 2010], leading to concerns about possible instrument drift. More recent efforts, however, led to reports of closure for more extended periods, including a NOAA report covering 2005 to 2013 ("The Budget of Recent Global Sea Level Rise, 2005-2013," by Eric Leuliette).
To capture changes in water mass accurately, shifts in atmospheric mass must be subtracted from GRACE’s gravitational measurements—along with changes in the mass of ocean basins, the lingering rebound effect from the loss of Ice Age glaciers [Tamisiea and Mitrovica, 2011].
One approach to achieving a high-precision dataset is the mass concentration (mascon) method, which breaks up GRACE’s gravitational measurements into discrete regions of higher mass. This allows more precise resolution of mass changes in smaller regions than more traditional “harmonic” solutions, which smooth gravitational measurements into a larger whole [Watkins et al., 2015].
Argo floats and in-situ observations
The deployment of floating sensors across the world’s oceans, known as the Argo project, reached a critical mass in 2007, with some 3,000 of the devices set adrift to measure temperature and salinity in the ocean’s upper 2,000 meters [Leuliette and Willis, 2011]. These sensors profile ocean expansion, the thermosteric sea-level rise that, as we have seen, is a consequence of ocean heat absorption. The Argo floats play critical roles in recent studies of trends in ocean heat content, which generally show increased warming over decades. One recent estimate, covering 1955 through 2010 [Levitus et al., 2012], relied upon historical data and more modern readings from the World Ocean Database 2009, additional data from NOAA through 2010, and Argo data that became available in early 2011. Some of the Argo data had been corrected by Argo quality-control teams, although uncorrected data also were used. (The authors say that temperature measurements from the floats, unlike salinity measurements, show few instances of data drift.)
Another recent study [Nieves et al., 2015] used Argo float data to determine that a decade-long slowdown in the rate of surface air warming—a “hiatus” reported by several researchers—can be explained by more heat accumulation in the Indian and Pacific oceans, at depth, than had been previously recognized.
Argo data also has been combined with ocean temperature measurements from the World Ocean Circulation Experiment in the 1990s to suggest a marked increase in thermosteric sea level through 2008 [Freeland and Gilbert, 2009]. The result comes with significant uncertainty, however, due to sparse sampling in the earlier project [Rhein et al., 2013].
Prior to deployment of Argo, ocean temperature analyses relied on bathythermographic instruments including expendable ones (XBTs) not designed for long-term monitoring of such tiny signals, conductivity-temperature-depth (CTD) probes and other methods [Levitus et al., 2009, Rhein et al., 2013]. Repeated hydrographic sections by ship (using ‘bottles’ that capture a small amount of water, with reversing thermometers) also have been used to create a reliable picture of temperatures below 2,000 meters, finding substantial indications of warming at depth [Purkey and Johnson, 2010].
The Earth keeps its own records of sea level change across thousands and even millions of years, though one that is often fragmentary and that tells a tale of relative sea level (that is, combining sea level rise or fall due to changes in ocean volume with the rise or fall of coastal land). Such sea level reconstructions must then be fitted into a shifting context of vertical land movement, including ongoing glacial isostatic adjustment, tectonic uplift, and the deformation of ocean basins by movement of water mass among ice sheets, terrestrial water sources and oceans [Lambeck et al., 2010].
These proxy records are preserved in a variety of marine and terrestrial settings: sediments and organisms in deep ocean cores, for example, or once-submerged shorelines and terraces, uplifted and exposed. Analysis of oxygen isotope ratios in the shells, or tests, of tiny ocean organisms called foraminifera reveal oscillations in relative volumes of ocean and ice [Shackleton, 1987]. Dating of uplifted fossil reefs using radiometric techniques, such as uranium-series measurements, have increased the precision of sea-level estimates for the Quaternary period, from about 2.6 million years ago to the present [Lambeck and Chappell, 2001].
Biological traces of sea level change can be seen in fossil creatures that lived in intertidal zones, distributed over a reliably vertical spread according to their species’ requirements. Barnacles, mussels, corals, tube worms, oysters and boring mollusks are among the organisms that have yielded enough data to track sea level changes across centuries and millennia [Lambeck et al., 2010]. One intriguing life form, the coral microatoll, can preserve complex records of changing relative sea level, written in the undulating shape of its growth bands. These disk-shaped colonial formations, as much as 20 feet across, consist of a dead interior ringed by living coral, which can grow no higher than low tide. Falling sea levels can kill the upper part of the organism, while the lower portion continues to grow outward, leaving a distinct pattern. Rising sea levels lead to new growth over the previous dead interior. And sea level fluctuations leave wavy patterns in the growth rings that again are legible to scientists [Woodroffe and McLean, 1990; see also Lambeck et al., 2014].
Geological markers for sea level change include tidal notches, surf benches and other erosional features [Lambeck et al., 2010]. Under certain conditions, archaeological sites, especially in the Mediterranean, can track relative sea levels, including the Phlagrean columns mentioned above. The oldest known such site is Cosquer cave with its flooded paintings in southern France, inundated some 10,000 years ago [Lambeck and Bard, 2000]. Roman fish tanks carved from coastal rock as long ago as the first century B.C., 8,000-year-old Israeli wells, and sunken cities such as Baia provide valuable records of changing sea levels [Lambeck et al., 2010].