by Judith Curry
Part IV of the Climate Etc. series on sea level rise focuses on the satellite era (since 1993), including the recent causes of sea level variations and arguments regarding the acceleration (or not) of recent sea level rise.
Part III considered historical sea level rise in the 19th and 20th centuries and Part II provided an overview of the ‘relatively’ recent geological evidence for sea level variability and rise.
Satellite altimetry
Since 1992, measurements of global satellite sea level have been obtained from satellites. Satellite altimeters measure the time taken by a radar pulse to travel from the satellite antenna to the sea surface and back to the satellite receiver. Converting the signal received by the satellite altimeter to global sea level heights is a complex undertaking.
As described by Ablain et al. (2016), the following corrections need to be applied to the SSH measurement from satellite altimeters: propagation corrections as the altimeter radar wave is delayed during atmosphere travel (ionospheric correction, wet tropospheric correction, dry tropospheric correction), and ocean surface correction for the sea state which directly affects the radar wave (electromagnetic bias). After making these corrections for interferences of the radar signal, the travelling time of the radar signal time is transformed into a distance, or ‘range’. Additional corrections are made for tides, and the ocean response to atmospheric pressure variations.
Determination of sea level height (SSH) from the altimeter range requires a reference to the mean sea surface at rest. To accomplish this, satellite altimetry datasets are complemented by satellite data that provides gravimetric measures — measures of the distribution of mass on the Earth and oceans. The force of gravity is affected by density anomalies in Earth’s interior, the rotation of Earth and topographic features. The Gravity Recovery and Climate Experiment (GRACE) and Gravity field and steady-state Ocean Circulation Explorer (GOCE) satellite missions are used to map the Earth’s gravity field and its changes through time. This gravity field provides the basis for calculating the height of the geoid that corresponds to the mean sea surface at rest. Subtracting from the measured SSH from a reference mean sea surface provides a ‘SSH anomaly’. Calculation of the SSH anomaly at each point in the ocean is repeated very 10 days – the satellite track repeat cycle.
Ablain et al. (2016) has assessed the errors in global mean sea level (GMSL) determined from satellites. Regarding the GMSL trend, an uncertainty of 0.5 mm/year was estimated over the whole altimetry era (1993–2015) within a confidence interval of 95%. The main source of error is the radiometer wet tropospheric correction with a drift uncertainty in the range of 0.2– 0.3 mm/year. Orbit error and the altimeter parameters instabilities add additional uncertainty, of the order of 0.1 mm/year. The uncertainties are higher in the first altimetry decade (1993– 2002). Differences between TOPEX-A and TOPEX-B (February 1999), TOPEX- B and Jason-1 (April 2003), Jason-1 and Jason-2 (October 2008) lead to errors of 2, 1 and 0.5 mm, respectively, result in a GMSL trend uncertainty of about 0.1 mm/year over the 1993–2014 period.
Significant work has been done to devise methods to accurately calibrate altimeter measurements against a global network of tide gauges. As a result, a number of drifts and bias changes have been discovered and corrected, including an early software error that caused the estimate to be nearly 7 mm yr-1 too high, drifts in the water vapor correction from the microwave radiometer,
and changes in the sea state bias model. Calibration efforts are ongoing, which is essential for obtaining an accurate climate record from satellite altimetry.
Summary. Satellite measurements of global sea level have been available since 1992, and the technology is under continuing development. Complex analysis methods are required to transform raw satellite measurements into sea level variations, including the correction and piecing together of records collected over many years by ageing and changing satellites. Estimates of sea level change made using satellite-collected data are associated with many uncertainties in the data processing; with time, the uncertainty in current analysis methods and datasets may be revised as addition errors are uncovered. There is some inconsistency between the results derived by different research groups for the interannual variability, owing to differences in making the complex adjustments. These uncertainties underscore the need for continual scrutiny of the satellite and in situ tide gauge data, plus the need for independent observing systems such as multiple satellite altimeters with differing instrument designs, the tide gauge network, in situ ocean temperature observing system, and gravimetric satellites.
Satellite observations of SLR since 1993
From the IPCC AR5, Chapter 3:
The rate of GMSL rise from 1993–2010 is 3.2 [2.8 to 3.6] mm yr-1 based on the average of altimeter time series published by multiple groups. As noted in AR4, this rate continues to be statistically higher than that for the 20th century. There is high confidence that this change is real and not an artefact of the different sampling or change in instrumentation, as the trends estimated over the same period from tide gauges and altimetry are consistent. Although the rate of GMSL rise has a slightly lower trend between 2005 and 2010, this variation is consistent with earlier interannual fluctuations in the record (e.g., 1993–1997), mostly attributable to El Niño/La Niña cycles.
Products from six processing groups are available for the altimetry-based sea level data, based on TOPEX/Poseidon, Jason-1 and Jason-2:
Figure 1 shows the GMSL time series from CU, which is current to 2/12/18
Figure 1. From University of Colorado,downloaded 2/12/18
Substantial interannual variations are seen, which are associated primarily with ENSO. The large trend between 2011 and 2016 is associated with a very strong LaNina (2011) and a very strong El Nino (2016).
Ablain et al. (2016) compared the Multivariate ENSO Index (MEI) with the global sea level time series after removing the mean trend (Figure 2). Recent studies have shown that the short-term fluctuations in the altimetry-based GMSL are mainly due to variations in global land water storage (mostly in the tropics), with a tendency for land water deficit (and temporary increase of the GMSL) during El Nino events [LINK] and the opposite during La Nina [LINK].
Figure 2. From Ablain et al. (2016)
A recent update to the altimeter sea level dataset is associated with a correction to the TOPEX altimeter data in the way that satellite drift is treated. Beckley et al. (2017) and Watson et al. (2015) have argued for adjusting the TOPEX sea level data downwards during the period 1993-1999. This adjustment is included in Figure 1 above.
Spatial variability of sea level rise
Arguably the most important contribution from the satellite altimeters to our understanding of sea level is associated with the spatial variability of sea level rise. Over the globe, significant regional variations occur in the rate of sea-level change. These variations are partly due to variations in the rate of warming and salinity changes between different regions, and the proximity to discharges of meltwater. But primarily these variations reflect the influence of major ocean circulation systems that redistribute heat and mass through the oceans. As a result, at any location around or within the oceans, the observed sea level behavior can differ significantly from the global average. Additionally, this understanding of the spatial variability of sea level rise is very important for interpreting the tide guage record of sea level rise.
Ablain et al. (2016) provides a map of sea level trends over the period 1993-2014. Regional trend errors generally range from 1 to 3 mm/yr. Trends are not significant in areas of high oceanic variability. Note the very large sea level rise anomalies in the tropical west Pacific and the South Pacific and Indian Oceans.
Figure 3. From Ablain et al. (2016)
Reconciliation with tide gauges
The IPCC AR5 states:
It is very likely that the mean rate was 1.7 [1.5 to 1.9] mm yr-1 between 1901 and 2010 and increased to 3.2 [2.8 to 3.6] mm yr-1 between 1993 and 2010.
Rates of GMSL rise during the altimeter period (since 1993) are about twice as large as the rates of GMSL rise during the period 1900-1993 (1.5 ± 0.2 mm/yr). Is this increase in the rate of sea level rise real, or does this reflect an apples-to-oranges comparison using fundamentally different measuring technologies and assumptions?
There have been numerous studies that compare the tide guage with altimeter values during the period since 1993:
Merrifield et al. (2009): After 1990, the global trend increases to the most recent rate of 3.2 ± 0.4 mm yr-1, matching estimates obtained from satellite altimetry.
Jevrejeva et al. (2014): There is a good agreement between the rate of sea level rise (3.2 ± 0.4 mm· yr-1) calculated from satellite altimetry and the rate of 3.1 ± 0.6 mm·yr-1 from tide gauge based reconstruction for the overlapping time period (1993–2009).
Hay et al. (2015): Our analysis, which combines tide gauge records with physics-based and model-derived geometries of the various contributing signals, also indicates that GMSL rose at a rate of 3.0 ± 0.7 millimetres per year between 1993 and 2010 . . . is also consistent with the estimate based on TOPEX and Jason altimeter measurements (3.2 ± 0.4 mm yr-1 for the period 1993–2010.)
Dangendorf et al. 2016: our estimate of 3.1 ± 1.4 mm⋅y−1 from 1993 to 2012 is consistent with independent estimates from satellite altimetry.
Figure 4 shows that there is some sort of SLR acceleration that occurred in the 1990s, and that this increase is evident in the tide guage record. Merrifield et al. (2009) examined the tide guage data from 1965 to 2000 for an interpretation. Prior to the late-1980s or so, the global trend is relatively steady. After 1989, there is a sharp increase in the rate of sea level that continues through the 1990’s.
Figure 4. From Merrifield et al. (2009)
Merrifield et al. (2009) note that the Northern Hemisphere oceans play a surprisingly small role in the acceleration. Dangendorf et al. (2016) found that this sharp increase was geographically dominated by the Indian Ocean–Southern Pacific region, marking a transition from lower-than-average rates before 1990 toward unprecedented high rates in recent decades. Merrifield et al. (2009) identified a covariation of regional sea level in the tropics and southern oceans that represents a shift from a state where the two regions once varied out of phase to now apparently varying more in phase. These ‘hot spot’ regions are clearly visible in the spatial map of rates of sea level rise (Figure 3, preceding section).
Note: the recent values of sea level rise of 1.1 or 1.2 mm/yr between 1900 and 1990 imply a greater acceleration in the 1990’s.
The ‘acceleration’ debate
The important question is not ‘is the long-term rate of sea-level rising’, since the geological, tide-gauge and satellite record all agree that it is and, all other things being equal, will continue to do so. Rather, in context of detecting a human influence on sea level rise, the key question is: ‘is the rate of sea-level rise accelerating?’
Recent headlines (February 2018) proclaim:
- CNN: Satellite observations show sea levels rising, and climate change is accelerating it
- USA Today: Miami could be underwater in your kid’s lifetimes as sea level rise accelerates
- The Weather Network: Global sea level rise on faster pace than expected.
Before delving into the recent Nerem et al. paper that is subject of these articles, I provide some context for interpreting acceleration or deceleration over the relatively short time period since 1993.
The IPCC AR5 Chapter 3 acknowledges that a long time series is needed to detect acceleration in sea level rise from human caused climate change. “While technically correct that these multidecadal changes represent acceleration/deceleration of sea level, they should not be interpreted as change in the longer-term rate of sea level rise, as a time series longer than the variability is required to detect those trends.”
Nevertheless, there have been several recent papers addressing acceleration/deceleration in sea level rise over the short time period (since 1993) of the satellite altimeter record, searching for a human fingerprint:
From Cazenave et al (2014) The rate of sea level rise.
Since the early 1990’s, sea level rose at a mean rate of ~3.1 mm/yr. However, over the last decade a showdown of this rate of about 30% has been recorded. It coincides with a plateau in Earth’s mean surface temperature, known as the recent pause in warming. Here we present an analysis that separates interannual natural variability in sea level from the longer-term change probably related to anthropogenic global warming. We find that when correcting for interannual variability, the past decade’s slowdown of global mean sea level disappears.
Fasullo, Nerem & Hamlington (2016), Is the detection of accelerated sea level rise imminent?
“Global mean sea level rise estimated from satellite altimetry provides a strong constraint on climate variability and change and is expected to accelerate as the rates of both ocean warming and cryospheric mass loss increase over time. In stark contrast to this expectation however, current altimeter products show the rate of sea level rise to have decreased from the first to second decades of the altimeter era. Here, a combined analysis of altimeter data and specially designed climate model simulations shows the 1991 eruption of Mt Pinatubo to likely have masked the acceleration that would have otherwise occurred. This masking arose largely from a recovery in ocean heat content through the mid to late 1990 s subsequent to major heat content reductions in the years following the eruption. A consequence of this finding is that barring another major volcanic eruption, a detectable acceleration is likely to emerge from the noise of internal climate variability in the coming decade.
Nerem et al. recently published a paper entitled Climate-change-driven acceleration detected in the altimeter era. The punchline conclusion:
Using a 25-y time series of precision satellite altimeter data from TOPEX/Poseidon, Jason-1, Jason-2, and Jason-3, we estimate the climate-change–driven acceleration of global mean sea level over the last 25 y to be 0.084 ± 0.025 mm/y2.
This follows recent publications by Cazenave et al. (2014) and Fasullo et al. (2016) that found no acceleration, or even found deceleration. What changed?
Apart from several additional years of data, the acceleration hinges on the new adjustment to the TOPEX record during the period 1993-1999 recommended by Beckley et al. (2017) and Watson et al. (2015). Nerem et al. (2018) use this revised data set as the basis for statistically ‘eliminating’ the effects of ENSO and the 1992 Pinatubo eruption. Using a simple quadratic fit the residual time series, they identify an acceleration that the attribute to ‘climate change’, without actually stating whether or why this climate change can be attributed to humans (instead of e.g. multi-decadal ocean oscillations). The implication that this is associated with human-caused climate change comes from this statement:
Coupled with the average climate-change–driven rate of sea level rise over these same 25 y of 2.9 mm/y, simple extrapolation of the quadratic implies global mean sea level could rise 65 ± 12 cm by 2100 compared with 2005, roughly in agreement with the Intergovernmental Panel on Climate Change (IPCC) 5th Assessment Report (AR5) model projections.
The flip from deceleration to acceleration hinges on a substantial adjustment to the first 6 years of the TOPEX record, which is associated with much greater uncertainty than the later JASON data. And this is not to mention the questionable statistical methods used to ‘eliminate’ the impact of Pinatubo and ENSO, and to determine an acceleration (these issues will be addressed in Part V).
UPDATE from frankclimate in the comments:
Re Nerem et al 2017: I think that the application of a quadratic fit is not justified at all. I digitized the Fig.1 of this paper http://www.pnas.org/content/pnas/early/2018/02/06/1717312115.full.pdf
for the years after 2000 to avoid the TOPEX and/or the Pinatubo issues. Thereafter I calculated the linear trends 2000…2016 ( for Nerem et al) and 2000…2017 for the “Colorado” data. The residuals in annual resolution:
https://i.imgur.com/X6HGci6.jpg
The “ENSO removal” didn’t work. IMO they didn’t remove the ENSO index but reduced the impacts by about a half. See 2011 and 2016. In the raw data the ENSO blop is reduced in 2017 but not in the paper whrere the data end in the end of 2016. The relation stands and in the end they have a ENSO-blop which influences a trend very much. Together with the data before 2000 this gives only a plea to estimate a quadratic trend. This could have been seen also in their Figure 2:
https://i.imgur.com/fGHsFUz.jpg
The monthly datapoints inscribe much noise but the not succesful removal of the 2015/16 ENSO-impacts are clearly visible and also the suspicious behaviour in the late 90s. As a reviewer I would have asked for an annual resolved record and would have pointed to the ElNino in the end of the record.
In my opinion, the value of the altimeter data is in understanding regional and interannual variability, and that the first 6 years of altimeter data should be pretty much ignored in climate change arguments. In any event, the altimeter data set is not useful by itself (owing to its short length) for detecting long-term accelerations that could be attributed to human-caused climate change.
The bigger issue of whether there is a detectable signal from humans in the record of sea level rise will be addressed in Part V.
Controversies surrounding the altimeter-derived SLR data
The controversies surrounding the altimeter SLR data set are associated with the myriad corrections/adjustments that are made to the data set.
A recent paper by Chen et al. (2017) illustrates the difference between the unadjusted and adjusted SLR data, where the red and orange curves are adjusted (CSIRO), and the blue-gray colors correspond to unadjusted measurements from three different teams:
Figure 5: From Chen et al (2017).
Specifically, the unadjusted data refers to omitting the glacial isostatic adjustment (GIA) or Global Positioning System (GPS) data set to correct for the effects of vertical land motion. Figure 5 shows a deceleration using the unadjusted rates of SLR. With regards to the sea level rate figure (bottom), details are not given regarding exactly what it represents, but it appears to be calculated from 10 year averages.
The implication is that any acceleration in rates is largely associated with the VLM adjustments. The generally recognized uncertainty in the GIA/VLM adjustments is ±0.3 mm/yr.
In (2004), Nils Axel Morner published a paper that points out that the raw satellite data shows barely any rise:
The raw data from the TOPEX/POSEIDON sea-level satellites, which operated from 1993-2000, shows a slight uptrend in sea level. However, after exclusion of the distorting effects of the Great El Niño Southern Oscillation of 1997/1998, a naturally-occurring event, the sea-level trend is zero.
Nerem et al. (2007) published a rebuttal to Morner, stating that Mörner’s claim that sea levels are not rising has been criticised for ignoring correctly calibrated satellite altimeter records all of which show that sea levels are rising.”
The rebuttal systematically goes through all the adjustments and corrections ignored by Morner:
Satellite altimetry is somewhat unique in that many adjustments must be made to the raw range measurements to account for atmospheric delays (ionosphere, troposphere), ocean tides, variations in wave height (which can bias how the altimeter measures sea level), and a variety of other effects. In addition, the sea level measurements can be affected by the method used to process the altimeter waveforms, and by the techniques and data used to compute the orbit of the satellite.
One of the errors is caused by a drift in the TOPEX Microwave Radiometer (TMR). It was first observed in sea level via a comparison to tide gauges, and was verified to be caused by the TMR via comparisons to other orbiting microwave radiometers and radio- sondes. It caused a drift of nearly −1.2 mm/year in measured GMSL until early 1998, and then a bias of −5 mm. A second major error was introduced when the redundant TOPEX altimeter was turned on in early 1999 due to degradation in the original instrument. Since the electronics of the redundant altimeter were different, it caused an apparent bias in the GMSL measurement related to the Sea State Bias (SSB). The sense of the bias was such to cause an incorrect sudden drop in GMSL from the end of 1998 to the beginning of 1999 of nearly 10 mm. This error is removed when an updated SSB model is applied.
The net result of this exchange is that this does not inspire confidence in the altimetry data, and makes me wonder whether their uncertainty/error assessment is reasonable.
A note on why I prefer blog dialogue to attempting to conduct this via journal publications. Morner’s 2004 publication was submitted in 2001. The Nerem et al rebuttal is published in 2007, and Morner’s response is published in 2013!
In his (2013) response to Nerem et al., Morner states:
“It is because of the introduction of additional calibrations — and those “calibrations” are subjective interpretations; not objective readings. Consequently, they are opinion-dependent. “We adopt the rate given by Douglas (1991,1995) of 1.8±0.1 mm/yr” [for long-term sea level rise], Mitchum (2000) states. This rate, however, is widely debated and far from generally accepted.
Wait a minute. The latest/greatest estimates for long term sea level rise is 1.1 or 1.2 ± 0.2 or .03 mm/yr. How would incorporating these lower long-term rates of sea level rise influence the interpretation of altimeter measurements? I haven’t yet dug into the weeds sufficiently on this, would appreciate any insights on this.
The JASON Products Handbook provides a very detailed technical reference on processing and interpretation of the raw satellite data. The errors in many of the processing steps are measured in centimeters; I have no idea how to reconcile these numbers with reported confidence intervals in sea level rise numbers. Apparently spatial averaging reduces these errors to millimeters.
The accuracy of Jason series altimetry is limited to +/- 3.4 centimeters (possibly as good as +/- 2 cm). Many of the uncertainties in the data processing are even larger than the inherent measurement uncertainty. The assessment of errors in the altimetry sea level data set seem to come more from the differences between ‘independent’ analyses of the dataset, rather than careful consideration of all of the uncertainties involved. The end result is that the ‘calibrations’ are far larger than the derived changes in global mean sea level. I have no doubt that many errors and sources of uncertainty in the altimeter data set have yet to be identified and understood. The recent error highlighted by Nerem et al. (2018) is a case in point.
Sea level rise budgets during the satellite era can in principle be used as a check on the altimeter-derived sea level rise. There is a plethora of such studies that will be explored in more detail in in Part V on attribution. Here I provide the budget from Chen et al. (2017) that includes both adjusted and unadjusted rates of sea level rise:
Figure 6. From Chen et al. (2017)
The components of the SLR budget reasonably match the adjusted altimeter record (not the unadjusted record). Notice that the sum of steric and glacier components follow the slope (but not magnitude) of the unadjusted SLR rates. Greenland melting is primarily responsible for increase in sea level rates. The ice sheet melting is determined from GRACE satellite using many adjustments, including the GIA adjustment. I haven’t deeply dug into exactly what is done for determining ice sheet contribution to SLR, but there may be some circular reasoning involved in the adjustments that promote a balance with the adjusted SLR rates determined from satellite altimetry — I would appreciate any insights on this (note I will be digging deeper into this issue for Part V).
As concluded by Wunsch et al. (2007) with respect to the satellite measurements:
At best, the determination and attribution of global-mean sea-level change lies at the very edge of knowledge and technology. Both systematic and random errors are of concern, the former particularly, because of the changes in technology and sampling methods over the many decades, the latter from the very great spatial and temporal variability. It remains possible that the database is insufficient to compute mean sea-level trends with the accuracy necessary to discuss the impact of global warming – as disappointing as this conclusion may be. The priority has to be to make such calculations possible in the future.
Forthcoming Part V: Attribution addresses arguments for the causes of sea level rise since the 19th century, with a focus on the period since 1950. In particular, we assess the arguments for attributing any of the recent sea level rise to human caused global warming.
JC note: I am continuing to update Parts I-III, based on your comments and additional information that I come across. Your emails and blog comments have been extremely useful to me in this endeavor. Keep the comments and emails coming!
