from CO2 Science: The authors write that “satellite altimetry measurements since 1993 have provided unique information about changes in global and regional mean sea levels,” suggesting a mean rate of rise of 3.2 mm/yr for global sea level over the period 1993-2012 (Boening et al., 2012; Cazenave et al., 2012), which “notably exceeds the estimate of 1.8 mm/yr sea level rise for the 20th century (Bindoff et al., 2007).”
So which rate is closest to the truth?
What was done
In a study designed to answer this question, Jevrejeva et al. (2014) say they “renew the global sea level [GSL] reconstruction by Jevrejeva et al. (2006), using monthly mean sea level data collected by the Permanent Service for Mean Sea Level (PSMSL) covering the observations from 1807 to 2010,” thereby improving the GSL reconstruction by increasing data coverage “by using many more stations, particularly in the polar regions, and recently processed historic data series from isolated island stations,” as well as by extending the end of the reconstruction from 2002 to 2009.
What was learned
Quoting the five researchers, “the new reconstruction suggests a linear trend of 1.9 ± 0.3 mm/yr [7.5 inches per century] during the 20th century” and “1.8 ± 0.5 mm/yr [7 inches per century] for the period 1970-2008.”
Global sea level reconstruction since 1807, blue shadow represents 5 and 95% confidence interval
What it means
Although some regions have recently experienced much greater rates of sea level rise, such as the Arctic (3.6 mm/yr) and Antarctic (4.1 mm/yr), with the mid-1980s even exhibiting a rate of 5.3 mm/yr (Holgate, 2007), this newest analysis of the most comprehensive data set available suggests that there has been no dramatic increase – or any increase, for that matter – in the mean rate of global sea level rise due to the historical increase in the atmosphere’s CO2 concentration.[Therefore, there is no evidence of any human influence on sea levels]
The paper:
Trends and acceleration in global and regional sea levels since 1807. Global and Planetary Change 113: 11-22. Jevrejeva, S., Moore, J.C., Grinsted, A., Matthews, A.P. and Spada, G. 2014.
We use 1277 tide gauge records since 1807 to provide an improved global sea level reconstruction and analyse the evolution of sea level trend and acceleration. In particular we use new data from the polar regions and remote islands to improve data coverage and extend the reconstruction to 2009. 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). The new reconstruction suggests a linear trend of 1.9 ± 0.3 mm·yr− 1 during the 20th century, with 1.8 ± 0.5 mm·yr− 1 since 1970. Regional linear trends for 14 ocean basins since 1970 show the fastest sea level rise for the Antarctica (4.1 ± 0.8 mm·yr− 1) and Arctic (3.6 ± 0.3 mm·yr− 1). Choice of GIA correction is critical in the trends for the local and regional sea levels, introducing up to 8 mm·yr− 1 uncertainties for individual tide gauge records, up to 2 mm·yr− 1 for regional curves and up to 0.3–0.6 mm·yr− 1 in global sea level reconstruction. We calculate an acceleration of 0.02 ± 0.01 mm·yr− 2 in global sea level (1807–2009). In comparison the steric component of sea level shows an acceleration of 0.006 mm·yr− 2 and mass loss of glaciers accelerates at 0.003 mm·yr− 2 over 200 year long time series.
Full paper with figures is available here: http://www.sciencedirect.com/science/article/pii/S0921818113002750#f0015
It looks like the sea level rise has the usual 60-year AMO signal it, like everything else, so the “acceleration” from 1990 – present appears to be nothing more than an artifact of that. It’s becoming increasingly clear that most climate alarmism has derived from a failure to account for the AMO.
This oscillation is also the explanation for Don K.’s confusion. Figure 3 shows the no/slow-growth period from 1960 – 1990, which offsets the fast growth after 1990.
“I ran WebPlotDigitizer on Figure 3 and got the following results”
Thanks AJ, better than nothing but it looks rather approximative. I may have a look later.
AJ says:
May 21, 2014 at 9:55 am
Greg Goodman says:
May 21, 2014 at 8:41 am
——————————————-
It looks like AJ Webplot Digitizer is good so we can now update with Jevrejeva’s 2014 data.
Comparing Jevrejeva 2008 and 2014 annual sea level increase numbers – pretty close.
http://s28.postimg.org/vdw078nlp/1900_2009_Jevrejeva_Sea_Level_Acceleration_Tide.png
Now I am most interested in is sea level accelerating? How does it compare to the satellite altimetry. So here is the 4 series from 1993 to 2009.
The numbers are all over the place. Potentially decelerating. The average over the period could something like the satellite altimetry of 3.26 mms per year but this is semi-deceiving because the numbers were higher at the beginning.
http://s28.postimg.org/cnzfcr5h9/1993_2009_Sea_Level_Acceleration_Tide_Gauges.png
Now we’ll add in the satellite altimetry from 1960 to 2014 for a comparison. Its hard to say there is a real acceleration in this data.
http://s4.postimg.org/ffs10eoot/1960_2014_Sea_Level_Acceleration_Satellite_Tide.png
—————–
Some of the sea level reconstructions from tide gauges are available here. (If you want the updated Church and White 2011 data, you’ll have to download a zipped file from CSIRO linked on this page).
http://www.psmsl.org/products/reconstructions/
Hi Bill,
You can also calculate the acceleration rate by using a quadratic fit. For example, this will work:
=2*LINEST(B2:B112,A2:A112^{1,2})
For an explanation see:
http://stackoverflow.com/questions/10857252/quadratic-and-cubic-regression-in-excel
This is how it was done in this study. From the paper:
3.4. Acceleration in global and regional sea levels
We estimate an acceleration of 0.02 ± 0.01 mm·yr−2 in global sea level by the conventional method, defining the acceleration as the second derivative of sea level with time (twice the quadratic coefficient), measured in mm/yr^2.
Really not sure if further fitting could be taken seriously without getting the real data.
It’s going to be pretty close to the author’s results and no one is going to get any more accuracy unless they start with the correct data, not an approximate “webscan” of a graph in a pdf.
It looks like there may be some interesting info in this dataset but I see no point doing any further d.p. unless I have the true data.
If the authors report 0.001 +/- ??? fitting post 1860 with a rising half a cycle at the end, there’s probably -0.002 if you remove the cycle.
It’s sod all either way. Call it linear.
The question this raises in my mind is what happened around 1850-1870 that changed a down slope into a upward slope. It was pretty punctual on this scale of events.
Turning it into a quadratic term applied to the full data range seems highly questionable.
http://s28.postimg.org/vdw078nlp/1900_2009_Jevrejeva_Sea_Level_Acceleration_Tide.png
Looking at rate of change it’s been dropping since 1995 and just gone -ve since about 2003. Reminiscent of SST.
Greg… I don’t disagree with you. The estimation process is probably offset by up to 0.5 years which usually won’t get you the peaks and valleys. IMHO, it’s probably ok for looking at trends over say 20yrs, but not if you were looking at year to year variability or say the correlation of ENSO with MSL.
And I agree that we can call this reconstruction linear from 1900 onwards. Also, I don’t really have an opinion about the merits of any of these studies. Most look plausible to me, including the satellite reconstructions. Heck I even rolled my own for the NH from 1930 onwards which seemed fairly consistent with Jevrejeva, C&W, and the satellite data.
Greg says:
May 21, 2014 at 1:50 pm
The question this raises in my mind is what happened around 1850-1870 that changed a down slope into a upward slope. It was pretty punctual on this scale of events.
—————————–
The issue in this case is that majority of the tide gauges in the 1807 to 1870 period were in Europe and the north-east US. They were mostly in areas where post-glacial rebound was occurring so sea level was falling on average for these gauges.
I have a chart using the PMSL database (1807 to 2009) and each point on here is an individual tide gauge’s change in sea level by year. You can see there are groups that are rising and groups that are falling and these various groups are post-glacial rebound/subsidence trends.
http://s17.postimg.org/si6ly8q27/Sea_Level_Measurements_PMSL.png
ie. without going back and re-adjusting each gauge’s numbers for the current estimates of rebound/subsidence from the quite accurate (after at least 3 years of measurement) GPS stations, one cannot be sure what the real sea level change was until about 1930 when there was enough global coverage so that rebound/subsidence was no longer a net factor to the average.
Thanks Bill, so it’s quite simply sampling bias.
So unless the authors were unaware of this issue, which seems unlikely after decades studying the subject, it most inappropriate of them to have fitted a quadratic to the combined record in the first place and even more so to have featured that “acceleration” in the abstract.
The true acceleration is 0.001 mm / yr^2 +/- 100%
IOW, sod all.
I don’t know if that explanation is sufficient Bill. Looking at your graph, even that group does not continue downwards.
So unless those were records that ended circa 1860 there is still a change of direction that requires explanation.
One of the biggest problems in this area, is that the PMSL database is extremely hard to work with. I don’t know why they can’t put together a nice, easily trackable, publicly available on-line database where one can trace trends for individual stations over time because they have all the data electronically …
… But, the simple fact is, the PMSL database is a complete mess.
Why do we only have 4 or 5 reconstructions of sea level from this huge database? Because it is a complete mess and it takes a huge amount of human effort in sorting and fixing just to put together a simple database of long-lived stations, or 400 stations, let alone 1207 stations.
I’ve been doing climate database work for a long time and have learned a few shortcut methods that made it at least possible to get an average sea level out of the measurements. Beyond that, I think, you need 2 nerdy grad students with great computer skills working for a summer to put together a useable database.
Between 1807 to 1857, the average of the gauges in the database fell by 100 mms. There seems to be some consistency in the type and number of tide gauges during this time. Did it really fall by that much or was it just subsidence. I think subsidence, Jevrejeva says it was partly real. She must have more nerdy grad students than I have.
After that, a large number of other stations were added and it becomes nearly impossible to track what sea level was doing in my version of the PMSL database. The average went up by 200 mms between 1857 and 1868. This is obviously mostly station sampling.
” I don’t know why they can’t put together a nice, easily trackable, publicly available on-line database where one can trace trends for individual stations over time because they have all the data electronically …”
Then they’d be out of a job. 😛
Why would they want to let us see the data. We only want to find something wrong with it. (TM) 😉
“Did it really fall by that much or was it just subsidence. I think subsidence”
Rising measured sea level could be subsidence, not falling. If it’s PG rebound it should be constant.
“The average went up by 200 mms between 1857 and 1868. This is obviously mostly station sampling.”
Very possible, But I don’t think it is scientifically “obvious”. It needs explaining .
There was a large sampling change in this period of marked “acceleration”. The trouble is they all seem to change, so it’s not just new samples. Also as I noted above, GAIA adjustments also “accelerated” in this period.
Clearly there are unresolved issues in the data and the adjustments so no one should be fitting quadratics to this nonsense and putting it in the abstract of the paper.
I still think it’s good paper in general, very detailed. It will be good to get actual numbers to work on.
“The average went up by 200 mms between 1857 and 1868. This is obviously mostly station sampling.”
Very possible, But I don’t think it is scientifically “obvious”. It needs explaining .
I say that because maybe something did happen. There’s too much “correcting” data to fit preconceived expectations going on already, right across the whole field.
A bias that large , if that’s what it is, must have a detectable cause.
http://www.nationmultimedia.com/home/2010/09/18/national/Bangkok-may-be-uninhabitable-%7Cin-seven-years-30138225.html
Anyone been to Bangkok recently? It should be underwater.
It should be underwater? Under martial law perhaps.
Funny how other catastrophic problems tend to come along much quicker than CGW.
USA will be under martial law before LAX is underwater too.
Bill,
PSMSL data is somewhat difficult to work with, but it can be done. IIRC Willis had something out a few weeks ago with the monthly data extracted into a spreadsheet. I personally took a crack at a reconstruction using R:
https://sites.google.com/site/climateadj/nh-sea-level-reconstruction
The following link provides individual station data:
http://www.psmsl.org/data/obtaining/
NOAA has a nice app:
http://tidesandcurrents.noaa.gov/sltrends/sltrends.shtml
Your most likely correct though, that improvements could be made regarding data analysis tools.
I’d also like to see more reconstructions. Perhaps something akin to BEST using kriging? Maybe Zeke could provide some pointers?
Thanks AJ, lost track of the thread in the last few days, but it looks to me that you have the skills to do a proper analysis of the sea level data.
What would be the most illuminating is the average increase/(decrease) in mms for each tide gauge, for each year, area-weighted. Then this average change by year can just be pinned to start at 6900 mms (in 1807 for example). Just add each year’s average increase in each tide gauge to 6900 (not the average of the tide gauges because we noted above that gauges coming into and then out of the database are distorting the numbers, we just want the change in each by year).
One is going to find lots of annual variability but just having a long properly done time series will allow one to answer the questions about acceleration, whether the satellites are accurate.
The skeptic community really needs this to be done because of the state of the PMSL database. We have to rely on these reconstructions and satellite measurements from climate scientists that have pressures on them to report results in a certain way. I just want an objective analysis done on sea level change because it is very important.
I’ve made this request to about 3 other people which appeared to have the skills to do the job, but nothing has come out of it yet.
Bill, the method you describe is similar to what I used. It was only for the NH after 1930 due to coverage issues. If you scroll down to the bottom, you’ll see my calculated results.
https://sites.google.com/site/climateadj/nh-sea-level-reconstruction