RealClimate logo


Sealevelgate

Filed under: — stefan @ 11 March 2010 - (Italian)

Imagine this. In its latest report, the IPCC has predicted up to 3 meters of sea level rise by the end of this century. But “climate sceptics” websites were quick to reveal a few problems (or “tricks”, as they called it).

First, although the temperature scenarios of IPCC project a maximum warming of 6.4 ºC (Table SPM3), the upper limit of sea level rise has been computed assuming a warming of 7.6 ºC. Second, the IPCC chose to compute sea level rise up to the year 2105 rather than 2100 – just to add that extra bit of alarmism. Worse, the IPCC report shows that over the past 40 years, sea level has in fact risen 50% less than predicted by its models – yet these same models are used uncorrected to predict the future! And finally, the future projections assume a massive ice sheet decay which is rather at odds with past ice sheet behaviour.

Some scientists within IPCC warned early that all this could lead to a credibility problem, but the IPCC decided to go ahead anyway.

Now, the blogosphere and their great media amplifiers are up in arms. Heads must roll!

Unthinkable? Indeed. I am convinced that IPCC would never have done this.


The North Sea (see Stefan’s photostream on Flickr)

But here is what actually did happen.

In its latest report, the IPCC has predicted up to 59 cm of sea level rise by the end of this century. But realclimate soon revealed a few problems.

First, although the temperature scenarios of IPCC project a maximum warming of 6.4 ºC (Table SPM3), the upper limit of sea level rise has been computed for a warming of only 5.2 ºC – which reduced the estimate by about 15 cm. Second, the IPCC chose to compute sea level rise up to the year 2095 rather than 2100 – just to cut off another 5 cm. Worse, the IPCC report shows that over the past 40 years, sea level has in fact risen 50% more than predicted by its models – yet these same models are used uncorrected to predict the future! And finally, the future projections assume that the Antarctic ice sheet gains mass, thus lowering sea level, rather at odds with past ice sheet behaviour.**

Some scientists within IPCC warned early that all this could lead to a credibility problem, but the IPCC decided to go ahead anyway.

Nobody cared about this.

I mention this because there is a lesson in it. IPCC would never have published an implausibly high 3 meter upper limit like this, but it did not hesitate with the implausibly low 59 cm. That is because within the IPCC culture, being “alarmist” is bad and being “conservative” (i.e. underestimating the potential severity of things) is good.

Note that this culture is the opposite of “erring on the safe side” (assuming it is better to have overestimated the problem and made the transition to a low-carbon society a little earlier than needed, rather than to have underestimated it and sunk coastal cities and entire island nations). Just to avoid any misunderstandings here: I am squarely against exaggerating climate change to “err on the safe side”. I am deeply convinced that scientists must avoid erring on any side, they must always give the most balanced assessment they are capable of (and that is why I have often spoken up against “alarmist” exaggeration of climate science, see e.g. here and here).

Why do I find this IPCC problem far worse than the Himalaya error? Because it is not a slip-up by a Working Group 2 author who failed to properly follow procedures and cited an unreliable source. Rather, this is the result of intensive deliberations by Working Group 1 climate experts. Unlike the Himalaya mistake, this is one of the central predictions of IPCC, prominently discussed in the Summary for Policy Makers. What went wrong in this case needs to be carefully looked at when considering future improvements to the IPCC process.

And let’s see whether we learn another lesson here, this time about society and the media. Will this evidence for an underestimation of the climate problem by IPCC, presented by an IPCC lead author who studies sea level, be just as widely reported and discussed as, say, faulty claims by a blogger about “Amazongate”?

p.s. Recent sea level results. A number of broadly based assessments have appeared since the last IPCC report, which all conclude that global sea level rise by the year 2100 could exceed one meter: The assessment of the Dutch Delta Commission, the Synthesis Report of the Copenhagen Climate Congress, the Copenhagen Diagnosis report as well as the SCAR report on Antarctic Climate Change. This is also the conclusion of a number of recent peer-reviewed papers: Rahmstorf 2007, Horton et al. 2008, Pfeffer et al. 2008, Grinsted et al. 2009, Vermeer and Rahmstorf 2009, Jevrejeva et al. 2010 (in press with GRL). The notable exception – Siddall et al. 2009 – was withdrawn by its authors after we revealed numerical errors on Realclimate. This is a good example of self-correction in science (in stark contrast with the climate sceptics’ practice of endlessly perpetuating false information). Rather bizarrely, Fox News managed to turn this into the headline “More Questions About Validity of Global Warming Theory“.

** About the numbers stated above. Regarding the actual IPCC AR4 numbers, adjust the IPCC upper estimate of 59 cm by adding 15 cm to make it apply to 6.4 ºC warming (not just 5.2 ºC) and 5 cm to make it go up to 2100 (not just 2095). That gives you 79 cm. Add 50% to adjust for the underestimation of past sea level rise and you get 119 cm.
For the hypothetical case at the start of this post, just introduce similar errors in the other direction. Let’s add 31 cm by going up to 7.6 ºC and the year 2105 (in fact that is “conservative” but it gives a nice round number, 150 cm). Now assume you have a model compared to which actual sea level is rising 50% slower (rather 50% faster): now you’re at the 3 meters mentioned above. For details, see The IPCC sea level numbers.


305 Responses to “Sealevelgate”

  1. 301
    Dave Burton says:

    Martin Vermeer #287, from your confusion it sounds like you haven’t looked at the tide gauge records yourself. May I make a suggestion:

    1. Go to my augmented version of NOAA’s GLOSS-LTT tide gauge table, here:
    http://www.burtonsys.com/climate/MSL_global_trendtable1.html

    2. Click on the “Year Range” column header. That will sort the list of tide stations according to the number of years they’ve been in operation.

    3. PgDn to the bottom of the table, to view the tide gauges which have been in operation the longest. (Over 40 of them have been in operation for more then 100 years.)

    4. Click on the station names for these stations to view the Mean Sea Level trend graph for each station.

    5. For each graph, look at the X-axis, and find 1993. Now look at the graph, and see if you can see an uptick of at least +1.2 mm/year starting about then.

    It just isn’t there. The ONLY visible uptick in MSL trend was in the late 1800s. That is true even though CO2 emissions have really taken off, starting in the middle of the 20th century.

    Note that 1.2 mm/year from 1993 to 2006 would add up to 13 x 1.2 = 15.6 mm (1.56 cm, or 0.0156 m). That amount of change would be hard to see at a location like Cuxhaven, Germany, or Helsinki, Finland, where there’s a lot of apparently-random fluctuation in sea level (“noise”). But some other stations, like Sydney and Townsville, Australia, and Korsor, Denmark, have a little enough year-to-year fluctuation that an increase of that magnitude would be noticeable. It just isn’t there.

    Moreover, recall that atmospheric CO2 levels started increasing rather dramatically starting in the 1950s, and global temperatures rose significantly starting in the mid-1970s, and are still not far from their peak of a decade or so ago. So if anthropogenic CO2 causes sea level rise, sea levels should have been rising at an increased rate for quite some time — for at least 30 years, if not more — not just since 1993. So where is the increase?? It just isn’t there.

    I encourage you to take a half hour and just click on the station names, and view the graphs. Look for the uptick.

    In the oldest tide stations you can see an uptick in the late 19th century, but where is the uptick in the 20th? It’s just not there.

    Well, there is one station, with a short tide gauge record, which seems to show a (unique) uptick (perhaps due to well drilling). But the stations with the longest and best data don’t show any 20th century increase in rate of MSL rise.

    Remember, the eustatic (global average) sea level trend is added to local factors, worldwide. So if it changes, it should change everywhere, and be noticeable everywhere that the “noise” isn’t too high. It just isn’t.

    Even Church & White (2006), who claim to have detected a slight 20th century acceleration in rate of MSL rise (0.008 ± 0.008 mm yr-yr), admit that “no 20th century acceleration has previously been detected” by other researchers. What’s more, they claim that the bulk of the acceleration which they detected occurred around 1930(!!). That immediately brings to mind a few questions:

    1. Note that the low end of their Confidence Interval is exactly zero. In other words, even Church & White (the researchers who are most confident that there has actually been an acceleration in MSL rise, cannot quite say with 95% confidence that there has been any acceleration of MSL rise during the 20th century. So why is the RC crowd so confident that there has been? Could it be confirmation bias at work?

    2. Why 1930? Worldwide human CO2 emissions were still low then; they didn’t really take off for another 20 or 30 years. So, even if C&W are right, how would an acceleration which (might have) occurred around 1930 in any way confirmation that increased emissions of anthropogenic CO2 cause increased sea level rise?

    3. If there really was a substantial uptick in rate of MSL rise around 1930 (rather than 1993), then WHERE IS IT IN THE TIDE STATION GRAPHS? Surely in the subsequent 75 years we should be able to see it in the longest tide gauge records, if it were really there? 0.008 mm/yr^2 for 75 years would add up to a difference of 0.6 mm/yr. For many tide stations, that represents a doubling or more of their MSL trend. So look at the graphs. Which ones show a rate of MSL rise at the end of the graph which 0.6 mm/yr higher than in 1930? (Good luck finding them!)

    4. If the amount of acceleration in MSL trend during the 20th century, when CO2 emissions soared, was so low that we can’t see it in the graphs, even after 75 years, then why on earth would we fret that it might become catastrophically large over the next 90 years? Even using C&W’s acceleration figure of 0.008 mm/yr^2, that would only add 0.72 mm/yr to the rate of annual MSL rise by 2010 — a near negligible amount.

    Hunt Janin #296 wrote, “It appears to me as a newcomer to this field that a reasonable, responsible guess about sea level rise is that, barring unforseen ice events, the rise likely be be only about 1 m by 2100.”

    A better guess is 5-20 cm. (10 cm if the current rate of 1.1 mm/yr continues; 20 cm if it doubles; 5 cm if it halves.)

    Of course, I’m talking about a global average. Some locations will see much larger increases, and others will see declines in sea level.

    -Dave

  2. 302
    Completely Fed Up says:

    “A better guess is 5-20 cm. (10 cm if the current rate of 1.1 mm/yr continues; 20 cm if it doubles; 5 cm if it halves.)”

    A better guess would be 80-200cm.

  3. 303
    Dave Burton says:

    First, I need to make a typo correction (#301):

    I wrote:
    “…over the next 90 years? Even using C&W’s acceleration figure of 0.008 mm/yr^2, that would only add 0.72 mm/yr to the rate of annual MSL rise by 2010 — a near negligible amount.”

    That should have been:
    “…over the next 90 years? Even using C&W’s acceleration figure of 0.008 mm/yr^2, that would only add 0.72 mm/yr to the rate of annual MSL rise by 2100 — a near negligible amount.”

    Please forgive the mistake!
     

    Second, despite the report of acceleration in Church & White (2006), there’s a very good reason to expect that sea level rise will not accelerate at 0.008 mm/yr^2 over the next 90 years. The reason is that (according to their latest data) it hasn’t done so over the last 90 years.

    In fact, according to Church & White’s latest data and methodology, the rate of sea level rise actually decelerated at 0.006 mm/yr^2 over the last 90 years for which they have data (1917 to 2007).

    C&W’s methodology for calculating the acceleration is simple. They plot the data, and fit a quadratic curve to it. The acceleration is twice the quadratic coefficient.

    Microsoft Excel can fit quadratics. So I downloaded C&W’s latest data, loaded it into Excel, selected the last 90 years, generated an Excel X-Y “chart,” and fitted a quadratic “trendline” to it. Here’s the graph; click on it to load the spreadsheet. As you can see, the quadratic term is negative, which indicates deceleration:

    Of course, -0.0062 mm/yr^2 is a very small deceleration. If continued for 90 years, it would amount to a decline in rate of global mean sea level rise of only about 0.56 mm/yr. (Likewise, C&W’s reported +0.008 mm/yr^2 20th century acceleration is also very small.)

    Now, I’m not claiming that the rate of sea level rise is actually decelerating. It is possible to cherry-pick starting points to show either acceleration or deceleration of sea level rise during the last century, using Church & White’s data and method. For instance, using 1940 as the starting year shows deceleration, and using 1950 as the starting year shows acceleration. Likewise, choosing the last 100 years (1907 to 2007) yields a very small acceleration, and choosing the last 95 years (1912 to 2007) yields a very small deceleration (though in both cases the amount is so tiny that the quadratic curve appears to be perfectly straight).

    The truth is that the tide gauge data for mean sea level for the last century is simply a straight line plus noise. There simply has been no measurable sustained acceleration or deceleration, regardless of whether you analyze the GMSL-LTT tide gauges (individually or averaged), or Church & White’s latest data.

    However, I have real concerns about the reliability of C&W’s “corrected” data. It indicates a linear trend which is substantially higher than that which the GLOSS-LTT tide gauges show. The reason is probably given in this remarkable admission in paragraph 5 of their paper:

    “An additional spatially uniform field is included in the reconstruction to represent changes in GMSL. Omitting this field results in a much smaller rate of GMSL rise…”

    In other words, they added an ad hoc fudge factor! They didn’t directly say whether that “correction” in rate of GMSL rise was temporally uniform, but if so then it would affect only the linear trend, not the quadratic coefficient. Still, I’d like to know exactly how they processed the data before I completely trust it, even for calculating acceleration & deceleration.

    Dave

  4. 304
    Dave Burton says:

    Well, it appears that I can’t embed images in my comments. So to see the graph of Church & White’s last 90 years of Global Mean Sea Levels, with the Excel-generated quadratic curve fit to it showing deceleration, click here:
    http://i831.photobucket.com/albums/zz231/ncdave4life/church_white_1917-2007_trimmed-1.gif

    To download the spreadsheet, click here:
    http://www.burtonsys.com/climate/church_white_2009_gmsl_90yr.xls

    Dave

  5. 305
    Dave Burton says:

    Another typo correction:

    “GMSL-LTT” should be “GLOSS-LTT” (of course).

    Dave


Switch to our mobile site