Skip to content

It is generally accepted that sea levels increased during the 20thC at a rate of about 185mm or about 7”. Furthermore studies suggest that there was no acceleration in this rate during that time. One of the best known studies was by Bruce Douglas, who produced this graph from Tide Gauge records for 23 geologically stable sites.


Satellite monitoring of sea level, which began in 1993, however, shows that the current rate of increase is about 2.8mm/year.


So, question – has the rate of sea level rise really accelerated in the last two decades. In this study, I will be extending Douglas’s work up to 2011 to find out.


Douglas based his work on tide gauge records from PSMSL, (The Permanent Service for Mean Sea Level), and used the following criteria :

  • Each record should be at least 60 years in length
  • Not be located at collisional plate boundaries
  • At least 80% complete
  • Show reasonable agreement at low frequencies with nearby gauges sampling the same water mass
  • Not be located in regions subject to large post glacial rebound.

    He subsequently identified 23 sites that met these criteria. (These are listed in Appendix A). The tide records at each site were adjusted for any post glacial (isostatic) rebound, i.e any rise or sinking of the land, so as to isolate the eustatic sea level, i.e. the quantity of water in the ocean.

    Of Douglas’s 23 sites, only 12 have full or nearly full records since. (Newlyn’s records go up to 2009, but are included). From these 12 stations, I have reconstructed the Douglas graph.

    A point I will keep emphasising is that I am not trying to establish “the actual amount of sea level rise”, but am looking to quantify “the relative rate of change”. In other words, is the current rate of rise greater (or smaller) than the rise during the last century.

    Bearing this in mind, my reconstruction is not adjusted for isostatic rebound. There are two reasons for this :-

    1) Any such adjustment is adding a certain amount of subjectivity, which is absolutely not necessary.

    2) As the objective is to compare the rate of sea level change between 1900-2000 and 2001-2011, the isostatic factor is irrelevant, as it is, to all intents and purposes, a fixed amount.


    Figure 1 shows the spaghetti graph for each station, which naturally does not tell us a huge deal. However Figure 2 averages all twelve stations together, with a cumulative three year running average plot, exactly as Douglas did.


    Figure 1


    Figure 2

    Two things stand out :-

    1) The new reconstruction indicates a sea level rise from 1900-2000 of 168mm, actually a pretty good fit with the original Douglas calculation of 185mm. The difference between the two datasets can be ascribed to two factors :-

    a) No adjustment for isostatic rebound in the new reconstruction.

    b) The elimination of the 11 stations, which do not have recent records.

    We have already discussed isostatic rebound and this does not affect the trend line of the graph one way or the other.

    As for the reduction in the number of stations, Appendix A shows that there are now no Southern Hemisphere sites in the new dataset. Could this be skewing the average? More on this later, though.

    2) The sea level rise has stuck pretty close to the long term trend (red line), both throughout the record, but more significantly in the last decade. This is our first indication that there has been no acceleration in the trend, at least in these 12 stations. Contrast the last decade, for instance, with the period 1940-60 when levels were consistently rising faster than trend.

    Before we look more closely at the figures, it is worth remembering that the three year average used above combines both the 2011 La Nina and the 2010 El Nino in the 2011 average, thus smoothing out any ENSO variability. (With no such smoothing, the 2011 figure would show a sharp drop.)

    A closer look

    We don’t have to rely on eyeballing the graph in Figure 2. Figure 3 shows the year on year changes in sea level (rather than the cumulative change).

    DOUGLAS UPDATE_htm_m55835242

    Figure 3

    While there is considerable inter annual variability, the 10 year running average indicates no upward trend. But we can actually go one step further, by focusing in on the 10 year average and therefore a much smaller range, as shown in Figure 4.

    DOUGLAS UPDATE_htm_ef9a71a

    Figure 4

    The red line is the trend and actually shows a small decline since 1900, although the last 10 years are slightly higher than trend at 2.15mm/year. However, the 10 year figures are comparing 2011 with 2001, when sea levels were lower than usual, having declined for two years, as Figure 2 illustrates.

    Figure 5 gives a broader perspective by listing the average annual increase to 2011, depending on which start year is selected.

    Start Yearmm/yr increase to 2011

    Figure 5

    Bearing in mind the increase in the 20thC was 1.68mm/yr, there has been nothing unusual in trends since the 1990’s, regardless of which year you compare with. Indeed, the evidence would suggest a declining trend.

    Southern Hemisphere

    As previously mentioned, the reconstruction now includes no Southern Hemisphere sites, with three New Zealand sites (Auckland, Dunedin and Lyttleton) and two Argentine sites (Buenos Aires and Quequen) disappearing because of the lack of recent records. Could the recent trend be biased because of this?

    Although there are now no sites with records back to 1900 in the Southern Hemisphere, there are a few with continuous records since the 1960’s. In New Zealand and Argentina, Wellington and Palermo respectively fit the bill. To these I have added two Australian stations, Port Adelaide and Port Lincoln. There are no such records available in other Southern Hemisphere countries. The sea level changes are plotted in Figure 6 and make interesting reading.

    SOUTH HEMISPHEE_htm_m557ab7ee

    Figure 6

    There is a clear decline in the trend, which the actual figures emphasise. Between 1966 and 2000, sea levels rose by 2.99mm/year, at the four stations averaged together. (1966 is the earliest year that we have records for all four sites). By comparison, between 2000 and 2010, they actually fell by 1.73mm/year. Furthermore, because there are no records for any of these sites in 2011, the plot finishes in 2010, when global sea levels appeared to be higher than normal.

    In other words, the exclusion of Southern Hemisphere stations, far from causing the reconstructed sea level trends to be understated, seems to have had the opposite effect. (For the record the New Zealand and Australian stations show falls since 2000, whilst Palermo shows an increasing trend).

    Let me make this clear. I am not claiming this small sample is representative of the Southern Hemisphere as a whole. But I would claim that it is a reasonable substitution for the stations excluded from the original Douglas study.

    Church & White

    In 2011, John Church and Neil White published their paper, “Sea Level Rise from the Late 19th to the Early 21st Century”. This attempted to reconcile sea level measurements from tide gauges and satellites. They concluded that between 1993 and 2009, sea levels rose by 3.2mm and 2.8mm per annum as measured by satellites and tide gauges respectively, as shown below in Figure 7.


    Figure 7

    Global average sea level from 1990 to 2009 as estimated from the coastal and island sea-level data (blue with one standard deviation uncertainty estimates) and as estimated from the satellite altimeter data from 1993 (red). The satellite and the in situ yearly averaged estimates have the same value in 1993 and the in situ data are zeroed in 1990. The dashed vertical lines indicate the transition from TOPEX Side A to TOPEX Side B, and the commencement of the Jason-1 and OSTM/Jason-2 records

    Two things stand out though.

    1) 1993 is used as the starting point (being the start of the satellite record). However, as Church and White themselves point out :

    “However, the reconstruction indicates there was little net change in sea level from 1990 to 1993, most likely as a result of the volcanic eruption of Mount Pinatubo in 1991. “

    In other words, the trends have been calculated from an abnormally low point in the long term record. (Shades of 1979 and satellite temperatures!). This dip can be seen very well on the original Douglas graph, shown again below.


    2) The exercise stops in 2009, when sea levels were at a cyclical peak, as confirmed by both satellite records and tidal gauges.

    So Church and White are comparing an unusually low point in the cycle with an unusually high one. Of course, their choice of dates was purely circumstantial, but, particularly over such a short period of time, such a choice of dates renders their results pretty meaningless. (Figure 5 shows just how much difference the choice of start dates can make). Satellites suggest a drop of 5mm from 2010 to 2011. Based on this, Church and White’s tidal gauge estimate of 2.8mm/year becomes 2.2mm/year, from 1993 to 2011.

    Is 2011 unrepresentative?

    Both satellites and tide gauges confirm that sea levels fell in 2011 with La Nina. So, is using 2011 as the end year disguising an accelerating trend?

    I would answer this in two ways :-

    1) As previously mentioned, I have opted for a 3 year average, in order to replicate the Douglas exercise. This has conveniently averaged together the 2009/10 El Nino and the 2010/11 La Nina within the 2011 “3 year average”, and consequently giving a robust underlying number. (Over the next two years this number should decline as the 2009 and 2010 figures are removed from the average, assuming, of course, no new El Ninos).

    2) I have also shown 10 year averages, which to some extent dilute and smooth out  changes in a single year. (For interest, if sea levels in 2010 and 2011 are averaged together in the reconstruction, the 2011 figure is increased by about 10mm. If the average annual increases in Figure 5 were calculated against the 2010/11 average, the increase since, for example, 1993 would be 1.56mm/year , instead of 1.00mm).


    In the reconstructed analysis, there is no evidence of an acceleration in the long term rate of sea level rise, which remains at below 2mm/year. Furthermore an analysis of Southern Hemisphere sites suggests a slowing down in the rate. The sample sizes in both cases are small and give limited geographical coverage. Nevertheless, they give a similar coverage to the original Douglas study, which has generally been accepted as giving an accurate assessment of 20th Century rise. (For instance, the IPCC quote a figure of 1.7mm/year).

    The divergence with satellite data can, logically, only be due to one or more of the following factors :-

    1) The original Douglas study is based on an unrepresentative sample or inaccurate records. If this is so, it would suggest that the rise of 1.85mm/year, that Douglas calculated for the 20th Century, is significantly understated.

    2) Satellite measurements are wrong.

    3) The relatively short satellite record is too short to give a accurate long term trend, particularly as it starts at a low point in the cycle.

    4) The exclusion of 11 stations from the reconstruction has reduced the true rate of sea level rise. It would appear, however, hugely unlikely and coincidental that these 11 stations had a rapidly increasing trend, while the other 12 showed no trend change at all. In any event, we have seen that this is not the case in the Southern Hemisphere analysis.

    One final comment. From this study, it appears that the number of reliable tidal gauge sites, with reasonably long and complete records, is on the decline. Is too much reliance being put on satellites? Maybe. But when sea level rise is such an important and controversial topic, I find it both astonishing and rather sad that this is being allowed to happen.