Skip to content

Analysis Of Sea Level Trends At Newlyn

July 21, 2017

By Paul Homewood


I have been playing around with the sea level data at Newlyn, which I published yesterday, in order to try and detect whether the rate of rise is accelerating.

To recap, this is the tide gauge record since 1915:



I have plotted the 10-year trends on an overlapping monthly basis:



At the end of 2016, the trend was 3.23mm/yr, ie between 2007 and 2016. This of course is higher than the trend of 1.83mm recorded over the full history of the gauge.

But does this mean that the rate of rise is actually accelerating?

What the data appears to show is that there have been many other 10-year periods when the rise was as great, or even greater. For instance, the 1920s saw the annual trend rise to 9mm, as did the mid 1940s to early 50s.

These peaks in trend oscillate with other periods when sea levels were actually falling for long periods.

It is well known that sea levels in the mid 20thC were rising as fast as now. But I must admit, this pattern of sharp peaks and troughs is actually quite an eye opener.

What it does emphasise is that we really cannot put draw any meaningful conclusions from the last decade or so.

What also stands out like a sore thumb is the sharp fall in sea levels in the early 1990s, due of course to Pinatubo. It is the natural recovery from this which gives the impression of “acceleration” since. In fact, as the chart shows, the 10-Year trend peaked in 2001 at 8.2mm/year. Since then, it has gradually subsided.

We can justifiably conclude that the current rate of sea level rise at Newlyn is far from unprecedented, and that it has been decelerating since the turn of the century.




In for a penny!!

While I am at it, I thought I would do North Shields as well.

As you can see, we see a very similar pattern of peaks and troughs. As North Shields data goes back to 1895, we can also see an even more dramatic peak culminating in 1915.


  1. quaesoveritas permalink
    July 21, 2017 6:26 pm

    I wonder if the cyclical nature of the overlapping trends may be due to some other factor (e.g. astronomical?). Clearly gravitational influence is very complex.
    The 20 and 30 year overlapping trends are also interesting.

  2. stephen m lord permalink
    July 21, 2017 6:46 pm

    Can you try fitting a quadratic equation to track these ups and downs

  3. Mike Jackson permalink
    July 21, 2017 6:53 pm

    Excellent stuff, Paul!

    It does on the face of it mean that maybe the “3mm a year” figure is not wrong but it puts the whole thing in a context that I don’t think anyone has tried before. The fluctuations would seem to suggest (to me at any rate) that trying to link sea level rise to climate change can only make the climactivists look even more ridiculous.

  4. CheshireRed permalink
    July 21, 2017 7:58 pm

    We’ve been told for years (maybe even 30…) that ‘climate is 30 years, anything less is just weather.’ Why should that rule not be applied now? You should demand alarmists come back in 20 years and then say it’s a 3mm long term trend.

    • duker permalink
      July 21, 2017 10:56 pm

      Tidal data should be even longer than 30 years, as you need 20 years just to allow for the physical variations of the positions of moon, sun and earths rotation. Even producing tide tables for a location used to be such a difficult process they created mechanical tidal calculators.
      There was a famous 10 component tide predictor created by Lord Kelvin who did the early
      work on Fourier analysis of tidal observations, now of course the complexity is ‘hidden’ in the programs used on everyday desk or hand held computers.

      • July 22, 2017 9:48 am

        That’s right. It was Bruce Douglas, a leading oceanographer, who said that nobody should look at trends over period of less than 50 yrs.

        I have run some more graphs this morning and will post them later

  5. Athelstan permalink
    July 21, 2017 10:29 pm

    OT Boss, but thought people might like a growl or two.

    Too many white men and dats der truff……….

    The scholars warn against perpetuating ‘white heteromasculine hegemony.’ Two geologists have warned against citing too many white men and/or too many “established scholars” in scholarly articles, because doing so contributes to “white heteromasculinism” and oppression.

    “oppression” yeah right.

    Link – here

    Er here we go………..and you know wotchamacallit – “man made” warming – it’s all a heteromasculine plan to destroy Amazonia in the West.

    • duker permalink
      July 21, 2017 10:59 pm

      A review of the movie Dunkirk – admittedly it was a US publication- said there werent any women or persons of colour in lead roles

      • Athelstan permalink
        July 22, 2017 6:30 am

        Weren’t enough then and now overload, there’s never a perfect balance…………


      • Mike Jackson permalink
        July 22, 2017 9:08 am

        The French aren’t happy either apparently!

      • dave permalink
        July 23, 2017 7:43 am

        “The French aren’t happy either…”

        A true picture of the French at the time would merely show disintegration and defeatism. Nothing shameful in that. It was Wellington who said “Any troops will break and run.”

        We do tend to forget that the Battle of Britain was largely won by the earlier action of both French Air and Royal Air Forces – in as much as the Luftwaffe lost one-third of their combat effectiveness during the Battle of France, and was badly shaken in its morale.

        Another thing we forget is that in a judgement of utter callous, madness by Churchill a second expeditionary force was immediately sent to France and had to be hurriedly withdrawn. In THIS evacuation 7,000 French and English troops went down in ONE ship. Naturally, this was kept from the people.

    • July 22, 2017 8:30 am

      Where is the world headed when quoting those greats who have gone before is no longer possible for fear of upsetting someone?

      • Athelstan permalink
        July 22, 2017 3:50 pm

        heading for trouble………… because not acknowledging expert analysis dooms those who come after to never moving forwards………….if you always have to start at the beginning…….. it’s always groundhog day for Geo wimin.

  6. July 22, 2017 4:56 am

    Why not decompose by doing a Fourier analysis. Subtract the cyclical component, possibly the 18.6 year lunar cycle plus something else.

    As it stands, your overlapping trends model does not do anything for me.

  7. Athelstan permalink
    July 22, 2017 6:43 am

    SL trends, I think that, the topic is an academic palaver but how bad could it get…………..switching to talking about the weather, who knows?

    From the IMO:

    Tonight at hrs. 21:15 (21.07.2017) a M3.1 earthquake was detected in the northeastern part of the Katla caldera on Mýrdalsjökull, a few aftershocks have followed, but there are no signs of volcanic tremors.

    “no signs of volcanic tremors” – thank the Gods for that small mercy.

    • July 22, 2017 7:32 am

      Yes, there has been signs of incipient activity around Katla recently The last time this volcano erupted around 100 years ago it caused major famines across Europe and some say triggered the French revolution.
      Statistically it is due to erupt sometime soon; but no one knows for sure what will happen.

      • Athelstan permalink
        July 22, 2017 7:35 am

        Indeed, it [Katla] ‘she’s’ overdue and crikey we will all suffer [NH] if and when she blows.

  8. July 22, 2017 7:20 am

    A number of years ago I earned my crust taking tourists for trips up and down the Thames under the bridges. On one occasion as we passed Traitors Gate , the main entrance to the Tower of London I recall an American asking why on earth did these Brits build this gate to be flooded on the high tide?. On another occasion there was a similar comment when we passed by Windsor Castle and it was asked why on earth would the Queen build it right under the Heathrow flight path.
    It’s a funny world.
    However Traitors Gate is, I suggest, a good but rough indication of what has happened to the London sea level over the centuries.

    • July 22, 2017 8:32 am

      Does this take into account the constrictions placed on the Thames by the construction of embankments?

      • Hivemind permalink
        July 22, 2017 12:03 pm

        The south of England is sinking due to rebound from the melting of the glaciers in the north.

      • Athelstan permalink
        July 22, 2017 3:53 pm

        if I lived anywhere near the middle reaches of the Thames, i’d be more concerned by the lack of dredging in the past 25-20 years, though of course tides are a significant factor.

  9. July 22, 2017 8:08 am

    No hockey sticks in sight.

  10. July 22, 2017 8:20 am

    This 2016 report (link below) has a curious view of the ‘Pinatubo effect’ on sea levels. It seems to suggest that because the rate of rise was suppressed by the eruption, we should expect acceleration in the future.

    A recovery period followed by a return to previous rates of rise would seem more logical?

    Mt. Pinatubo Eruption Hid Rate of Sea Level Rise

    • dave permalink
      July 22, 2017 9:18 am

      Maybe the global SEA-BED is RISING? Maybe the tectonic plates are quivering? How would we know? What could we do about it – except step back from the beach a little?

      I declare an interest. I live just above maximum sea-level. The local authorities raised the sea-wall a couple of feet and beefed up the defenses, at reasonable cost, ten years ago. Problem sorted, for me.

  11. dave permalink
    July 22, 2017 9:04 am


    Essentially your model is “time-varying, non-causal, linear regression coefficient.”

    Why not run your data through a program that calculates the evolution of the coefficient?

    Running such a program as a matter of routine on data [should!] also stop all the ad nauseam talk about “pause or non-pause” in RSS data, blah-blah X data, that infests internet blogs.

    More generally, the whole area is known as Flexible Least Squares approaches.
    I am far too old to be bothered with it.

    • July 22, 2017 10:14 am

      I wish I had the programming skills!

      • dave permalink
        July 22, 2017 11:44 am

        I will give it a go myself, then.

        I have the old programme set up somewhere – on an ancient, retired computer. I think It is a C programme running under DOS. This may take a while!

        First, I will watch the rest of the Open.

      • dave permalink
        July 22, 2017 3:50 pm

        OK…inspired by that record 62 low round in a Major…I relearned the Windows proxy for DOS (why did they drop the Edit function in 64-bit versions??!!).. and reread the Manual for VCC, and ran the programme for ANNUAL Newlyn data (three years were recorded wrongly!!).

        The result is that almost all of the variability in trend in the last hundred years disappears. There is neither a cycle in the coefficient on time nor anything more than a very gentle upwards tilt to the trend.

        The annual “underlying” rate of increase was just under 1 mm a year from 1916 to the end of WW2, reaching 1 mm a year in 1947 which is where it stayed until about 1978. Since then it HAS increased to 1.16 mm a year.

        You have probably noticed that these are all less than the 1.84 mm calculated from the monthly data. This is because:

        (a) the outliers in the monthly data bugger up the intercept term;
        (b) VCC uses maximum likelihood in a clever way and applies it in thousands of iterations;
        (c) It is what it is.

        So the signal of “modern” AGW, 1978 to 2016, might be guesstimated as an EXTRA sea-level rise now at Newlyn of 0.16 mm each and every year super-added to the pre-existing “natural” rise of 1 mm a year.

      • dave permalink
        July 22, 2017 3:52 pm

        VCC stands for Variable Coefficients Calculator.

      • dave permalink
        July 23, 2017 4:31 pm

        Having resurrected the program, I applied it to the whole of the UAH monthly LTT anomaly data from January 1979 to June 2017. I included lagged regressors for ENSO episodes and for the two cooling volcanic eruptions in the period since there is clear theoretical reason for treating these regressors as true causal variables.

        The coefficient on the volcanos regressor was significant. A similar volcano now might knock 0.20 C off the anomalies for a year.

        The coefficient on the ENSO index was significant. A small El Nino or La Nina might, respectively, add and knock off 0.10 C for a year or two; and a large one might change the anomalies by 0.20 C (as was probably the case in 1998 and 2016)

        The trend coefficient on Time was about 0.14 C per decade and did not change much over the period – no acceleration or pause. The intercept coefficient on Time varied irregularly by several tenths C over the period.


        The coefficients on time were not statistically significant!

        Now what is that all about? It seems as if there is a conclusion about Trend which is pretty much what Dr Spencer (and others, with THEIR temperature anomaly series, say about trend) and in addition their method says “Significant.”


        Time is not a causal variable, it is merely a metronome. The canned programs which non-statisticians use – at their extreme peril – were designed for causal variables with a single non-stochastic intercept. The VCC program uses the mathematics of state-space in which “doubts” about the meaning of the intercept are quantified, and THIS METHOD NOTICES if
        “the trends” seem little different from the “trends” which random walks do produce when integrated, and applies appropriate wide error estimates.

        To repeat an old point:

        to “note a trend” over the last 40 years in something like UAH data is only to note that it had a net move. If you look at it for ANOTHER forty years and it makes the SAME net move…then you could say, That is not a random thing!

  12. July 22, 2017 10:30 am

    Date: 22/07/17 Roy Spencer

    A face-value reading of the two main studies together results in the conclusion that sea level rise since 1993 has been revised downward.

  13. Malcolm Swinbanks permalink
    July 22, 2017 10:30 am

    The output from calculating a running trend is closely related to bandpass filtering. Least square averaging to form the trend suppresses high frequency, short duration fluctuations, which is a low-pass filtering action. The trend itself is a measure of the average gradient over the period of the trend. Measuring the gradient or derivative of a time-varying function emphasises high frequencies ( the Fourier transform of a time-derivative corresponds to multiplying by the values of the frequencies, thus amplifying high frequencies.)

    So this combination of low-pass filtering and high-pass filtering implicit in a running trend represents a form of band-pass filtering, loosely centred around the duration of the trend.

    The same process can be applied to the original global temperature data, (as defined prior to all the recent changes.) Choosing a 30-year running trend (30 years being the consensus time period necessary to assess “climate”) emphasises a very clear ~ 60-year oscillation. Removing this component from the data leaves a residual rising component which if correlated with CO2 emissions yields much lower climate sensitivity than the exaggerated figures of the IPCC.

    Regretfully, this analysis will no longer yield the same result, since the more recent recent approach to evaluating global temperature has been to iron-out the historical peaks and troughs !

    • dave permalink
      July 22, 2017 5:04 pm

      “…recent approach…iron-out…peaks and troughs.”

      UAH have stayed honest. Their revisions of methodology have been driven by real issues.
      They have little effect on peaks and troughs.

  14. July 22, 2017 5:28 pm

    A question: How is the Sea level data compiled, i.e., what is reference point? Is it a land based mark on a rock or something? My point is that the land itself moves up and down, in some cases quite substantially, and that would affect such readings.

    • dave permalink
      July 23, 2017 3:12 pm

      This is the good old fashioned way, using a tough bit of rock with its feet in the sea:
      From the late,lamented John Daly who lived near this spot.

      At first, one imagines the green line to be the marker, but that is seaweed exposed at low tide. The actual marker is the etched V, as shown in the first picture of the second link. Made in 1841 near Port Arthur on the extreme South-East peninsular of Tasmania where the tidal range is about a meter.

      Tasmania is fairly stable tectonically and is not likely to have experienced geological changes in the last few hundred years. As for glaciation and possible rebound – that is a non-starter:

      Click to access igs_journal_vol03_issue024_pg298-303.pdf

      Since the water of the world’s oceans finds its own level, one really only needs ONE clear counter-example (as is the principle in Pure Mathematics) of unchanged level to bring down a whole structure of speculation and special pleading.

  15. July 24, 2017 10:23 am

    talking about sea levels when the earth itself is largely liquid seems nonsensical and pointless. The continental land masses are moving on the surface of the globe, so what can looking at the water level tell us about climate?

    • dave permalink
      July 24, 2017 11:54 am

      “…tell us about climate>”

      It IS a bit a**backwards, since the climate is quite well monitored in a DIRECT way from space:

      AND we know enough to state that the upward blip in 2016 was largely caused by an ordinary, quasi-cyclical phenomenon, whose positive phase is returned to neutral for the present.

      The “fact” that sea-levels rose from 2011 to 2016 –

      – is telling us that there was an El Nino episode. But we already knew that!

  16. Jack Broughton permalink
    July 25, 2017 6:32 pm

    As a person who avoids the seas as far as possible,(e.g. live in the midlands), can someone who understands please tell me whether mean sea level is a proven indicator of maximum level, and why we use the mean level when the maximum seems to be the dangerous level to humans? Thus, if it were the minimum level that was increasing this would seem to be of little interest other than academic??

Comments are closed.

%d bloggers like this: