The dangers of making assumptions about data

I commend to you this YouTube presentation by Frank Lansner, from October, which explains and updates his 2018 paper, which is unfortunately behind a paywall:

The information presented here appears to me to be potentially very significant in the matter of climate change. In essence Lanser’s team hit on the idea, or else spotted in the data, that the distinction usually made between “land surface temperature” and “ocean surface temperature” is misleading, because many land regions, and therefore temperatures, are greatly affected by air masses moving from the oceans.

Weather stations close to the oceans will essentially be reflecting the temperature changes of the upper ocean. However, the true land temperature, ie that produced by the direct affects of sunshine or global atmospheric temperatures (eg from greenhouse gases) will need to be measured in areas sheltered from the influence of sea temperatures, which in effect means measuring them in inland valleys.

When the team took datasets from across the world, and divided them in this way between ocean- and land-influenced temperatures, two dramatically distinct patterns of climatic temperature change emerged for the last century, which differ markedly from the aggregated, homogenised-data graphs published in most of the literature that leads climate policy.

Essentially, the usual homogenised data produces, if not a hockey stick, then a more or less steady, or slightly rising, temperature, with a great acceleration in the last few decades attributed (of course) to CO2 levels.

Separating out the two types of temperature, however, gives an entirely different picture. The ocean-influenced temperatures show a gradual, and more or less steady, small increase from when records began. It correlates with the massive loss of glacial ice recorded from the end of the eighteenth century on. This likely reflects either the warming of the upper ocean following the end of the little ice age, or perhaps more likely the longer term warming after the end of the ice age. Either way, it predates any possible effect of anthopogenic CO2.

The land temperatures, however, once uncluttered from the influence of the oceans, show a clear sinusoidal pattern with something like a sixty year periodicity: a peak of temperature in the 1920s and 30s (corresponding to historical records and individual memories of excessively hot weather around the world), followed by a cooling thereafter (corresponding, towards the end, with the apocalyptic fears of a new ice age), until the recent upswing, markedly similar to that of the early twentienth century. Since there is little to distinguish this from the earlier warming, nothing warrants pinning it to CO2 levels.

Hence taken individually, neither type of data gives any signal whatsoever of a correlation with rising CO2 levels, and both show evidence of different causes preceding any possible anthpogenic effect.

It is only when the two kinds of data are confused and combined that the earlier land peak is largely ironed out by the lower ocean tempertaures back then, and the illusion is given that unprecedented increase has taken place over the last forty years or so.

This is interesting science, which clearly challenges the whole anthropogenic warming theory by re-examination of the raw data. But the interesting (and saddening) part of the story is that in a number of ways this evidence has been hidden by the pre-existing science, and rendered difficult to see, let alone correct.

In the first place, as Lansner points out, the homogenisation of data has not simply been a question of mixing the different types of signal. Instead, the sheltered valley stations, already by their nature lower in number, have been seen not to match the general pattern, and have therefore often been either excluded, or statistically down-graded, as being anomalous to the “best” data predicted by the theory of anthropogenic warming.

Lansner’s presentation shows how, the more this has been done (or the more a particular study is based close to the oceans) the more the early-century maximum and the mid-century minimum has become invisible.

But by performing this homogenization at a stage prior to writing the articles, and presenting the “corrected” temperatures as if they were data (whilst the raw data languishes in an unpublished database), any alternative interpretation is effectively lost to science.

In some cases, giving even more cause for concern, Lanser shows that the graphs produced in the literature simply cut off the period before the mid-century minimum, thus making the present look unique by omission. This is worrying because it suggests deliberate, rather than inadvertant, bias.

Now, all this is of less importance to the advancement of knowledge if some keen-eyed researcher like Lansner notices a phenomenon, re-examines the original data, and realises another story is hidden there – in this case a story that, if confirmed, might save the world trillions of wasted dollars, millions of deaths from fuel poverty, huge numbers of mental breakdowns amongst teenagers taught to feel doomed, massive exploitation of rare-earth miners in poor nations, mountains of un-recyclable solar panels in future decades… and, of course, the simple fact that living in untruth produces multitudes of unforeseen evils. As we are always told, science advances by learning from its mistakes, so all this can be put right, by science.

But sadly, Landsner begins his presentation by explaining how, in order to produce their paper, they contacted twenty or thirty meteorological sources to obtain the raw temperature data for their re-analysis. The response was, it seems, universally negative. So the paper, and a further one due for publication in the New Year, is far less comprehensive than it would otherwise be.

The refusal to release original data is a recurrent feature of climate science, from “Climategate” onwards, and it is done by scientists themselves, not by politicians, the press, or any other scapegoat. The excuses have ranged from protecting private intellectual property (the same excuse that brings universal condemnation of pharmaceutical companies for suppressing negative research on their drugs) to preventing negative criticism of years of the originators’ scientific work – which is as much as to confess they are interested in their own careers, not in science and, accordingly, not in the public good.

The extreme instance of this attitude is the claim that releasing the data might strengthen the hand of “denialists,” much like those biologists keeping uncertainties over evolutionary theory out of text-books and press-coverage lest creationists’ arms be strengthened. Both positions are religious or political, in a bad sense, in that belief in the theory takes priority over what the data actually shows. If truth favours “denialists,” “creationists” or any other out-group, then that’s the way the cookie crumbles. Give stronger evidence for your own view, or admit you’re wrong.

I’m reminded of Jeremy Corbyn’s words following last week’s complete demolition of his Labour Party in the UK general election: “We won the arguments.” Well yes – I too can win every intellectual argument, provided the criterion of victory is my own belief that I am right. In Corbyn’s case, the likelihood is that the criterion of “winning the argument” is correspondence to Marxist theory, which by definition is correct. As my old boss at the pest control lab used to say, “The people have failed us – we will elect a new people.”

Suppressing information from others to win our arguments, however, is about ideology or personal corruption. It is not just bad science.

It is just bad.

Jon Garvey

About Jon Garvey

Training in medicine (which was my career), social psychology and theology. Interests in most things, but especially the science-faith interface. The rest of my time, though, is spent writing, playing and recording music.
This entry was posted in Creation, Politics and sociology, Science. Bookmark the permalink.

7 Responses to The dangers of making assumptions about data

  1. GD GD says:

    Interesting stuff Jon. We have discussed the uncertainty in climate change models and this adds to that.

    Btw an extract from my book is available at CPS and also a featured review by an eminent academic. If you want to see these let me know and I will provide the links.

    • Jon Garvey Jon Garvey says:

      George, the UK “independent advisor on climate change” (a historian by training!) was on the BBC news this morning criticising the lack of government zeal in phasing out petrol cars, banning fossil fuels for heating, etc.

      Not the least hint of uncertainty there – the world will reach a “tipping point” after the next ten years, after which life on earth will be difficult to sustain. Needless to say, us stabilizing our 1.02% of the world’s CO2 emissions will completely offset the coal-fired pollution of the proliferating Chinese solar-panel industry and the escalating use of your coal in India!

      Please do provide some links to your book extract and the review, as I think some of the Hump’s readers might also be interested to gain some idea, at least, of the practical work being done that could help the environment, once the wheels come off the renewables wagon.

        • Jon Garvey Jon Garvey says:

          Good review, George. Let’s hope your book does get on to a lot of influential desks in governments around the world!

          • GD GD says:

            Thanks Jon – btw I have the paper you referred to. If it important to your post, the authors should not mind if I send you a copy. I would remind you however, that GHG concentrations are rising, and the contention is more on assessing the impact on the global scale. Qualitatively we should be concerned; quantitatively the models are uncertain.

            • Jon Garvey Jon Garvey says:

              That would be nice for future reference at least, George.

              The “greenhouse problem” still, it seems to me, depends entirely on forty-year old guesswork about water-vapour feedback. As far as I can tell from the satellite data, the models are all running far too hot.

              Lansner’s work, if validated, seems to mess up any plausible correlation between CO2 levels and temperatures, because today looks pretty identical to 60 years ago.

              That’s an awful lot of uncertainty on which to make predictions dictating we change the entire economic structure of the world… rather than, for example, spending money on adaptation and cleaning up coal: and employing the wise medical strategy of “watchful waiting.”

              The panic all seems to be about the long promised “tipping point” when runaway warming turns us into Venus. I haven’t seen much evidence that most serious scientists believe that will actually happen, and since CO2 has been much higher than now for most of the period since the Cambrian, there are good reasons for doubting it, it seems to me.

  2. Jon Garvey Jon Garvey says:

    Here are a few quotes extracted from the original Lansner paper, which seem to me either explanatory or scientifically significant.

    For all individual stations, we have aimed to use yearly averaged unadjusted temperature data. During the work, we encountered the problem that numerous data series for individual stations have been regarded as outliers and thus adjusted to resemble data series from neighbor stations. This is a potential problem for the present work since OAS [“land type” JCG] and OAA [“ocean-type” JCG] temperature stations can often be located just a few kilometers apart depending on the topography.

    From discussion:

    The OAS temperature data worldwide reveal the existence of a period from 1920 to 1950 with a temperature over the Earth surface warmer than the OAA data show and resembling temperatures seen in recent decades. We have not found any larger OAS area worldwide with a majority of the stations showing temperature trends in disagreement with this general observation.

    For Greenland, most stations are coastal to some degree, but Bjørk et al. have found that retreat rate for glaciers terminating on land (typically hundreds of kilometers from the ocean and often in territory with hills and mountains giving some shelter) in the decade 2000–2010 was smaller than the retreat rate in the last warm period 1930–1940.17 Thus glaciers on Greenland in more OAS-like conditions do not retreat faster in recent years than they did in the 1930s, whereas the opposite is found for marine-terminating glaciers in OAA conditions.

    From conclusions:

    Around year 1900 – after the little ice age – the oceans had been affected by cold conditions for centuries. The oceans (and thus OAA data) after 1920 responded slowly to the rather sudden strong warming of the Earth 1920–1950 while the warming 1920–1950 was detected well and instantly in OAS data.
    The implication from this is therefore that new and fast-changing heat balances over the Earth are better captured in temperature data when no large volume of water as larger lakes or oceans can act as a heat buffer absorbing and delaying faster changes in the heat balance over the surface. The difference between OAA and OAS temperature data may thus help in determining the effect of internal climate variability which appears to be most significant in the ocean-affected stations.
    In contrast, we would expect the OAS regions to show a temperature signal which was less affected by internal variations in the climate system. The OAS temperature data are therefore best suited data type we have to reflect variations of the heat balance over the Earth.
    Therefore, the lack of warming in the OAS temperature trends after 1950 should be considered when evaluating the climatic effects of changes in the Earth’s atmospheric trace amounts of greenhouse gasses as well as variations in solar conditions.

Leave a Reply