A distraction due to errors, misunderstanding and misguided Norwegian statistics

A friend asked me if a discussion paper published on Statistics Norway’s website, ‘To what extent are temperature levels changing due to greenhouse gas emissions?’, was purposely timed for the next climate summit (COP28). I don’t know the answer to his question.

But this discussion paper is problematic for sure. It was, authored by Dagsvik and Moen, and already in its introduction it claims that it is difficult to explain and predict weather and temperature (giving the false impression that it’s more or less impossible). This claim is both misleading and a distraction – weather forecasting has been one of the most important success stories in science. 

Today, the warnings conveyed via YR.no are used all over the world, based on calculations that quantify the effects of physical mechanisms and the way different parts of the atmosphere are connected to each other. In other words, we manage well to forecast the weather and temperature in a good way. 

A common misconception is that weather forecasts and climate calculations are based on statistical models, such as the kind used by statisticians or in economic forecasts.

Presumably Dagsvik and Moen are used to this kind of model, but they seem to be inexperienced with the models used for weather and climate, which on the other hand are based on the laws of physics.

The physics-based models describe how energy flows through the atmosphere and ocean, as well as how the forces from different air masses push against each other.

Dagsvik and Moen seem to be sceptical about the merit of estimating the global mean temperature, and their doubt is based on a misunderstood notion that it should have no basis in physics. The irony is that they are no physicists, but many climate researchers, like myself, have a background in physics. 

The truth, on the contrary to their claim, is that the simplest physical models for both the greenhouse effect and the earth’s energy balance are precisely based on the earth’s average temperature. See for instance Benestad (2016). In other words, the claim shows a lack of knowledge and is a well-known old argument that I commented on Realclimate.org in 2007.

We don’t expect to discern a clear effect of greenhouse gases on local measurements, and mixing up local and global climate is a basis for a common strawman argument. Dagsvik and Moen make this classic mistake when they confuse local temperature measurements with Earth’s global mean temperature. 

It is well known within the meteorological and climatological communities that local temperatures are strongly characterised by spontaneous and chaotic fluctuations, and it is also no problem to find examples where temperature correlates with (almost) anything. This is a well-known flaw called “cherry picking“.

The temperatures can nevertheless show ‘artificial’ variations because the thermometers are not evenly distributed over the earth’s surface. We have examined the effect of the geographical distribution of measurements in a study that was cited by New Scientist in 2019, and we found that the uneven distribution of thermometer measurements can give a misleading impression that the perceived warming rate has been slower than the actual global warming.

Dagsvik and Moen write that they only relied on land temperature. In plain language, this means that the data they used represented less than 30% of the Earth’s surface.

It is well known that 30% of Earth’s surface does not provide a good description of global warming, and it is not unexpected that a more limited sample will have a greater impression of random and natural variations, which apparently indicates that random variations play a greater role.

While temperatures provide a measure of the Earth’s climate, it is even better to use the global sea level, which provides a far more reliable measure. 

The global sea level acts like the mercury in a thermometer because warmer water expands. Furthermore, the volume of the oceans increases from the melting of land ice.

The rise in global sea level is to a far lesser extent influenced by random variations and reflects a global warming that is not characterised by any natural cycles.

Dagsvik and Moen claim that recent research indicates that variations in the sun’s magnetic field are of great importance for long-term fluctuations in solar activity. The figure below shows the sunspot number, which is a common proxy for solar activity, and the global mean temperature. The resemblance is striking (not!).

Since the 17th century, the idea that the Earth’s climate is influenced by sunspots has been commonplace, but over time it has proved difficult to explain the variations and the trend observed over the past centuries. 

In other words, the debate about the role of the sun is old, and the explanation about the importance of sunspots has not stood the test of time, something I have described in the textbook ‘Solar Activity and Earth’s Climate‘ from 2002.

One would have to massage the sunspot and temperature curves well so that they can even resemble each other (this has been attempted many times, but not in a convincing way). 

The most important problem with the sun-climate explanation is that there has been no long-term change in either solar activity or cosmic radiation over the last hundred years that can explain the global warming we are now seeing. In 2013, there was a report in the Guardian about scientific studies that refuted the opinion that solar activity or cosmic radiation should explain climate change.

In 2015, an extensive European research collaboration, COST-TOSCA, produced a handbook that dealt with the question of how solar activity affects the Earth’s climate. None of these studies are mentioned in Dagsvik and Moen’s discussion note, which instead refers to think tanks and well-known climate sceptics.

Dagsvik and Moen purports that according to theory and reconstructed temperature data, climate is affected by cyclical variations in the Earth’s orbit, the Earth’s axis and the planetary orbits of Jupiter, Saturn, Neptune and Uranus.

The Earth’s orbit around the Sun is indeed slightly affected by the gravity of the heavy planets in the Solar System, known as the Milankovitch cycles. But the variation in the planet’s gravitational fields lead to very slow changes over many thousands of years, while the global warming we see today takes ten to a hundred years.

One thing we can learn from the previous cycles and the ice ages that they created, is that the Earth’s climate is sensitive to changes in physical conditions. In other words, the extent to which climate is affected by varying physical conditions, whether it is the sun’s radiation, volcanic eruptions, changes in the Earth’s orbit around the sun, changes in the Earth’s landscape or the chemical composition of the atmosphere (which also includes “climate gases”).

Today, the Earth is under constant surveillance from advanced satellites orbiting the Earth (e.g. Copernicus), measurement programs in the ocean (ARGOS floats) and land-based instruments that can measure all physical conditions that can affect our atmosphere. The results from this monitoring show that it is primarily the man-made increased greenhouse effect that is responsible for the observed global warming.

Dagsvik and Moen try to make the point that it can be difficult to say anything about global warming by comparing the temperature with a so-called ‘fractional Gaussian noise process‘ (FGN), which actually involves a kind of smoothing of a bunch of arbitrary numbers.

When such a framework is applied to local or a limited range of temperature measurements, which are characterised by the chaos laws of the local weather, we also do not expect to see any strong effect of CO₂. On the other hand, we expect to see the effect of greenhouse gases when we look at the temperature averaged over the entire globe.

This debate is not new and there was a similar public discussion about long-term memory in the climate system and trends in 2014 on a Dutch website, Climate Dialogue. Although there was no clarification on that at the time, we can say that the discussion is characterised by tunnel vision that does not take into account all our knowledge and how physics plays a role. I also wrote a post on this topic in 2005 (Naturally trendy?), and even if it is a blog, it could provide some clues for Dagsvik and Moen.

One thing we know for sure is that there is always one or more physical mechanisms behind cycles and variations in the Earth’s climate. Even internal variations such as El Niño, which I did my PhD on in 1997, are driven by physical mechanisms such as Kelvin waves, the Earth’s rotation and the interaction between ocean and atmosphere.

Another thing is the totality of the increase in the global sea level, the increased precipitation, snow and ice melting and permafrost thawing. All this can be explained by physical processes and an enhanced greenhouse effect.

In summary, we can say that FGN is a red herring that takes us back to old beliefs that have not proven to be convincing, mostly because those who advocate this idea cannot account for any credible physical mechanisms that can explain how the global warming is a fluctuation in such a stochastic fashion. For the sake of consistency and the lack of a hidden physical mechanisms, it would imply high climate sensitivity – and guess what – there are some physical conditions with a trend that can match the observations: CO₂. 

We do know that CO2 is a greenhouse gas because we can measure this effect in our laboratories and explain why with the help of quantum physics, which is some of the best knowledge we have in science.

Dagsvik and Moen doubt whether the climate models can be evaluated against historical observations, when the same observations have also been used to calibrate them. It is easy to understand this scepticism, but we must also concentrate in order to keep balance and avoid misunderstandings.

The climate models are adjusted so that they create average temperatures that match the observed average, but are usually not adapted to the observations to reproduce the changes and variations that have occurred. The adjustment is about quantifying the statistical effect of, among other things, cloud formation on climate, within a plausible range.

This calibration (often referred to as “tuning“) is thus misinterpreted in the discussion note as meaning that the models are calibrated against the observations to produce the same variations. There is no blind calibration either, but is based on deeper knowledge of the microphysics of the clouds.

There must be great freedom in science to research various things and to communicate the results obtained. With that freedom, there is also a risk that one misses the mark completely and is left with misconceptions.

Since I started working as a climate researcher at the Meteorological Institute in 1998, I have come across a number of erroneous conclusions about climate change which included, among other things, scientific publications. Our job is to be thorough and verify questionable results. It has taken time.

About ten years ago, I became curious if there were any typical common features of scientific results that our verification showed to be erroneous. That led me, together with several colleagues, to do an analysis which we published and which was picked up by Scientific American ‘Why Climate Skeptics Are Wrong‘.

We discovered that erroneous analyses tend to ignore relevant information and are often written by researchers who have experience in fields other than climate. It is also common that they do not evaluate their methods in a sufficiently good way. I also see these traits in the discussion note of Dagsvik and Moen.

The arguments that Dagsvik and Moen present in their note are old and have already been checked out, and it is a little frustrating that this verification has been completely overlooked. It feels like we start from scratch again because of inexperienced and lazy newcomers. The note can also be interpreted as an insult because it suggests that we climate scientists do not know our subject.

We assume that people do thorough preparatory work to ensure that the analysis does not completely derail. It is wise to be humble when moving into an unknown professional field, and especially if the results are conceived as being controversial. I hope Dagsvik and Moen are interested in my feedback, especially because they presented their claims through a discussion note.

They show a lack of knowledge and reveal, among other things, that they are not up to speed as they used an old temperature curve (HadCRUT3) which we have long since abandoned. Newer curves (e.g. HadCRUT5, which is also more easily available) are based on a more extensive database and agree better with other data sources (e.g. from satellite measurements and archived weather forecasts).

The lack of knowledge also becomes apparent when Dagsvik and Moen claim that it is difficult to get an overview of the scientific consensus when it comes to climate change. They apparently have not realised that both the UN climate panel and the professional communities in the USA and Europe have written a number of reports to do exactly that (IPCC AR6 2021, NCA 2017, ESOTC2022).

The discussion paper also creates a naïve, superficial perception when it both repeats a number of old well-known climate myths that have been refuted and has a selective reference list with a dominant element of sources that are not scientific and often from climate-sceptic think tanks such as the Heartland Institute and GWPF. Only a small part of their references are from scientific journals that focus on climate research.

If Dagsvik and Moen had better knowledge about their chosen topics, they would have referred to other relevant information, not been so selective and been more critical of their own sources.

One challenge is that erroneous research results are costly. We have to spend time and resources to review them and explain why they are wrong. This cost is something we have to reckon with, but it is a waste of time if it is really obvious that the analysis was wrong in the first place.

We generally have limited resources for obtaining and disseminating climate knowledge that will protect lives and values. Thus we should preferably spend our time on more sensible things than debunking frivolous claims that surprisingly come from Statistics Norway.

We know from previous experience that it is likely that such cases will be used uncritically and selectively for political and economic goals. Hence my friend’s question about COP28.

Books have been written about how disinformation is deliberately used to cast doubt on climate change and prevent restrictions on the use of fossil energy sources (e.g. The Heat is On, Climate cover-up, The Merchants of Doubt), and during the UN’s last climate meeting, California’s governor, Gavin Newsom, explained how interests from the fossil fuel industry work hard to prevent emission cuts and undermine climate science.

This issue is therefore not only about academic freedom, because doubt is often used to undermine the climate issue quite deliberately by groups with financial interests.

In August, the global mean temperature was close to 1.5°C above pre-industrial times in August, according to Copernicus. In other words, we are approaching the limits defined in the Paris Agreement, even if it is only a monthly measure, as opposed to a permanent increase. 

It is very dramatic because we continue to emit CO₂ and other greenhouse gases that intensify the greenhouse effect and create further climate change. We are increasingly receiving shocking reports about how extreme weather creates disasters and tragedies.

In this drama, the Statistics Norway memo creates a misleading impression based on old myths while at the same time dismissing well-established knowledge. In the light of hindsight, we can say that there are many possibilities for publishing a working note and that it is not necessary for Statistics Norway to put its stamp on it.

My experience is that it is depressingly rare that people become wiser from debates about the causes of climate change, but society rather becomes more polarised.

References

1. 
   R.E. Benestad, “A mental picture of the greenhouse effect”, Theoretical and Applied Climatology, vol. 128, pp. 679-688, 2016. http://dx.doi.org/10.1007/s00704-016-1732-y