Recently, two disparate strands of my work produced publications within a week of each other.
Curiously they both concerned one of the commonest measurements made on Earth today – the measurement of air temperature.
- One of the papers was the result of a humbling discovery I made last year concerning a common source of error in air temperature measurements. (Link to open access paper)
- On the other paper I was just one amongst 17 authors calling for the establishment of global reference network to monitor the climate. My guess is that most people imagine such a network already exists – but it doesn’t! (Link to open access paper)
I am writing this article because I was struck by the contrasting styles of these papers: one describing an arcane experimental detail; and the other proposing a global inter-governmental initiative.
And yet the aim of both papers was identical: to improve measurement so that we can more clearly see what is happening in the world.
Paper 1
In the middle of 2018 I was experimenting with a new device for measuring air temperature by measuring the speed of sound in air.
It’s an ingenious device, but it obviously needed to be checked. We had previously carried out tests inside environmental chambers, but the temperature stability and uniformity inside the chambers was not as good as we had hoped for.
So we decided to test the device in one of NPL’s dimensional laboratories. In these laboratories, there is a gentle, uniform flow of air from ceiling to floor, and the temperature is stable to within a hundredth of a degree Celsius (0.01 °C) indefinitely.
However, when I tried to measure the temperature of the air using conventional temperature sensors I got widely differing answers – varying by a quarter of a degree depending on where I placed the thermometer. I felt utterly depressed and humiliated.
Eventually I realised what the problem was. This involved stopping. Thinking carefully. And talking with colleagues. It was a classic case of eliminating the impossible leaving only the improbable.
After believing I understood the effect, I devised a simple experiment to test my understanding – a photograph of the apparatus is shown below.

The apparatus consisted of a set of stainless steel tubes held in a clamp stand. It was almost certainly the cheapest experiment I have ever conducted.
I placed the tubes in the laboratory, exposed to the downward air flow, and left them for several hours to equilibrate with air.
Prior to this experience, I would have bet serious amounts of money on the ‘fact’ that all these tubes would be at the same temperature. My insight had led me to question this assumption.
And my insight was correct. Every one of the tubes was at a different temperature and none of them were at the temperature of the air! The temperature of the tubes depended on:
- the brightness of the lights in the room – which was understandable but a larger effect than I expected, and
- the diameter of the tubes – which was the truly surprising result.

I was shocked. But although the reason for this is not obvious, it is also not complicated to understand.
When air flows air around a cylindrical (or spherical) sensor only a very small amount of air actually makes contact with the sensor.
Air reaching the sensor first is stopped (it ‘stagnates’ to use the jargon). At this point heat exchange is very effective. But this same air is then forced to flow around the sensor in a ‘boundary layer’ which effectively insulates the sensor from the rest of the air.

For small sensors, the sensor acquires a temperature close to that of the air. But the air is surprisingly ineffective at changing the temperature of larger sensors.
The effect matters in two quite distinct realms.
Metrology
In metrology – the science of measurement – it transpires that knowledge of the temperature of the air is important for the most accurate length measurements.
This is because we measure the dimensions of objects in terms of the wavelength of light, and this wavelength is slightly affected by the temperature of the air through which the light passes.
In a dimensional laboratory such as the one illustrated below, the thermometer will indicate a temperature which is:
- different from the temperature of artefacts placed in the room, and
- different from the temperature of the air.

Unless the effect is accounted for – which it generally isn’t – then length measurements will be slightly incorrect.
Climatology
The effect is also important in climatology. If a sensor is changed in a meteorological station people check that the sensor is calibrated, but they rarely record its diameter.
If a calibrated sensor is replaced by another calibrated sensor with a different diameter, then there will be a systematic effect on the temperatures recorded by the station. Such effects won’t matter for weather forecasting, but they will matter for people using the stations for a climate record.
And that brings me to Paper 2
Paper 2

When we see graphs of ‘global temperatures’ over time, many people assume that the data is derived from satellites or some ‘high-tech’ network of sensors. Not so.
The ‘surface’ temperature of the Earth is generally estimated in two quite distinct parts – sea surface temperature and land surface temperature. But both these terms are slight misnomers.
Considering just the land measurements, the actual temperature measured is the air temperature above the land surface. In the jargon, the measurement is called LSAT – the Land Surface Air Temperature.
LSAT is the temperature which human beings experience and satellites can’t measure it.
LSAT data is extracted from temperature measurements made in thousands of meteorological stations around the world. We have data records from some stations extending back for 150 years.
However, it is well known that data is less than ideal: it is biased and unrepresentative in many ways.
The effect described in Paper 1 is just one of many such biases which have been extensively studied. And scientists have devised many ways to check that the overall trend they have extracted – what we now call global warming – is real.
Nonetheless. It is slightly shocking that a global network of stations designed specifically with the aim of climate monitoring does not exist.
And that is what we were calling for in Paper 2. Such a climate network would consist of less than 200 stations world-wide and cost less than a modest satellite launch. But it would add confidence to the measurements extracted from meteorological stations.
Perhaps the most important reason for creating such a network is that we don’t know how meteorological technology will evolve over the coming century.
Over the last century, the technology has remained reasonably stable. But it is quite possible that the nature of data acquisition for meteorological applications will change in ways we cannot anticipate.
It seems prudent to me that we establish a global climate reference network as soon as possible.
References
Paper 1
Air temperature sensors: dependence of radiative errors on sensor diameter in precision metrology and meteorology
Michael de Podesta, Stephanie Bell and Robin Underwood
Published 28 February 2018
Metrologia, Volume 55, Number 2 https://doi.org/10.1088/1681-7575/aaaa52
Paper 2
Towards a global land surface climate fiducial reference measurements network
P. W. Thorne, H. J. Diamond, B. Goodison , S. Harrigan , Z. Hausfather , N. B. Ingleby , P. D. Jones ,J. H. Lawrimore , D. H. Lister , A. Merlone , T. Oakley , M. Palecki , T. C. Peterson , M. de Podesta , C. Tassone , V. Venema, K. M. Willett
Published: 1 March 2018
Int. J. Climatol 2018;1–15. https://doi.org/10.1002/joc.5458
Like this:
Like Loading...