How is climate change affecting the interpretation of temperature records?

Dear Carlotta Gotti

Temperature records are similar to other record-breaking events, such as those in sport: these are absolute rankings dating back to the start of record-keeping in that area. They can therefore be surpassed time and again, meaning that the month currently classified as the ‘fifth warmest January’ may, in a few years’ time, have slipped to seventh or eighth place.

Reliable measurements for 160 years

Several factors need to be taken into account here: Firstly, these figures initially refer to the start of reliable meteorological measurements. In Switzerland, these measurements date back to the 18th century and have been carried out since 1863/1864 by an official, state-run network – the predecessor organisation of today’s MeteoSwiss. In other countries, the start of these regular weather measurements may have been earlier or later. What is clear, however, is that most of these record-breaking months or seasons have occurred during the last 25 years, a period characterised by progressive global warming.

A cluster of records over the last 25 years

Secondly, this also means that whilst a single year with record values says little about climate change, a cluster of such warmest, third-warmest or fifth-warmest months naturally does. This is because, in scientific terms, climate change is primarily reconstructed by comparing the averages of longer periods (usually 30-year periods, known as climatological normals, or CLINOs for short). So, if – as has been the case in recent decades – one monthly or seasonal record temperature follows another, this automatically raises the average for the entire 30-year period.

Records depending on the observation region

Thirdly, when reading news reports about such record values, one must always bear in mind which observation region is being referred to. Is it the ‘fifth-warmest January’ purely within Switzerland, or in Europe, or globally? Unfortunately, this is sometimes not communicated clearly enough in the media (partly because these articles are subject to very strict character limits). Consequently, it may be the case that, to you personally, a particular month or season does not seem particularly hot, cold, wet or dry, but that the figures recorded record values in many other locations within the observation region.

Side note: flood records

One more point regarding another area of climate history where the accumulation of record values brings further consequences: In hydrology – that is, the study of watercourses and thus also of high and low water levels – there are also frequent reports of record values, such as the highest ever recorded flood level on River X or a ‘once-in-a-century flood’ on River Y.

Probability of recurrence and flood protection

Here, the increasing frequency of extreme events also has implications for what is defined as the statistical probability of recurrence. In hydrology, for example, a flood with a so-called HQ value of 100 means that, statistically speaking, such a flood is likely to occur once every 100 years in that river system.

Flood protection, for example in the form of bank reinforcement or the protection of public buildings, must often comply precisely with this HQ100 value in accordance with building regulations. This means that the protective measures would provide safety against all events up to and including an HQ100 event.

Increased frequency of ‘once-in-a-century’ events

If, as a result of global warming and the resulting higher levels of precipitation during extreme events (because warmer air can hold more moisture), a ‘once-in-a-century flood’ (i.e. one with an HQ100 rating in hydrological terminology) occur multiple times within a relatively short period, this often leads to astonishment amongst the general public.

The importance of historical flood research

However, as the calculation of what exactly constitutes a flood with a statistical return period of 100 years is based on long-term average values, the many extreme events of recent decades have caused a shift in the average: A flood previously classified as an HQ100 would then have to be reclassified, for example, as an HQ80 event due to the higher average values. Conversely, flood defences and other safety measures would need to be adapted to the ‘new’, even more severe HQ100 flood events – that is, to events which, statistically speaking, used to occur, for example, every 120–150 years. Historical flood research can therefore help to adapt the structurally prescribed protective measures to the statistically probable hazard scenario.