What are the chances of another hot summer like 2018?

The ‘UNSEEN’ approach

We know that the atmosphere is a chaotic system – with the flap of a butterfly’s wings, very different weather outcomes can arise from near-identical starting conditions. One of the questions that follows is how the summer temperatures that we have witnessed over the years could have looked different, or in other words: have we been lucky in the extreme temperatures that have affected the UK? Could conditions have been more severe? We only have one version of how the climate evolved in reality but what if we could simulate lots of different versions? The temperature extremes produced by these alternative versions would enable a better representation of the frequency of plausible extreme events in today’s climate.

This is the idea behind ‘UNSEEN’, UNprecedented Simulated Extremes using ENsembles. A large set of climate model simulations – or ensemble – provides the different versions of the climate. They simulate many more virtual summers than are available from observations alone, and some of these have higher temperatures across the UK than in 2018. These are conditions that are possible in the current climate but have not yet been experienced. The set of simulated summers is used to quantify the chance of seeing unprecedented high temperatures, while also enabling us to estimate how hot high-end extremes could be in today’s climate.

Estimating the chance of exceeding summer 2018 temperatures in the UK

The likelihood of exceeding the UK’s record breaking summer 2018 mean temperature of 15.76°C is estimated to be ~11% each year (Figure 1a), and much warmer temperatures are possible. An extreme of a 1 in 100-year magnitude would bring an average summer temperature of around 1°C greater than 2018. Against the backdrop of a warming climate, the chance of seeing temperatures higher than 2018 has increased rapidly over the last decades, from <1% in the 1960s to ~10% in the 2010s (Figure 1b).

Figure 1. Applying UNSEEN to UK summer temperatures. a) UNSEEN estimate of likelihood of exceeding summer 2018, with central estimate in dark green and 95% confidence interval obtained via bootstrap resampling of the ensemble in pale green. Dotted line indicates the magnitude, relative to summer 2018, of a one in 100-year event. b) Change in likelihood of exceeding 2018 over time, calculated each year from the ensemble adjusted to represent that year’s climate. 95% confidence intervals calculated as before.

The dynamics of hot summers in the UK

An important benefit of the UNSEEN dynamical approach over simple statistical estimates is that the model simulations provide us with the wider climate conditions connected to the extreme of interest, which enables insight into the dynamics behind hot UK summers. By sub-selecting the model set of 4720 simulated summers based upon high UK temperatures, we get the average atmospheric circulation signature of hot summers (Figure 2).

Figure 2. Sub-selecting the ensemble to characterise anomalous atmospheric circulation. Composite 500 hPa geopotential height anomalies (m) selected on warmest 5% (236) simulated UK summers. Stippling marks the 90% confidence level, obtained via a t-test.

The atmospheric circulation anomalies marked by red shading over the UK and northern Europe are indicative of warm temperatures at the surface. There is a clear wave pattern of alternating high (red) and low (blue) circulation anomalies reaching around much of the northern hemisphere, suggesting that there is influence on the UK from remote regions. The study describes possible influence from both the tropics and the Arctic, which projects onto the summer circulation over the north Atlantic and northern Europe, contributing to high temperatures over the UK.

Applying UNSEEN to other impact-relevant extremes

UNSEEN has been applied in several other cases, and research in this area is ongoing for impact-relevant extremes in the UK and around the world. Some examples follow:

UK applications of UNSEEN

• Winter rainfall: virtual observations of around 100 times more data than we have in the observed record (from 1981-2015) have been analysed, including winter months that were wetter than the record-wettest January in 2014. Research shows there is now a 7% chance of record monthly rainfall over South-East England each winter and a 1% chance of a very large increase (15-35%) beyond the current historical record (2).
• National Flood Resilience Review: the UK Government National Flood Resilience Review (2016) (3), aims to understand the risks of river and coastal flooding from extreme weather over the next 10 yrs. The inclusion of UNSEEN adds an important component to the assessment, allowing enhanced estimates of record rainfall and beyond those based only on historical observed records of river levels and flows during floods.