How can you accurately calculate the chance of extreme weather, when there’s little precedence on which to base predictions? Here’s where an innovative research technique using the Met Office’s supercomputer is breaking new ground.
The UK has been hit with a series of extreme rainfall events in recent years. The winter of 2013/14 brought storms with record rainfall and flooding in many regions. In December 2015, Storm Desmond caused widespread flooding and storm damage in the North West, prompting the government to launch the National Flood Resilience Review.
As part of this review, the government asked the Met Office to assess the likelihood of a worstcase scenario for extreme rainfall over the next 10 years. However, predicting extreme weather is not straightforward. It is, by its very nature, a rare occurrence, which means there are very few observations on which to base research.
Not only that, historical observations from before 1980 may no longer be relevant, since the climate has changed considerably. To overcome the lack of real data, the Met Office did something entirely new.
A game changer
Using the supercomputer, Met Office scientists were able to simulate thousands of possible weather scenarios and therefore create a considerable body of virtual observations. These simulations were based around the Met Office decadal prediction system where large ensembles are initialised with observations and allowed to evolve freely. In the end, one hundred times more possible simulations of the current climate than is available from real observations.
The researchers have named this new research method Unprecedented Simulated Extremes using ENsembles, or UNSEEN, to emphasise that the analysis anticipates possible events that are yet to be seen.
This is the first time that model simulations have been used in this way – and the results were striking. Analysing these simulated events showed there is a 7% risk of record monthly rainfall in south-east England in any given winter. When other regions of England and Wales are also considered, this increases to a 34% chance.
In other words, even with the current climate, it is likely there will be one or more monthly regional record extreme rainfall events in the coming decade.
It can also be used to highlight month where certain regions have so far by chance been ‘lucky’ not to have experienced a more extreme rainfall event so far. For example, in the South East, the model simulations show a record rainfall is most likely to take place in December.
Moreover, as Vikki Thompson, Climate Dynamics Scientist at the Met Office, explains, “Some of the climate dynamics we’ve seen in the research are dynamics we haven’t seen in the real world. Yet when we assessed them, we found they are plausible.”
In one simulated event, for example, the atmospheric conditions across the North Atlantic were different to what would normally be expected for a winter of heavy rainfall in the UK. On further examination, it transpired that the rainfall was being caused by moist air coming from the tropics. This would cause a warm but much wetter winter.
A better handle on risk
The applications of this ground-breaking research are significant. Now, the Met Office is able to specify how much rainfall a flood mitigation strategy would need to prepare for.
“Whilst statistical methods exist using observations alone to estimate the risk of record rainfall,” says Vikki, “our new technique allows us to give a more precise estimate for a one in- 100, or one-in-20 scenario.” For contingency planners and engineers building a bridge or flood defences, for example, this new information is crucial.
However, the applications go far beyond extreme rainfall. Since this new methodology is based on a global model, it has multiple applications around the world.
Taking the research further
The Met Office is already applying this new methodology to assess the likelihood of other extreme weather events. Met Office researchers are currently assessing the likelihood of a winter even colder than that of 2010 in the UK.
Another significant project is investigating the likelihood of the maize crop in China and the United States failing simultaneously as a result of extreme heat and dry conditions – an event that would have a catastrophic impact on the global food market.
It is a global model and methodology with almost endless possibilities – and it could be an essential tool in helping countries around the world implement effective contingency strategies for the future.
Read the profile on Vikki Thompson, Climate Dynamics Scientist.