Carol Kirkwood
Chief weather forecast presenter
Sometimes someone comes up to me in the store and says, "I'm organizing a barbecue on Saturday, and you said it would rain," he says.
"And it didn't fall. Why were you wrong?"
Or vice versa, they planned to spend the whole day in the sun, and were greeted by gray, cloudy skies.
Or in March, a parent asks me what the weather will be like on the day of his son's wedding, which is scheduled for September.
These people are always extremely kind, and conversations like this are an integral part of the job of a weather forecast presenter, which I have been doing for three decades and which brings me great joy.
But these people's questions also reveal an interesting fact.
During my career, weather forecasting has improved almost beyond recognition.
Today we can predict weather much more accurately and with much finer detail than when I started working in the mid-1990s.
Liz Bentley, professor of meteorology at the University of Reading in England and director general of the Royal Meteorological Society, says that the forecast for the next 24 hours is accurate more than 90 percent of the time.
However, despite these advances, people still do not have complete confidence in the forecast.
When last summer an online market research firm YouGov asked adults in the United Kingdom (UK) whether they trust the weather forecast, a significant minority, 37 percent, said they "don't trust it very much" or "not at all." (Encouragingly, 61 percent of respondents said they still trust presenters, like me.)
There are many jokes about the weather forecast.
At the opening ceremony of the 2012 Olympic Games in London, a clip from 1987 was also shown, when weather presenter Michael Fish told viewers not to worry, because there would be no hurricane, but just a few hours later a powerful storm hit the south-east of England.
(It turns out that Michael was actually right, as hurricane-force winds were recorded that night, but technically it wasn't a true hurricane.)
However, that event has become synonymous with a failed weather forecast.
So why, with all our knowledge and powerful weather forecasting technology, do some people still think the forecasts are inaccurate?
And are we forecasters really wrong, or is the problem more complex and related to the way we communicate the weather forecast?
High accuracy – and high expectations
Part of the challenge is related to people's expectations, which have increased in today's world, where constant access to information is possible.
Today, we can adjust the temperature of our refrigerator or detect a car malfunction in a second using our smartphones.
So why can't we find out with 14% certainty whether it will rain on our street on Sunday at XNUMX pm - isn't that simpler?
The second part of the challenge is how to summarize and present this wealth of information to viewers.
Meteorology produces a huge amount of data that is difficult to summarize into a clear and concise weather forecast that fits in a TV report or on a mobile app.
This means that, even when we are technically correct, some viewers may still be confused.
The answer, however, lies in the very nature of meteorology.
It is a very delicate science, and any slightest inaccuracy in the data can change the details or the entire forecast.
Every day, forecasters across the British Isles collect data on temperature, wind speed and other parameters from a network of more than 200 weather stations, operated by the National Meteorological Service.
The data is then fed into mathematical models that are processed by extremely powerful machines, so-called supercomputers.
Earlier this year, the National Weather Service unveiled a new supercomputer, replacing physical machines with cloud computing.
The new system will enable "better forecasts and help scientists advance important climate research around the world," the Weather Service says.
But, like any other science, meteorology has weaknesses.
- 'Don't be afraid': BBC erroneously announced a hurricane in Europe and hellish temperatures
- How hurricanes form and why they are so common in America, Mexico and the Caribbean
- Parts of the Sahara under water for the first time in half a century
Chaos Theory: When Time Goes Wrong
The atmosphere is known as a "chaotic system," meaning that even the smallest error, even as small as 0,01 degrees Celsius, in the initial measurements can produce a drastically different result.
"It's called chaos theory," explains Professor Bentley.
"Or the butterfly effect. The analogy is that if a butterfly flaps its wings in Brazil, it can affect the atmosphere across northern Europe six days later."
A particular challenge is weather forecasting for very small geographical areas.
In the 1990s, a weather event had to be larger than about 160 kilometres to be fully observed, but today the weather model used by the Met Office for the whole of the UK can map weather events as large as three kilometres, says Professor Bentley.
But zooming below that size is still difficult, so predicting weather like dense fog, which can cover an area as small as a kilometer, is particularly challenging.
Even with enormous scientific advances, technical problems still sometimes occur, although, fortunately, very rarely.
Last fall, on a weather website BBC Weather For a brief moment, completely impossible data appeared: winds in London faster than 21.000 kilometers per hour, and a temperature of 404 degrees Celsius in Nottingham.
The BBC has apologised for "problems with some of the weather data from the service we get our weather forecasts from".
Data compression problem
The biggest challenge of my job is to translate all this data into a short television report.
"There is no other science that people examine, test and evaluate so much," says Scott Hosking, director of the environmental forecasting sector at the Alan Turing Institute in the UK.
"It's almost as complex as the physics of nuclear fusion, but most of us don't encounter it every day, so we don't have to think about how to present this science to a wide audience."
It's also easy to forget that weather forecasting is just that - forecasting.
Over the years, we have improved a lot in this subtle art of "communicating uncertainty."
Meteorologists today can run up to 50 different models, each with slight variations.
If all of these scenarios point to a similar outcome, meteorologists can be confident that the forecast is accurate.
If they give different results, then their self-confidence is much lower.
That's why you might see a "ten percent chance of rain" for your area on weather apps.
Is it time for a new way of communicating forecasts?
Presenters often face a difficult challenge - how to make the weather forecast as clear and understandable as possible.
Recently the BBC renewed cooperation with the National Meteorological Service, which was officially discontinued in 2018, since when the BBC has been using the Dutch organization's forecasts MeteoGroup.
The aim of the new agreement is to bring together the expertise of both organisations and "turn science into stories", said Tim Davey, director general of the BBC.
Some believe that more creativity needs to be brought into communicating weather forecasts.
Dr Hosking, from the Alan Turing Institute, says presenters should avoid dry data, such as percentages of probability of precipitation, and adopt what he calls a "narrative approach".
Such a style would involve the presenter saying, for example, "What we are seeing now is similar to the situation from a few years ago," something that people could relate to the events they remember.
That's one of the reasons the Weather Service started naming storms in 2015.
However, Professor Bentley believes that numbers sound powerful and that it may be better to provide people with concrete data.
He says that in the United States (US), a percentage probability is given for "everything" - from the chance of rain to the possible temperature range.
"People are used to it," she says.
"Because they get this information all the time, they just understand it."
A new supercomputer for weather forecasting
Weather forecasting could soon change drastically with the development of artificial intelligence.
The application of machine learning in weather forecasting has been advancing rapidly in recent months.
It is often said that meteorologists gain an additional 24 hours of accuracy every ten years, which means that the Meteorological Service can now issue weather warnings up to seven days in advance.
But the artificial intelligence models he developed Google DeepMind they are already accurately predicting the weather up to 15 days in advance, says Hosking.
Earlier this year, a team of researchers from the University of Cambridge unveiled a weather forecasting system that is completely based on artificial intelligence, called Aardvark Weather.
The results were published in a scientific journal. Nature.
While the traditional way of getting weather forecasts requires hours of work on a supercomputer, researchers say Aardvark can be run on a regular desktop computer in just a few minutes.
They also claim that it uses "thousands of times" less computing power and provides more detailed weather forecasts.
They also state that it will improve forecasts for West Africa and other poorer regions (while the best traditional forecasting models are mainly made for Europe and the USA).
"It could be revolutionary, it's really exciting," says Richard Turner, a professor of machine learning at the University of Cambridge, one of the model's creators.
But Professor Bentley also sees weaknesses in AI-based weather forecasting models.
He says they are fed vast amounts of historical data and are trained to recognize patterns, which makes it very difficult to predict events that have not yet happened.
"We are facing new records due to climate change," she says.
"We may experience temperatures of 41 degrees Celsius in the UK."
"But if artificial intelligence keeps looking into the past, it will never predict 41 degrees because we haven't had it yet."
Professor Turner acknowledges that this is a challenge for artificial intelligence models like his, and says his team is working to find a solution.
The "so what?" factor
Analysts believe that weather forecasts will be more detailed in the future.
Instead of just predicting whether it will rain, they will also talk about how the rain will affect your trip or your gardening plans.
Professor Bentley calls this the "so what?" factor.
"Would you put a notification on the app like, 'If you're planning a barbecue, you better have it at lunchtime, because there's a chance you'll get wet in the afternoon'?" she explains.
This coincides with a trend that I myself notice in my work, which is that people increasingly want to understand meteorology as a science.
Viewers are no longer just interested in knowing if there will be a heat wave, they also want to know why.
That's why we publish more content that explains, for example, the physics of the aurora borealis or why the city is bigger due to climate change.
As for artificial intelligence, it could certainly increase the accuracy of the forecast, but there is also a risk of overwhelming viewers with information.
Hosking says that because AI is more flexible and can adapt weather models more quickly, users will soon have access to constantly changing forecasts.
Also, these forecasts could be "much more localized" (perhaps providing data not just for your city, but, as some analysts predict, for your backyard).
The result could be a huge amount of data in applications, which would keep people glued to their smartphones.
And in such a world, it will be even more important for presenters to communicate the weather forecast in a clear and understandable way.
Of course, there are also advantages, especially in terms of much longer-term and more accurate weather forecasts.
Maybe one day, when my mother asks me what the weather will be like on her son's wedding day, scheduled for six months from now, I'll be able to give at least a slightly more precise answer.
Additional reporting: Luk Minc
BBC is in Serbian from now on and on YouTube, follow us HERE.
Follow us on Facebook, Twitter, Instagram, YouTube i Viber. If you have a topic suggestion for us, please contact bbcnasrpskom@bbc.co.uk
- How climate change affects Serbia's economy
- Spectacular sight: A tornado and a rainbow intersect
- Why atmospheric rivers are becoming more destructive
Bonus video:
