by Sam Burton-Weiss
On the morning of the 15th October 1987, BBC weather forecaster Michael Fish presented the forecast for the days ahead, and begun so by delivering perhaps the most infamous gaffe in British TV forecasting history:
“Earlier on today, apparently, a woman rung the BBC and said she heard there was a hurricane on the way… well, if you’re watching, don’t worry, there isn’t!” [1]
He went on to detail definitively how, while some strong winds were expected, there was nothing to be concerned about as “Reports of a hurricane are unfounded.” [1]
That night, a huge storm, now known as the Great Storm of 1987, ripped through south-eastern England, killing 18 people and causing up to £5bn of damage. Such a weather event hadn’t been seen in the region since 1703, with measured wind speeds as high as 185km/h being characteristic of a category-3 hurricane. Earlier in the week, UK Met Office forecasts had predicted this event, but later forecasts rebutted these predictions. [2]
As the unfortunate face of this disaster, Michael Fish was a public figure of ridicule. In an interview many years later, he describes his confusion at being blamed for the historic error. In his words:
“The truth is, of course, it had nothing to do with me at all. The forecast is produced by a computer, and if there’s anything to blame at all it’s the computer.” [3]
On the face of it, it is easy to empathise with his perspective. He was not responsible for failing to predict the storm, nor was he even the final forecaster to do so (the evening TV broadcast was similarly inaccurate in its prediction, much to Fish’s chagrin [3]). So why was he blamed?
The answer to this question lies in the language he used when delivering his forecast.
In 1987, the primary source of data upon which forecasts were based was weather balloons. Their ability to collect data was far less comprehensive than modern satellites and, as an accurate understanding of current atmospheric conditions is required to produce accurate forecasts, this meant that forecasts at the time were even less reliable than they are today. So, when Fish delivered his dismissive, almost derisive, statement, he did so with an assuredness not at all warranted by the unreliable nature of the forecast he was presenting. In short, he failed to communicate the fact that his forecast was laden with epistemic uncertainty. And, with his mischaracterisation being highlighted in the most extreme way possible only hours later, it is unsurprising that he received such a huge public backlash.
In the years since 1987, at least in some part as a consequence of Fish’s famous blunder, several improvements have been made to weather forecasting. There have, of course, been inevitable advances in atmospheric data collection and computer models which have vastly improved forecasts’ accuracy. However, as detailed in 2007 by Ewen McCallum, the Met Office’s chief meteorologist at the time of the Great Storm,
“The other aspect that has been improved is the communication of risk. One of the big lessons from 1987 was how to get your message across.” [4]
Specifically, in the wake of this disaster, the Met Office set up the National Severe Weather Warning Service (NSWWS) to provide a coordinated way of delivering warnings to government, emergency responders and public broadcasters. The well-defined and easily understandable 3-tier warning system clearly conveys the risk associated with a potential extreme weather event, and is updated frequently such that the public, and governmental agencies in charge of protecting their safety, can prepare accordingly. Had such a system existed prior to the 1987 storm, Fish’s statement may not have been so damning as it would have been contextualised by the publicly available NSWWS. Even so, it has been documented that, since 1987, British TV forecasters have also been subjected to a so-called ‘Michael Fish Effect’, whereby they are inclined to predict a worst-case scenario in their forecasts in order to avoid being caught out [5].
Michael Fish is, overall, a little unfortunate to have been so intensely ridiculed for his mistake. After all, the Great Storm of 1987 was an extremely rare weather event the likes of which had not occurred for over 200 years, and there was simply no precedent for how the risk of such an event should be communicated publicly. While the assuredness with which he dismissed the possibility of the severe storm was undeniably his downfall, his position as a TV forecaster, the sole authority on the weather which the public could rely upon at the time, made him an obvious target.
This case does, however, demonstrate that risk communication must always include an acknowledgment of the uncertainties inherent in forecasts. In the context of the ever-worsening climate crisis, the future of our planet and society is communicated through models that link greenhouse gas emissions to global warming and its potential societal impacts. As these models are used to justify governmental climate policies related to emission reduction and adaptation, and with the lives and livelihoods of people around the world at stake, transparency with regards to the uncertainty inherent in their predictions is more crucial now than ever before.
Perhaps it is time, then, that we stop mocking Mr Fish, and instead thank him for providing the perfect case study of how NOT to communicate the likelihood of extreme weather events, around which improvements to forecasting have been made ever since.
[1] Youtube video: BBC Michael Fish 15th October 1987 hurricane forecast full version!
https://www.youtube.com/watch?v=NnxjZ-aFkjs&t=2s
[2] Met Office: Lessons and Legacy of the Great Storm of 1987
https://www.metoffice.gov.uk/about-us/who/our-history/lessons-and-legacy-of-the-great-storm-of-1987
[3] Youtube video: Fred Dineage talks to Michael Fish about the great storm of Oct 16th 1987
https://www.youtube.com/watch?v=U1R3mxu4e7w
[4] BBC News Article: Lessons Learnt from the Great Storm (October 2007)
http://news.bbc.co.uk/2/hi/science/nature/7044050.stm
[5] Yahoo News Article; Wayback Machine (May 2009)
Thanks for introducing us to this very interesting case.
A must-read book for you and anyone who enjoyed reading this blogpost is Gary A. Fine (2007) Authors of the Storm – Meteorologists and the Culture of Prediction: https://press.uchicago.edu/ucp/books/book/chicago/A/bo4343084.html.
Thanks for the recommendation Simöne! I read the blurb, and the idea of social and cultural factors playing a large role in shaping weather forecasting certainly tracks with what I found when researching my article. I’ll be sure to look for the book the next time I’m at the library.
This is a very well written article and I enjoyed reading it. Thanks for the links as well.
In fact, there is another very famous case, in which false communication of risk (in that case somewhat intentionally) resulted in losses: The Aquila Earthquake and following court trial against politicians and scientists. You may see a summary that I prepared at the time: https://www.newton.ac.uk/seminar/12002/
In that case participants of an expert group were held accountable for not communicating the risk appropriately. And while the court trial was later revised, acknowledging that the uncertainty of the occurrence of an earthquake prevents accurate prediction, it is worth looking at the details of the procedure on how to derive the warning and statements for public release.
Oh cool! I’ll watch your seminar when I get the chance and let you know my full thoughts.
Based on your quick summary though, I’ll say that I don’t see how uncertainty can be an excuse for lack of effective communication of risk posed by extreme events. The only exception would be events that are so low-probability that they could not have reasonably been expected to be included in risk evaluations. Though I suppose it’s difficult to draw that line.
Thanks for your response 🙂