Skip to content

The UK Met Office’s Model Muddle


The Met Offices’s model-based rainfall forecasts have not stood up to empirical tests, and do not seem to give better advice than observational records.

In July this year, scientists from the UK Meteorological Office released a new study estimating the risk of UK regions suffering a record-breaking monthly rainfall between the months of October and March.

Vikki Thompson, Met Office: We want to understand the chance of these extreme rainfall events in our current climate and the likelihood of exceeding the rainfall levels that we have seen so far. We have used the new Met Office supercomputer to run many simulations of the climate, using a global climate model.

But rather than producing an estimate of the risk based on the observational record, they claimed that computer models provided a far more useful and accurate result to inform policymakers about what to prepare for.

A Met Office infographic explaining the work even shows the historical data being thrown into a waste bin.

Vikki Thompson, Met Office: This is the first time that we have used what we have called the UNSEEN method, which stands for Unprecedented, Simulated, Extremes, Using Ensembles.

To get their estimate, they ran thousands of climate simulations and found that there was a seven per cent risk of a record monthly rainfall in Southeast England in any given winter.

But this result does not give any better information than what could be obtained using a piece of paper, rather than a £97 million super computer.

The seven per cent chance of a month between October and March exceeding the record for that month in any year is equivalent to a new record being set every 86 months.

New monthly records were set twice in the 216 October-March months between 1980 and 2015. Therefore the ‘risk’ of a new record for monthly rainfall is 5.5% per year, according to the record.

Although the researchers claim that the climate has changed, the results for the two preceding 36 year periods in the 20th Century record suggest similar results.

Between 1944 and 1979, there were three new record monthly rainfalls – an 8.7 per cent chance of any month in a year exceeding the existing record.

And between 1908 and 1943, there were 4 record events – a risk of 14.5%.

The risk of monthly rainfall exceeding the monthly record in the Southeast of England has not risen, contrary to many claims..

The existing monthly rainfall records for October, November and December were set in 1987, 1939 and 1915, and the records for February and March were set in 1951 and 1947.

Therefore, the Met Office computer models do not give any more reliable insight than the historical data.

And recent history suggests that the Met Office is over-reliant on computer modelling.

CH4 News, April 2012. Britain is facing its most severe water shortage 1976. Now, another seventeen counties have been added to the official English drought zone.

This new research follows the Met Office’s prediction of the continued drought conditions that affected much of the UK in late 2013.

January 2014 was the wettest on record for that month. Rain and floods caused widespread disruption and flooding.

Only a few years earlier, the Met Office predicted that the summers of 2008 and 2009, would bring average temperatures, and drier than normal conditions. But far from the ‘barbecue summer’ the British public had been told to expect, both summers were cold and wet.                    

And Met Office’s winter predictions have been no better than its summer forecasts. In 2008, they predicted the trend of mild winters would continue. But the winter of 2008/9 was the coldest for a decade. 

The following year, they repeated their prediction, and the winter of 2009/10 was the coldest for 30 years.

One reason for these mistakes maybe an assumption of anthropogenic global warming, built into the Met Office’s forecasts.

In early 2007, Scientists at the Met Office’s Hadley Centre claimed that year’s average global temperature was likely to be the highest on record.

They claimed that the model they had produced had a 0.06 degree centigrade margin of error.

But by the summer, it was announced that the world’s temperature had fallen. The Met Office retired its old model, and announced a new one. But that warming didn’t appear, and prompted many to point out that the Met Office’s models were consistently running much warmer than observations.

increasingly, seasonal, annual, decadal, and century-long forecasts are being provided to policymakers and planners, providing the basis of import decisions and policies.

These model-based forecasts have not stood up to testing, and do not seem to give better advice than observational records.