Waspsnest’s Contributors

The Wasp
Mr Raccoon

Met Office Forecasts

The Met Office have annoyed me for a long time now with their pitiful attempts at forecasting despite costing £83 million per year :


The PWSCG is mainly funded by the Ministry of Defence on behalf of Government. In addition, the Civil Aviation Authority pays towards the service to underpin its services for UK and global aviation. Some research and development projects receive additional funding from other sources, including the EU. The current total PWSCG funding is approximately £83 million p.a.

As for their accuracy, I have heard many stories relating to forecast accuracy improving but finding hard evidence that is understandable is somewhat difficult to come by. The Met Office website has this section on accuracy but it is filled with trends of numbers showing an improvement but no actual description of what the magical numbers mean.

For example :

Forecast verification

Verification is needed to measure how we are doing against the Key Performance Targets (KPT) for forecast accuracy, as set by government. These KPTs are part of our ongoing long-term commitment to improving the accuracy of our forecasts to the public. We will do this by reaching accuracy targets set for:

  • Probability of precipitation forecast, for the next day and the day after, for 139 places within the UK (Met Office observing sites).
  • Temperature forecasts, for the next day and the day after, are assessed at the same sites as the probability of precipitation forecasts.
  • The forecasting skill of our computer models

UK NWP index for recent months

UK NWP index for recent months

Interesting indeed and looks impressive to a casual observer but I have no idea what an accuracy of 118.0 means and more to the point, why is the target for March 2011 118.0 when they have achieved 118.5 only 12 months ago?

Out of interest, I decided to conduct a little experiment to see whether it was my personal perception of the forecasts that was at fault or the forecasts themselves.

For clarity, I am not talking about a rain shower not turning up exactly at 12 noon or scattered showers turning into downpours. I am talking about the 3 to 5 day forecasts which give overall information about the weather ahead such as wet on Friday or cold and windy on Saturday, which are the kind of things most people would look for when making plans for the weekend for example.

The experiment simply involves taking a snapshot of the current 5 day forecast as shown on the BBC news website over a period of 4 full days to see how the daily weather overview changes with time.

The results below are rather interesting in that it suggests to me they have no idea what the weather will be like in 3 days time (look at Thursdays forecast for example) let alone 5 and, if that is the case, why bother having a 5 day forecast at all?

The first image below is a composite of forecast results starting with Monday nights forecast for the week in the first column, followed by the evening forecasts from Tuesday to Friday nights with the forecast days aligned against the first Monday forecast.

You will see, for example, that Fridays forecast progresses as Heavy Rain, Grey Cloud, Light Rain, White Cloud – the latter being last nights (Thursday’s) forecast.

Interestingly, it pissed it down here (in Sheffield) all morning today and didn’t really get light until lunchtime which means Monday’s forecast appears to be more accurate than last nights – so much for those improving graphs of forecast accuracy.

That is not a one off either – Wednesday nights forecast for Thursday shows sun and clear skies when in fact it was overcast from lunchtime and rained heavily from late afternoon onwards.

Indeed, if anything it suggests that the earlier view of a forecast is more accurate than the day before view of the same weather which is just bizarre.

Overall, I don’t think I would pay them in washers, let alone £83 million.

Composite forecast :

For completeness (in case any weather types wish to look at wind and pressure data too) I have posted below all the forecast snapshots I gathered to produce the composite.

Images are in order, Monday evening to Thursday evening :

Pinecones and seaweed anyone?

Related Posts with Thumbnails

1 comment to Met Office Forecasts