How accurate are our forecasts? 

Weather forecasting and the methods by which we get forecasts have come a long way in the last 20 years.

Weather forecasting and the methods by which we get forecasts have come a long way since all we had was the shipping forecast for the next 24 hours, only available via BBC long wave, and which we scribbled down, transferred to a DIY synoptic chart to put flesh on the bare bones of the Shipping Forecast and even deduce an outlook! 

Very hairy; difficult to teach and even more difficult to maintain the skills. It did, however, mean that sailors had to think in basic terms about weather. These days some sailors seem to expect the same sort of accuracy from weather forecasts as they do from their GPS and navigational software.

So how much accuracy should we expect from our forecasts? It is a common criticism that sailing days are lost because the predicted Force 7 did not arrive, or that others got caught out in a nasty little blow where the wind strength was far greater than forecast. Somehow we expect better.

RYA Work on improving forecasts

The RYA's Cruising Manager, Stuart Carruthers belieces people often have unrealistic ecplanations when it comes to forecasting: "A weather forecast is just that; a prediction, it's not set in stone. These days, perhaps in part due to the introduction of detailed forecasts on the web, expectations of forecasting accuracy seem to me to be a little unrealistic.

"Over the past three years the MCA MSI Group of which the RYA is a key member has, with a great deal of input from practical sailors, accomplished much on behalf of all coastal mariners, especially those in recreational craft."

The Group has argued for and encouraged the MCA and the Met Office in providing the following enhancements:

• Improved 3 hourly Inshore Waters schedules on VHF, not bettered anywhere.
• Use of NAVTEX 490 kHz for  regular Inshore Waters Forecasts (IWF) and 24 hour outlooks.
• Use of NAVTEX 490 kHz for coastal station reports.
• The availability of coastal station reports on the Met Office website.
• The publication of the NAVTEX 518kHz extended (3-5 day) outlook on the Met Office website.
• Addressing the need to improve NAVTEX reception to the west of Scotland.
• The splitting of the inshore waters area 16 Hebrides into two separate forecast areas.
• Adding a Channel Islands area forecast to the Niton NAVTEX 490 kHz broadcast.  

These improvements collectively mean that any mariner at sea around the coast of the UK has access to 24 broadcasts every day that are free and based on 4 entirely new Met Office Numerical Weather Prediction (NWP) runs.

Using NAVTEX, much of this is also available up to 250+ miles out to sea, and sets these days record the information so that it can be referred to later. Throughout the negotiations, the MCA and Met Office have been highly proactive in these improvements.

Financial realities

There is a popular belief that more weather information could be justified by cost savings to Searh And Rescue and other services; however, this cuts no ice with HM Treasury. Their philosophy is that sectoral interests should not benefit at the taxpayer's expense and the ‘user pays’.  

This financial reality and the obsolescent terrestrial technology inherited from the days when BT ran the coastal network has been a major constraint on what has been possible.

Nevertheless, a great deal has been achieved which is of direct benefit to the recreational sector. Clearly with more resources, it might be possible to provide more information, but sadly some of these resources are finite.

For example: there are simply not enough VHF frequencies available for the Coastguard to dedicate frequencies to continuous loop broadcasting (at least four would be need to achieve the current level of coverage).  

The VHF should always be monitored at sea when in range, so broadcasts should not be missed. In port, there are other sources such as the Internet for texts of HMCG forecasts.

Contrary to a general impression, the HMCG’s role and responsibilities extend well beyond providing yachtsmen with forecasts for the day. So remember that when you call an MRCC to ask for a forecast 10 minutes after the scheduled broadcast! You might well get a polite rebuff.

New technology, unrealistic expectations

Beyond the realm of VHF and NAVTEX, new technology is also helping cruisers gain access to forecasts, but again, many are falling into the trap of thinking these forecasts are absolutely bulletproof and becoming infuriated when they are wrong.

The RYA's resident weather guru Frank Singleton explains: "Weather forecasting has moved a long way in the last few years, in particular, there are now GRIBs (Gridded Information in Binary) which provide wind and other data straight off the weather forecast computer.

What a difference a decade or so makes: DIY Synoptic charts and modern charts on the internet. Just remember, they're not without their drawbacks.

"These look very convincing and it is all too easy to fall into the trap of thinking that computer generated outputs must be ‘accurate’. Some web sites now offer ‘precise forecasts’ for up to ‘7 days ahead’ anywhere in the world which are ‘location specific’.

"These are unrealistic claims and can lead users to expect the unachievable – it is simply impossible to produce specific forecasts for, say, Start Point or the Needles Channel; yet we see claims to do just that!"  

Accordingly, mariners should be careful how they use these or indeed, any forecasts. Professional forecasters use computer generated information carefully and will amend and interpret it as necessary.

Few of us have such skills or the necessary knowledge. In order to understand the realities and practicalities of weather prediction critics should, first, try describing weather experienced in a 30 mile coastal passage as though it were a text suitable for broadcast.

Then they should imagine doing that for a 24 hour period for an Inshore Forecast area bearing in mind all the localised weather effects. Would the text give a description recognisable by a sailor in the same coastal area? Would we have time to listen to it? Probably not in both cases.

Limitations

The limitations of detailed coastal forecasts and the benefits of using the strengths of the longer term predictions as sensible planning tools should be recognised. Above all, any forecast has to be interpreted using experience and common-sense; that is so for even the most detailed computer predictions.

Significantly improved 5-day forecasts allow sensible, considered decision making and greatly enhanced safety; such questions as 'When and where do we go next?' or 'When and where should we seek shelter?'

Systematic daily consideration of forecasts from all available sources (and there are plenty of them) allows 4 or 5 day planning with some confidence. The key in all of this is the word ‘forecast’.  In this context, a 'forecast' is the best prediction possible within the constraints imposed by computer power and data availability.

The Met Office Numerical Weather Prediction model is a set of mathematical equations that are crunched through the Met Office super computers.

The output is fine tuned by Met Office forecasters using actual observational data. It should be emphasised that the Met Office is always concerned with, and continually monitors, the performance of their models; this leads to continual, gradual improvements. 

A look at their website will show that there have been significant improvements over the past 20 years as technological advances have made scientific improvements possible.

In addition, the Met Office and MCA are currently analysing the accuracy of the Inshore Waters Forecasts; this is showing that the 24 hour IWF is 90% accurate to within 1 Beaufort force.  

However, as has been stressed above, mariners must recognise the current impossibility of using NWP models to predict the weather on the space/time scales we experience – say 2 miles/30 minutes. The grid spacings used by the Met Office are as follows

• About 20 NM for large, synoptic chart scale prediction to 6 days ahead. 
• About 5 or 6 NM for smaller features, around 25 to 30 NM size over the North Atlantic and Europe, up to 48 hours.
• About  2 NM for the short term prediction of small scale weather, down to about 10 NM size, over the British Isles, up to 36 hours ahead.

It should be remembered that the lifespan of weather features varies with their size. Large lows and highs can last for days, or longer. A small thundery low of size 30 or 40 miles may last one to two days. Weather on a scale of a few miles, a single thunderstorm say, may only last a few hours. Gusts last seconds. These times determine their predictability.

Most optimistically at present, the Met Office, world leaders in the field, can only predict on a 10 mile space scale and, more realistically, 25 miles. That is for 'today' and perhaps 'tomorrow'. When it comes to periods beyond two days the size of the weather feature likely to be predictable is more like 100 miles.  

This means that the computer forecasts used for the IWF are, effectively, averages over an area of about 15 radiuses each the grid point. The numerical information has to be condensed into a manageable number of words for broadcasts that can convey the general meaning of what is going on.

Future developments

However, the Met Office is now experimenting with 1.5km grid models so who knows what might be possible in the future. Of course, this opens up other problems such as communication of large volumes of data and how a mariner would use them.

To read more about forecasting, the uncertainties in forecasts and how to use forecasts check out Frank's weather website.