Early warnings and probabilities
Although it doesn't always come across very well in broadcast presentations, when an 'Early Warning' is issued for a particular meteorological event, forecasters are issuing what is known as a 'probabilistic' forecast not a 'deterministic' one. It is important to understand the distinction between the two.
When a forecast says something like ... " strong winds and heavy rain will sweep north from SW England, across Wales reaching SW Scotland this evening ", then that is a 'yes/no' (or deterministic) forecast - that is, the expected evolution of the weather with time.
When a NSWWS 'Early Warning' is issued, the probability of those conditions occurring somewhere within the UK must be 60% or higher. That is, on the balance of probabilities, the event is more likely than not to occur (> 1 in 2) .. BUT it is not a certainty.
Increasingly, the level of risk will be set using objective techniques based upon NWP ensemble output - some very sophisticated routines are being developed which will take such output and, based on past cases and the knowledge of current meteorological & infrastructure conditions, produce a probability - which of course can be modified manually if required.
So, what does a probability figure mean? Well, if a figure of 60% probability is issued, it means that given 10 situations with the same set of precursor ( & forecast ) indicators, on 6 of those, 'something' extreme (meteorologically) will occur which will lead to a significant impact on the community - with consequences for the emergency services, local authorities, public utilities etc. However, this also means that on 4 occasions, nothing 'extreme' (in terms of public impact) is likely. The primary customers understand this - unfortunately the general public is not too well educated about the latter point. Even in the dizzy heights of 90% probability (not often used for events more than 12 hours ahead), then there is still a slim chance that nothing untoward will occur.
Within the individual warning, the probability for any particular area may be even lower. It may be that the 'threat' to Wales, SW England and SW Scotland is 60%, but for adjacent areas, such as NW England, the West Midlands etc., perhaps 40%. In this latter case, it is even more likely that nothing untoward will occur (i.e. 60% probability of a 'non-event').
This gives rise to the charge ... "well, they can't get it wrong, can they!" Well, believe me if forecasters continually issued warnings and nothing happened time after time, then the 'powers-that-be' would soon get involved. Each warning is carefully monitored after issue to find out how well (or otherwise) both the forecasters & the end-users coped. Central (UK) government also gets involved directly and indirectly and would view over-use of these warnings with concern. In fact, it turns out that forecasters, being human beings, actually try to get each one exactly right - which is not the idea of the scheme!
So, why use probabilities then? If you are a county planner, or a team leader with the emergency services, or responsible for having 'stand-by' resources available to support the public utilities (electricity supply, rail transport, airport management etc.), then the idea of attaching a % level of risk is supposed to help in the provision of stand-by/back-up services: it allows planners to avoid over-preparing by bringing expensive resources 'on-stream' for a low-threat event (say 20% or one-in-five possibility), but have plenty of back-up available for a 80% (or roughly four-in-five probability) event. Even in this last occasion though, remember that the wind may blow, the snow may fall, but there is still that slim chance that little 'disruption' will occur.
For the general public, the probabilities tend to get 'smeared-out' into very broad "low", "moderate" or "high" risks of disruption on radio and television presentations. But, even when a "high" risk is indicated, it must not be assumed that trouble will ensue. The information is presented to enable YOU the user to decide whether to carry on with your planned journey, building project etc., or to delay perhaps by 24 hours, setting this against the cost in monetary and other terms of either the delay, or carrying-on regardless and 'taking a chance'. This is what 'probabilistic' forecasting is all about - you, the user have extra information to assess the likely outcome of the event, and the costs/risks associated.