All outdoors activities precautions include taking note of the weather forecasts for the day.

However, I have seen those forecasts be terribly wrong in many places, even for a two-hours ahead prediction.

So, what are the conditions that make weather forecasts reliable? Up to what duration?


In the French riviera, near the Alps (Nice backcountry), where both the sea and high (~1200m) mountains are within 10 km, the weather forecasts are totally unreliable, from one day on the next one, but even from 8 AM (storms announced, rain visible) to 11 AM (storms announced, bright sun visible).

Inversely, in central France (Burgundy), in a much more continental weather, predictions are reliable up to 2 days ahead with good confidence, both for temperatures and rain. 1 hour-ahead to-the-minute rain predictions are correct 90% of the time.

Based on that limited experience, I would tend to say that sea and mountains are reasons for unreliability. But up to which distance? What about oceans? From which heights to mountains impact? Are certain seasons more error-prone than others? etc etc

  • Weather predictions in the Southeastern US in the mountains aren't very reliable at all. We routinely get weather 10 to 20 degrees (F) outside of the prediction and the running joke is to never hike if there is a 30% chance of rain because you'll get soaked every time. May 20, 2012 at 19:11
  • 1
    A little trick for the checking the weather predictability remotely: Find two or more forecast sources for the same area and compare the results - the bigger the differences, the less sure you can be about the weather. Of course, you need to be sure the sources are different. May 20, 2012 at 20:49
  • In Scotland, same day is usually okay, however I have seen a morning forecast suggest sun and 15 degrees end up with snow and 0 degrees (centigrade)
    – Rory Alsop
    May 20, 2012 at 21:11
  • 1
    Great question, I'd be interested in a comprehensive answer too if anyone has one!
    – berry120
    May 21, 2012 at 13:45
  • @RussellSteen you are miss understanding what 20% chance of rain means in the US. It really means there is 100% chance of rain over 20% of the forecast area. May 23, 2017 at 18:31

1 Answer 1


The usual method, and actually the one meteorological offices use the world over is to run multiple models with slightly different parameters. When all the outputs show generally the same result they have confidence in the forecast, and when they come up with different results they have low confidence. They say this on weather reports, often giving percentage chance of rain, cold etc.

All you can do is try and look at reports from differing sources and run your own comparison. It is not an exact science by any means, and if you are anywhere near the sea the results could be very erratic. In the centre of a continent, the forecast can be much more predictable, as there are less uncertainties in the modelling - the hard bits are where sea meets land.

All you can do is gain a degree of confidence by listening to local experts, national forecasters who use massive modelling runs and taking all of it with a degree of skepticism.

Thanks to @RowanHawkins who pointed out that:

An addition to this is that the offices in the US have a few extra rules to abide by. For instance where forecast areas meet, there cant be more than a 5 degree (f) forecast difference between the sides. They do talk to each other and some modeling is done on a local basis


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.