
Sometimes models come up with extremely unlikely forecasts that need huge amounts of scepticism so here is a good example why it's incredibly unlikely to hit 49c in SW France in 2 weeks time.
The American GFS model is widely viewed as to overegg temperature projections for Europe in the summer by around 2-2.5c. There's a couple of reasons why but a key one is that it dries out soil too fast in its models. If soil has moisture that can evaporate and cool the air a bit - a process known as evapotranspiration. If it doesn't have moisture, no evapotranspiration is possible. That's already one reason why its projection of 49c is not that likely.
A bigger reason though to not fixate on these temps is that this is a projection for 17/07, that's two weeks from now. It's extremely hard to get good accuracy for five days out never mind fourteen.
Why then do we run models beyond five days? To get a general idea of likely trends for the month ahead.
What can we get from this single model - nothing worth hanging your name on. Even after taking off the obligatory 2.5c and another 5c in case this is an outlier from the various GFS runs, all this particular run can tell us for now is that SW France might be in for a notably hot spell by mid July.
Yet if you look at projections from other models such as ICON, AIFS, GEM etc, the trend - for now - is relatively normal temperatures.
In summary, always look at the bigger picture from multiple sources 🙂