accurate weather forecasting is more than a bit tricky. A few years ago I attended a fascinating lecture given by the head of the Met Office who explained the way they use extant data and computer models to make forecasts and to assign them levels of confidence.
Turns out that quite small variations in the 'present data' (i.e. from live weather monitoring stations) can make large differences to the output from a computer model. So one of the things they do is they run the model many times over, each time with small variations in the input parameters. They then compare the predictions of these various models. If the predictions are always the same then they have a confident forecast. However if they vary wildly then the confidence level is much smaller.
Needless to say the predictions vary more as you predict further ahead. This means that only rarely can you assign a very high confidence level for a forecast (say) even a week ahead of time.
The proud boast of the modellers is that their predictions a week ahead are as good now as their predictions for tomorrow were a few decades ago. This sounds great until you realise that this means 'around 50%, probably worse than that'.
This is worth a look
https://weather.slimyhorror.com/but the priorities used to judge accuracy are not the same as yours. The links from that page are interesting too.
IIRC the BBC don't use the Met Office any more. Duh. Hence their forecasts have been a bit shit for a little while now.
[ they said they were looking to change in 2015, and because they couldn't get anything that worked for them, ended up extending the contract with the Met Office until march 2018. Since then it has been 'meteogroup'. ]
cheers