UKC

What's going on when the met office mountain forecast, for example this one:

https://weather.metoffice.gov.uk/specialist-forecasts/mountain/snowdonia#?d...

gives a warning for thunderstorms but none of the summit specific forecasts in the area give the same? For example on Sat 22nd of March the met office mountain forecast gave a medium likelihood of thunder and lightening in Snowdonia, but the summit specific forecasts of mountains in Snowdonia didn't have the warning, or the thunder/lightening symbol, and the text description just said something like 'heavy showers'.

I'm guessing the situation on Sat was that there was a general risk of thunderstorms in the area but uncertainty was too high to pin it down to any specific summits? But this seems strange as the overall likelihood was medium so not insignificant, so why not say 'chance of thunderstorm' on the summit specific forecasts? It sounds nitpicking but I'm trying to understand how best to interpret this kind of thing to avoid thing going wrong.

The area forecasts are written by human forecasters, the summit specific ones are entirely computer generated. I would suggest using the latter for additional localised information, and not as a substitute for reading the area forecasts. 

Thanks. Wouldn't the met office be a bit alarmed if an expert human thinks there's a medium risk of lightening but the computers haven't spotted it? Or is the medium probability for an area reduced to very low probability when looking at an individual summit?

That's possible, though Snowdonia is quite a small area and I would not have thought that there would be much difference between summits that was resolvable by the forecast models.

Could be that the the probability was not high enough to result in a thunderstorm symbol: producing a single symbol that represents the weather accurately is not straightforward. I get the sense that the Met Office has erred towards more optimistic symbols in recent times - it doesn't seem to show a rain symbol until the probability exceeds 50%. I think (though can't immediately find a reference) that the Met Office definition of medium probability is 40-60%, so it is possible that a 40% chance of thunderstorms gets into the text forecast as medium probability, but doesn't result in the thunderstorm symbol. 

I couldn't see a reference to thunderstorms in the forecast you linked. The only medium hazard was strong sunlight (there's reduced ozone layer over the UK just now).

More generally, as Mark says, the area forecast is written by a human. This won't be based on just one model, but will consider a range of different of sources of information, including other models and observational data. The summit specific forecasts are generated from a mix of models and don't have any human input.

The other factor is that the way the symbols for the summit specific forecasts are selected is a bit complicated, but essentially looks at the likelihood of each different weather symbol, but with a weighting applied to favour more hazardous weather types.

If the area forecast is showing a medium likelihood of thunderstorms, but no thunder forecast in site specific forecasts I suspect the most likely reason would be it hasn't quite reached the threshold to be shown in the site specific, but the human forecaster has judged that the risk is significant enough to flag up.

I linked to the current snowdonia forecast not the one on that date sorry. It gave a medium likelyhood of thundery showers which also means lightening. I just find it odd that a human thinks there is a medium likelyhood of something quite dangerous in an area but the automated systems haven't flagged it up for any actual points in the area. Not that it really matters for me as I look at the human generated mountain forecast but maybe not everyone knows to do that.

They are showing different things. The summit specific forecast shows a definitive forecast for a given location, i.e. it's saying the weather will be X at time Y at that location. I can't remember off the top of my head the probability threshold for showing a thunderstorm symbol, and think it varies depending on what other weather types are happening at the same time.

The hazard on the area forecast is giving a likelihood, albeit a crude one, so can still flag it when there's a lower risk. Given the very hazardous nature of a thunderstorm, a human forecaster might be inclined to go medium likelihood for a lower risk.

So hopefully you can see how in marginal situations lightning is more likely to be highlighted in the area forecast than the summit specific forecast. If the summit specific forecast showed lightning even if there was a marginal risk, it's a bit of a 'boy who cries wolf' situation, since it doesn't give any info on the likelihood, and people would loose trust in the forecast.

I'm not sure its wise to replace common sense with a weather forecast which by definition is looking into the future.   If you are going into the hills then by all means use the weather forecast but use your own knowledge or others on what the weather may turn out to be.

I know that I was just wondering about the differences between the two forecasts and how the met office handles these things.

I might have been misinterpreted as being naive or not knowing much about forecasting (I've been in the mountains a fair bit and seen a lot of weather) but I'm asking how come a human forecaster sees fit to give a medium probability of something dangerous but none of the automated forecasts include it, and whether this discrepancy is a big deal given that people don't always look at both.

No problem, I guessed that it's just hard to know in text only which made my reply sound more defensive than I intended. My reply was more because I wanted to clarify what I was actually asking, because I wasn't entirely sure at the beginning of the thread.