Odd values from #forecast and #cumulusforecast

[freddie]

The above web tags are outputting a strange forecast of "Rain at times, very unsettled". I know the forecasts calculated by the weather stations and Cumulus aren't brilliant, but this one is well wide of the mark (pun unintended) and have made me curious.

I have checked the usual suspect of pressure extremes defined in my MX configuration.

I initially suspected something I had done - so I checked what MX was actually outputting with the shiny new MX endpoint for webtags - http://localhost:8998/api/tags/process. ... usforecast

Both display the forecast text from the first paragraph. I have MX configured to generate the forecast, rather than use that supplied by my VP2 - so there is no surprise that they are identical.

Am I missing something?

[ConligWX]

"Rain at times, very unsettled" is from the Zambretti forecast scale

viewtopic.php?t=9318

[freddie]

Thanks Simon. Yes, MX uses Zambretti, and I have MX configured to supply the forecast.

[mcrossley]

It's giving me "Rain, very unsettled" - so v. similar.
I wonder what the Sager algorithm is giving across the UK at the moment? I don't run it anywhere at present.

[mcrossley]

The answer for Sager appears to be... "Increasing cloudiness or overcast followed by rain or showers and warmer. South or Southwest winds, Probably increasing."

[ConligWX]

mine says "its raining" - typical weather for NI

[freddie]

Sager sounds about what I would've expected. Strange that you've got a similar forecast to mine. I wouldn't have anticipated that forecast as there's nothing in the data that would trigger it. Not 100% sure how it works, but I would've thought you would need either low pressure values or rapidly falling pressure. I've got neither - and I don't suppose you have too.

[freddie]

I used the @beteljuice JS implementation of the Zambretti algorithm to confirm the text I was seeing. Indeed it is as expected (by Zambretti).

In that case, I will go back to ignoring it again

[sfws]

From memory, I have not looked up what MX does, Negretti and Zambra's wartime forecast worked on
-actual s-l pressure and whether steady, rising, or falling
-wind direction, plus veering or backing
-that min & max pressure range you mention

I used the @beteljuice JS implementation of the Zambretti algorithm

I believe beteljuice advised Steve Loft, so they use same basis, but I might be wrong on that too.

Here I have had a mix of fair weather cloud types all day, but the blue sky is increasing now it is evening so I don't expect rain. I think I trust the sky more than an algorithm - MX is telling me "Stormy, much precipitation". Given how little rain I have had in June and July, the local fields and gardens might be wishing the algorithm was true, but I have been outside painting a rusty complex antique object the last few days and I don't want rain.

[Mapantz]

Mine was/is saying rain, and so is the Davis forecast.

[freddie]

My Davis wasn't earlier, but is now. My point was that the pressure didn't seem an overly low value (and was only falling very slowly) so I thought that a forecast of "rain, very unsettled" seemed at odds with the measurements. Now I have read up on the Zambretti algorithm I can see that it was reporting the correct (if unrealistic) value. So I am going to go back to ignoring this type of forecast - and probably will remove it from my website as it has the potential to be so far off the mark.

[mcrossley]

I use the met office forecast on my site, it's a bit more accurate, but not always!