Cumulus Support

the Cumulus forecast number is just that - a forecast of what will happen in 12-24 hours time, not the current conditions

PaulMy wrote:

Hi BJCKiwi,
The cloudbaseCU has been showing very well yesterday, overnight and to this afternoon using the $useCUheight = true and the height changing whenever Cumulus recalculates the cloud height as per Cumulus Cloud Base Calculator . I periodically checked the local Metar and the Clouds and conditions where fairly close.

However this afternoon I noticed it was showing rain and displaying "Drizzle" when the sun was shining brightly. In some search I found that the uploaded cloudbaseCU.txt was 15 for <#forecastnumber> which according to the samplestrings.ini in my Cumulus folder is "Showery, becoming less settled" (I don't have a "strings.ini" in my Cumulus folder). The Davis station is showing "Increasing clouds with little temperature change. Precipitation possible within 24 to 48 hours." The Sun icon was/is shining on the Cumulus console.

Is this discrepancy of being quite clear with sun shining condition compared to the cloudbaseCU.php showing rain and displaying "Drizzle" just within the margin of error in the calculations or is there other settings I can experiment with?

I have now changed the settings back to use the local Metar CYXU which is 26 km away and that is currently presenting a more accurate display - no rain and displaying "Mostly cloudy". The Metar shows Clouds at 811 m but that was about 30 minutes ago and Cumulus console is showing between 750 to 900 m which is reasonably similar, but the rain and drizzle is the discrepancy.

Thanks,
Paul

Unfortunately this seems to be a common problem with the 'Zambretti' forecast (used in Cumulus).
I might add I have found the same thing with the Davis forecast. bt-Sager weathercaster is similar if the conditions are not manually entered as recommended.
The common element here is that these systems all use a 'best guess' based on current data and past trends - none of which indicates the actual cloud conditions which is what the script is about.
Metar however uses actual observations (manual and/or instrument based depending on the site).

Unless you have some local cloud sensing instrument, Metar is the best option provided it is relevant. This depends on how far you are from the metar source location and on how fast the weather is changing. Standard Metar updates are hourly on the hour. I got annoyed with the Metar saying it was raining when it wasn't which is why the overrides were added so raining was determined by local data. The overrides have been configured only to work with Metar as Metar provides separate cloud and rain data where there is only one data point from Zambretti.

If you don't want to use Metar, You could try changing the $curcond = $vars in the switch statement around line 380. These are used in place of the normal text strings and the ones used were my 'best guess' at the equivalents to the original Zambretti text. Note - you can only use the vars ($sunny, $scattered etc) shown as these are the only ones understood by the script. You can change/move them around but not create new ones.

If you want to experiment further, you could try the rain overrides with the zambretti by changing (just above the $curcond case statement);

Code: Select all

if ($getMetar && $useWXrain) {   // override non-rain Metar cloud type if station has rain
to
//if ($getMetar && $useWXrain) {   // override non-rain Metar cloud type if station has rain
if ($useWXrain) {   // override non-rain Metar cloud type if station has rain

Trust this helps.