Page 1 of 1

This Year's Rainfall Total

Posted: Sun 14 Oct 2018 10:11 pm
by Grimers
Hi everyone,

I hope you're all doing well.

Got a slightly problem. Cumulus MX has calculated that this year's rainfall total is 485.6 mm (at the time of writing), but adding up the daily rainfall values for the data available this year (incl. today so far) gives 484.0 mm. I've read the Wiki and apparently Cumulus uses the dayfile.txt to work out the yearly rainfall total, but it's calculated 1.6 mm to much.

I've checked the dayfile.txt (which is used on one of my website's pages here: http://www.newton-poppleford-weather.co.uk/tables.php), and it agrees with the 484.0 mm quoted above. So, the question is, where's Cumulus MX retrieving this extra 1.6 mm of rainfall from? If the dayfile.txt is saying 484.0 mm, then surely this year's rainfall total should be the same?

Any ideas?

Thanks,

Will

Re: This Year's Rainfall Total

Posted: Mon 15 Oct 2018 12:25 am
by water01
Rounding?

Re: This Year's Rainfall Total

Posted: Mon 15 Oct 2018 6:30 am
by ConligWX
was the 1.6mm, todays rainfall and hasn't been added to the yearly rainfall until rollover perhaps?

Re: This Year's Rainfall Total

Posted: Mon 15 Oct 2018 9:28 am
by Grimers
Nope, it's still at 485.6 mm, so it's still invented 1.6 mm from somewhere.

Re: This Year's Rainfall Total

Posted: Mon 15 Oct 2018 10:18 am
by Mapantz
Hi William

I see you have missing data in your data summary for May 26th. Did you have rain on that day?

Re: This Year's Rainfall Total

Posted: Mon 15 Oct 2018 11:41 am
by sfws
A few additional thoughts, that might all be red herrings!...

The table web page you link to mentions some problems with rain gauge and with Davis commmunications, did you make some adjustments in connection with those problems that amount to your discrepancy?

As you use a Davis station, Cumulus will be using the "calendar year to date" rainfall from the console, as the total rain count (http://wiki.sandaysoft.com/a/FAQ#.28Inv ... in_Counter) it uses to work out today's rain.
As far as I know this does not affect what you have read in
http://wiki.sandaysoft.com/a/FAQ#Where_ ... ls_from.3F. But as I don't know exactly how Cumulus is coded, perhaps it is a factor that might affect one of the figures you quote?

I assume you don't have a year to date figure in your settings being added for this year, and all figures represent year from 1 January?

As your rollover is at midnight the discrepancy can not be from rainfall in any period between midnight and rollover, unless there was a faulty rollover on one of the days in the period.

Re: This Year's Rainfall Total

Posted: Mon 15 Oct 2018 12:41 pm
by Grimers
@Mapantz

Hi, how are you?

Yes, I think so. But, surely if this year's rainfall is using the data in the dayfile.txt and that day isn't there, then it shouldn't use it?
sfws wrote:A few additional thoughts, that might all be red herrings!...

The table web page you link to mentions some problems with rain gauge and with Davis commmunications, did you make some adjustments in connection with those problems that amount to your discrepancy?
No, AFAIK.
swfs wrote: I assume you don't have a year to date figure in your settings being added for this year, and all figures represent year from 1 January?
Not AFAIK.

Re: This Year's Rainfall Total

Posted: Tue 16 Oct 2018 8:03 am
by sfws
William, my final suggestion (prompted by Water01 suggestion).
Does your version of the daily summary on your 'table' web page treat single tip rainfall amounts as not significant?

If so, the various ignored 0.1 (or whatever value is set) might sum to the discrepancy you report. Some versions of that presentation, as discussed many times in the forum, ignore a single tip, as it might be dew or might be partly due to rain falling on a previous day. Indeed the UK Met O set a minimum threshold for rain to be counted, and in handwritten reporting days used the term "trace".

Instead of relying on the web page, take a copy of the dayfile.txt lines relating to this year into a spreadsheet and see what the relevant column adds up to.

Re: This Year's Rainfall Total

Posted: Tue 16 Oct 2018 10:29 am
by Grimers
sfws, you're a star! Thanks very much! It is something to do with that table! The dayfile.txt amount matches with this year's amount. Now, to try and work out how to fix the table. :lol:

Re: This Year's Rainfall Total

Posted: Tue 16 Oct 2018 2:45 pm
by sfws
Grimers wrote:sfws, you're a star! Thanks very much!
I am pleased my brain still works!
Grimers wrote:Now, to try and work out how to fix the table.
Beteljuice explains this in https://cumulus.hosiene.co.uk/viewtopic.p ... 45#p107031.

Re: This Year's Rainfall Total

Posted: Wed 17 Oct 2018 8:45 am
by vr34
Hello,
I read carefully all your explanations to understand and correct the difference between 484 mm and 485.6 mm, which represents a difference of 0.3 %, negligible in my opinion on the year! It is also necessary to take into account the limit of detection and accuracy of our electronic rain gauges. That of my rain gauge is 0.2mm. Which means that if it rains 0.1mm, it does not record anything or if it rains 20.1mm, the collected data is 20mm. So cumulatively over a year, the error is well above 1.6mm.
Data calculations are not everything. The accuracy of the material is also essential.
Here we talk about the difference between accuracy and precision ;)
Regards,