Rain fall data does not match

[Pat]

The one item that my partner has an interest in is the amount of rainfall this month. Any thoughts on following?

Looking at the display on the unit it shows 65.2 for the month.
I check the Feb10log.txt and that shows 65.3 (can live with rounding)

but Cumulus displays 80.3mm for month on alltime records and on website. http://www.comer.eclipse.co.uk/cumulus/

[daj]

Something happened on the 18th February -- some sudden rain but I'm not sure that is relevat, however 28mm of rain in one day seems out of characters for your area based on your other days.

I take it you do not run 24 hours per day? Your logging interval on Cumulus looks as if it is set to 10 minutes, and your station to 60 minutes, so when Cumulus if off we have large gaps in the data so it is hard to tell what happened. I can only assume there was a slight error during that time.

Cumulus has certainly logged 76.7mm in February, plus the 3.6mm today = 80.3mm -- this according to the day file. However it does not match that of the FebLog.

I also see a correction for the first entry in your dayfile of -341.9 for rain. Have you done any other corrections to the day file?

you should possibly change your station logging interval to something nearer the 10 minutes so that when Cumulus is off you are still getting sensible snap shots of data.

There does seem to be something odd going on -- if you look at the entry in the monthly log file for 18 Feb at 00:08, the "rainfall so far for the day" column is 4.2, it should be 0 (or just over if it was raining). It did not seem to zero. There are a several other examples of it not zeroing (another two....23 Feb 00:08, 24 Feb 00:08)

[steve]

It's failed to reset the daily rain counter on a few occasions - but it appears it has actually done the day rollover, so I don't understand how that's happened:

17/02/10,23:20,1.8,93,0.8,2.5,2.5,39,0.0,4.2,979.1,17.6,20.8,44,2.5,1.8,1.8
18/02/10,00:08,0.6,92,0.1,3.6,3.6,68,0.0,4.2,979.1,17.6,19.7,43,3.6,0.6,0.6
18/02/10,01:08,0.6,92,0.1,0.0,0.0,0,0.0,4.2,979.4,17.6,18.0,42,0.0,0.6,0.6

22/02/10,23:10,0.5,87,-1.4,5.4,5.4,35,0.0,3.1,979.0,46.6,20.2,36,5.4,0.5,0.5
23/02/10,00:08,0.1,88,-0.7,0.0,0.0,0,0.0,3.1,979.6,46.6,19.1,35,0.0,0.1,0.1
23/02/10,01:08,1.1,89,0.4,1.8,1.8,68,0.0,3.1,980.0,46.6,17.6,36,1.8,1.1,1.1

23/02/10,23:10,3.9,95,3.2,8.3,13.0,122,0.0,7.8,977.7,51.3,19.2,49,8.3,1.6,3.9
24/02/10,00:08,4.8,96,4.5,0.0,0.0,0,0.0,7.8,977.6,51.3,18.5,49,0.0,4.8,4.8
24/02/10,01:08,5.4,96,5.1,2.5,2.5,158,0.0,7.8,978.0,51.3,17.1,51,2.5,5.4,5.4

The effect of this is that those totals have been counted twice - 4.2+3.1+7.8=15.1, which is your discrepancy (80.3-65.2=15.1).

Could you please zip up and attach your diags folder? Hopefully one or more of those will be still in there so I should be able to see what went wrong.

If you want to fix it in the meantime, stop Cumulus and edit dayfile.txt, and subtract 4.2 from the 18th, 3.1 from the 23rd, and 7.8 from the 24th rain totals. For completeness, you could also edit the 'daily total so far' column for those days and subtract the same amount from each one, which is a bit more tedious - sorry about that.

[steve]

Um, right. Pretty much what David said

[daj]

LOL -- I added my last paragraph later after studying the logs a little more (probably when you were typing your reply)

[steve]

I think there's a clue in the way the log intervals change from 10 minutes to an hour around that time, indicating that Cumulus was closed somewhere between 11 and midnight. I think that, combined with the one hour station logger interval, is exposing a bug in Cumulus. There used to be bugs where it wouldn't do the day reset if these circumstances (closing down with your station logger interval before midnight) but I seem to have fixed those. It seems there may be a similar bug remaining that just affects rainfall totals somehow. Hopefully the diags files will help me.

I would second what David said about your station logger interval. One hour gives you nearly six months running - do you really need that? Reducing it to 10 minutes would also probably eliminate your chances of hitting this problem, whatever it turns out to be.

[Pat]

Thanks for response.

11 to midnight is when I tend to get time on the PC so that could certainly have an effect. will amend day files etc, over next few days.

Have read the manual a few times but cannot see how i can change the data logging interval on the station. La Crosse WS 2355

[steve]

Could you let me have the diags files zipped up, please? I still think there's a bug in Cumulus here.

I think that to change the logger interval, you'll need to use the software supplied with the station (Heavy Weather?)

[Pat]

Hi Steve
Found the setting in heavyweather and have changed to 10 minutes.

Diag files attached

Pat

[steve]

Thanks, Pat. I've found the problem; it's to do with the rounding of the rainfall total and thinking that the counter in the station has been reset. I'll see if I can fix it.