Incorrect Cumulus daily recordings

[Downwest]

Hi Steve, (and anyone else who's interested!)
For some reason, on checking today's records, I'm told by my Cumulus on-screen display (using a W-8681 weather station) that today was the warmest day this year....(65.1F) when it probably went no higher than 52F. (I use Fahrenheit cos I'm over 65!!)
In addition we apparently had 14.5mm of rain today when it actually remained totally dry.
And my air pressure reading also dropped from 1022 to 1012mb.
And just to add insult to injury, I couldn't log in (as Souwesterly) and had to open a new account.
All this happened around 7.50 pm today......in moments.....
What happened?
And what should I do now?

[steve]

These may well just be the 'usual' odd readings that Fine Offset stations tend to produce, particularly if they all happened at around the same time - are you using any 'spike removal' settings? Did the values come from 'live' data or from the logger? Regarding the pressure drop, did it drop and stay there, in which case have you looked at the FAQ?

If you zip up the diags folder and attach it, along with your March log file, I'll see if I can see anything.

[Downwest]

Hi Steve,
Files duly attached.
Hope you can find something to explain the workings.
I could perhaps understand some kind of temperature jump on some days but not last Thursday!
And no-one else in the locality recorded any anomolies.
On Thursday a cold front did come through here but it brought no rain and nothing but a gradual decline in pressure and temperature according to local sources.
So, over to you!
Chris

Mar12log.zip

PS All readings back to normal today - even a rainfall rate of 32mm per hour - genuine! (Got 10mm in 20 mins)

[steve]

The time you mention corresponds to the time your station got to the end of its memory and started again at the beginning. I don't think there's a bug in Cumulus, it looks like it's doing the correct thing. I'll try to find some time to look at your data in detail to see if I can see what exactly happened.

Edit: It looks like the problem was that your station hadn't actually got to the end of its memory and started again at the beginning, but the 'current data pointer' had been reset to make it appear that had happened (i.e. the station started using the locations at the beginning of its memory before it had got to the end of its memory). So when Cumulus downloaded the logger data to fill in the gap from when it was closed down, it was reading old data from the end of memory, hence the incorrect values.

[AllyCat]

Hi Chris,

Do you run Cumulus 24/7 or have you checked/replaced the batteries in the Console? Odd things can happen when the Console switches from (low) battery voltage to power from the USB cable.

With the "Select-A-Graph" facility you might be able to see when the spurious "end of buffer" data was originally recorded (i.e. identical graph shapes from some time ago) and thus patch it out. Your logging interval and the information that there should be 4080 records in a full buffer might help. A FAQ or Wiki entry may explain how to get Cumulus to read back further through the buffer than it "thinks" it needs to. But I don't think that correcting all the dates/times in the log file will be a trivial task.

Cheers, Alan.

[Downwest]

Hello Alan,
Yes - I run Cumulus 24/7 - and I had thought that perhaps the batteries were running low - so I'll change them tomorrow as a precaution.
An interesting idea to check back for matching graphs - but at the end of the day I reckon I'll just accept that no machine is perfect! Humans, maybe - but not machines! LOL

Steve - while writing - I'm now getting some error reports.
19/03/2012 19:06:37 : Error on line873 of C:\Cumulus\Data\Mar.12log.txt. List index out of bounds (1)
This is repeated at 06:38 (three times) and at 06:46
What's all that about?

[steve]

It's telling you that there's a problem at line 873 in that file:

C__Users_steve_Downloads_Mar12log (1).png

A line of 'NUL' characters was written to the file instead of data. You can edit the file and remove it.

[Downwest]

Finally had time to check the data folder.....
And 'yes' as you say the date seems to be missing - at first glance.
Or at least it's not in it's proper place - because everything is displaced to the right.

So to cure the problem, all I had to do was to plant my pointer at the start of the row and hold delete until the proper information scrolled left and into place.
Problem solved!

Thanks for your help on that - and for your comments re the strange spike the other day.
Just have to put that down to a stray cosmic ray or something!

Cheers folks!

Chris