Strange temperature flat-line

[Gina]

Just been checking my weather sites and found something very puzzling. The temperature (and derivatives) shows a flat line between about 0930 and 1200 this morning. But the wind speed and direction show no flat line. Also, this only occurred with Cumulus - G-Weather shows a continuous temperature plot over the period. If the problem was due to loss of signal I would have expected it to affect the wind too. Both console/computer/software sets receive the same transmission from the outdoor sensors and yet one seems to have lost temperature data only.

Cumulus :-

temp.png

wind.png

windd.png

G-Weather :-
(The spikes are due to interference from our electric fence - I haven't added software spike suppression yet.)

24hrs-Temperature-11-02-07.png

[tjaliwalpa]

I've seen some strange things like that on my Cumulus graphs as well. Never been able to get to the bottom of it. I have managed to remove them by restoring a backup and getting Cumulus to read from the logger again.

PS Don't look at my graphs right now because the errors there are to do with me taking my station down to put up a new shield and to replace the batteries while I was at it. DAM, BUGGER, BLAST I broke the wind sensors when a gust of wind blew and my ladder fell (no I wasn't on it at the time). Ordered new parts already.

[captzero]

PS Don't look at my graphs right now

Sorry Bob, I just had to have a look (bit like a 'wet paint sign' to me)
Hope you are up and running again soon.

[Ned]

Gina, I'm guessing you have temperature spike suppression active in Cumulus?
I quote from the Cumulus calibration help:
The "Spike removal" section allows you to specify the maximum differences between successive readings that you wish to allow, for Fine Offset and La Crosse stations. If a reading differs from the previous reading by the value you specify, or more, then it will be ignored. Be careful in your choice of values; if the difference is genuine, (i.e. not a 'spike') and exceeds your supplied value, then that reading will be 'stuck' at the previous value until a new reading returns to within your supplied difference.

[steve]

Yes, Ned's explanation is almost certainly the correct one. It doesn't quite seem to tie in with the plot for your own software, though.

[mcrossley]

It looks to me like (if spike suppression is enabled) that Cumulus followed the small downward spike around 9:30 as it just within the suppression limit, but the rise after the spike went to a slightly higher temp and was just outside the spike limit. Cumulus then waited until the temp fell back into range again at just before 12:00. The scale on Gina's plot is a bit too compressed to see it clearly though.

[Gina]

Ah yes! Thanks for all your replies :- Yes, I have the EWTempdiff set at 0.5 - too small. I'll change it.