Removing spurious readings

[Blackmyre]

Yesterday my Fine Offset seems to have recorded a record gust of 85.2mph. I have no idea how or why it did that, but I think we'd have noticed had it really occurred!

Today I've noticed a record high rainfall rate of 460.8mm/hr. This one's presumably more explicable: I would guess it's due to melting accumulated snow suddenly gushing through into the collector.

What's the best and safest way to remove the spurious data? Presumably it's more involved than just finding and deleting the individual observations in the monthly datafile.

[daj]

You can modify the all time records within Cumulus itself -- Edit, All time records. This would assume you knew what the previous highs were for your gust and rain rate

How often do you shut down and restart Cumulus -- if it was done recently, look in the 'backup' folder under Cumulus as you could copy the alltime.rec file from a recent backup into the Cumulus 'data' folder (shut down Cumulus before copying it)

You can also remove the offending readings from the log files -- dayfile.txt and the Monthly Log files. Also, check today.ini and yesterday.ini

See the Wiki for the structure of these (again, quit Cumulus to edit these)

http://wiki.sandaysoft.com/index.php?ti ... :Log_Files

[Blackmyre]

This would assume you knew what the previous highs were for your gust and rain rate

That would be a wildly optimistic assumption in my case Good point about the automatic backup though - that's worth a look.

I think there might be something I need to do about resetting the rain total or something - ISTR having to tweak it before after disturbing the rain gauge but it was a while ago.

[clas]

Hi
I have a similar problem, I just started using cumulus today and downloaded 3 months of data from my weather station (WH1080). I found that there has been some strange values recorded from time to time witch resulted in extreme all time high values. I tryed to fix this by changing the value in the dayfile for the recorded date, and remove the line with the erroneus data from the month file, but the value in all time high does not change when I start cumulus again, is there a way to make it go throug the month files again?

[steve]

but the value in all time high does not change when I start cumulus again, is there a way to make it go throug the month files again?

No - only 'lowest max' and 'highest min' temperatures are extracted from the log file. The others you have to edit using Edit | All time records.

[daj]

... is there a way to make it go throug the month files again?

Hi clas

I had a similar problem a few months ago so what I did was take a copy of the dayfile.txt in the Cumulus 'data' folder. Import the copy into Excel (or another spreadsheet) and then sort one column at a time to get the high and low figures for each reading. Then manual types those into the All Time Records within Cumulus (Edit, All Time Records). It only takes a few minutes.

The format of the dayfile.txt is here ... http://wiki.sandaysoft.com/index.php?title=Dayfile.txt

[steve]

Yes, that's the way to do it. It's the sort of thing that Cumulus could do for you, of course, if only I wrote the code...

[clas]

but the value in all time high does not change when I start cumulus again, is there a way to make it go throug the month files again?

No - only 'lowest max' and 'highest min' temperatures are extracted from the log file. The others you have to edit using Edit | All time records.

Ok, tnx, i used excel to find the max rain/day and max rain/h in the month files and edited the All time records. All looks great now.

[Blackmyre]

Damn, it's done it again - We did not have a 100.0mph wind at 22:35 last night!

Does anyone have any idea why such oddities might have started to creep in? I've also noticed quite a lot of "Lost sensor contact!!!" errors of late, but I got the impression that these were new reporting in a recent version of Cumulus rather than a new fault. Looking at the log, though, there were a string of them last night with the last one being just before the spurious 100mph wind speed...

[daj]

Now that it very interesting as I had the same problem last night -- I have had no wind recorded since 18 December as wind sensors are frozen solid, however last night at around 22:40 I had a gust of 39 mph. I edited it out, but I also lost contact with my sensors for the best part of an hour on and off. (which is very rare for me).

I see we are not very far from each other -- I wonder if there was a cosmic blip (I am sure there is a better explanation, but I mean some kind of electrical/atmospheric interference)

Last time I lost contact, several other people (all over the UK) reported a similar issue.

[dane]

Did you hit an extremely low temperature last night?
Are the batteries in your transmitter getting old and running low?
Such a combination could explain the strange spurious data.

[Blackmyre]

Hmmm... Very interesting. My pattern of lost connections seems rather different from yours though, so I'm inclined to think it's more likely to be a coincidence.

The log displayed by Cumulus starts at 18:29:25 yesterday and reports an error every 10 seconds for a few minutes, then there's a gap of 7min20 and one of 21 seconds. It then settles into a repeating sequence of 9,10,11 seconds with the odd 21 or 10 second gap thrown in. That goes on until about 18:50, when there's an 8min14 gap followed by a bunch of 10 and 20 second gaps. This behaviour continues with a ~25 minute gap around 19:34 and a 2hr40 gap around 22:16. After that it's 10 & 20 second gaps through to 22:34:56, with the spurious 100mph wind at 22:35. There are no errors after that until a single one at about 03:10 this morning.

Regarding the frozen wind sensor, I had that a few times around Christmas. I thawed it with a hair dryer then sprayed it liberally with WD40. I did that on perhaps three or four occasions, but gave up when it next seized, only to find it freed itself fairly quickly despite the sub-zero temperatures. It's been fairly good since then, only seizing very occasionally and freeing itself fairly quickly.

[Blackmyre]

Did you hit an extremely low temperature last night?
Are the batteries in your transmitter getting old and running low?
Such a combination could explain the strange spurious data.

Good point - I don't think I've changed the batteries since I bought the station just over a year ago. Certainly worth changing them, and I guess they might be the cause of the lost connections too. Don't know why I didn't think of that

Temperature last night only dropped to -1.9C, and around the time of the spurious 100mph wind it was close to zero. However, there's a suspicious looking sudden drop and flatline for 15 minutes or so immediately before the spurious reading before resuming the previous line (albeit with an brief very low wind chill, unsurprisingly!).

Thanks for the suggestion.