oddball 'extreme' readings

[stevew]

running a 2350 for about 4 or 5 months now have been using wuhu to update WU (http://www.wunderground.com/weatherstat ... =IBERKSRE4) site while I found time to start my own site, which now I have. Having toyed with using wuhu to write a curdata.lts and then upload, parse, report and graph that for my site I came across Cumulus.

Excellent work here Steve, well done.

I really only started playing with it last night but have easily been able to see how the web pages are created and been able to fiddle with them (http//weather.60redwood.com).

I have noticed, twice (last night and this morning) odd 'recent extremes' for temperature(53.6c and last time was 213c - approx from memory) and wondered if anyone else had seen this or had any ideas what it might be?

[steve]

Your rain seems to be problematic too. How is your station connected to your PC?

Steve

[stevew]

serial to usb and appears as com14. haven't noticed issues previously with wuhu though (but then I haven't gone and looked in detail yet either)

interestingly I have cumulus logging to WU and there is no 'odd' reading there, though that doesn't really prove anything one way or another.

Do you suspect the serial/usb 'magic' as a suspect? The driver was the one provided with the station, alas I don't have a pc in my kit of sufficient vintage to try a proper serial port

if the problematic rain you refer to is the '77mm' that's me, I set that value as the start annual rainfall

[steve]

interestingly I have cumulus logging to WU and there is no 'odd' reading there, though that doesn't really prove anything one way or another.

The bad readings are transient, so it's only because Cumulus keeps highs and lows that you get to see them, unless they happen to coincide with a WU upload.

Do you suspect the serial/usb 'magic' as a suspect? The driver was the one provided with the station, alas I don't have a pc in my kit of sufficient vintage to try a proper serial port

If you look in the forum here, and elsewhere on the 'net, you'll see quite a few examples of invalid data with La Crosse stations. Virtually all of these are where serial/USB converters are in use. I can't explain why you haven't seen these problems with WUHU - unless it is because WUHU doesn't keep extremes, so you have to be unlucky to hit a bad reading at the time of an upload. I don't know anything about WUHU. I did have a quick glance at the site though, and it appears that WUHU tries to time its reads when the station isn't communicating with the remote sensors (a strategy which has been suggested by someone here). It isn't clear whether WUHU is doing this to try to avoid the bad data.

if the problematic rain you refer to is the '77mm' that's me, I set that value as the start annual rainfall

You seem to have some unlikely values in your all-time rainfall records.

Steve

[stevew]

Thanks Steve, yes I have heard of the issues previously, wuhu does have some filtering going on and no doubt some other tricks but I tend to agree that more by luck than judgement I haven't noticed the issue before. I will do some research on other threads to see what else has been found. Changing weather station isn't an option but as my goal is to use cumulus to provide the data feed to the web site I could look at the parsing on the site removing wildly varying values to smooth the issue out (not ideal, but not unheard of when using sensor based measurement- often there is some data smoothing/extreme cleansing)

[steve]

I'm happy to try things to try to improve the quality of the data (I already do a bit of "obvious" filtering) but I'd like to avoid having to start averaging values etc to deduce bad data. The problem doesn't appear to be related to the sensors, it's simply bad communication between the console and the PC.

Any information you can supply on the conditions under which this happens would be useful. For example, is it data downloaded from the logger at startup, or read during normal running, or both?

Steve

[stevew]

the data logger was my first 'easy' target so just cleaned up all the cumulus data (deleted it all) and turned off the logger which was active, set it going and had a dodgy temp reading of 30.6 today at 16:51

In your previous exposure to this type of issue was it the PC/Logger coms in general or specifically using a usb/serial converter?

[steve]

the data logger was my first 'easy' target so just cleaned up all the cumulus data (deleted it all) and turned off the logger which was active, set it going and had a dodgy temp reading of 30.6 today at 16:51

So this was a 'real time' reading, i.e. Cumulus was running at 16:51 and had been running for a while previously?

In your previous exposure to this type of issue was it the PC/Logger coms in general or specifically using a usb/serial converter?

Specifically using a usb/serial converter. I'm not sure whether it's been established that errors occur with both logger data and real-time data - but I have a feeling that they do.

Steve

[stevew]

yes data logger mode off, cleaned up all data, restarted, error as previous post and another of -21.2 at 17:04

[steve]

Maybe it would be worth trying to avoid reading the data while the console is talking to the remote sensor. I'll see what I can do.

Steve

[steve]

I've uploaded build 756 which attempts to avoid reading the data when the station is talking to the remote sensors. Please give it a try and see if it makes any difference.

Steve

[stevew]

Thanks Steve, I'll download it and set it running today and report back.

[stevew]

the download still gives me build 726? which is the same as previous one

http://sandaysoft.com/download/CumulusSetup.exe

[steve]

It's a beta, see announcements and news section in the forum.

http://www.nybbles.co.uk/downloads/beta ... sSetup.exe

Steve

[stevew]

thanks