Filtering out ridiculous LaCrosse temperatures?

[Gordon-Loomberah]

About twice per week my WS2350 produces a temperature reading of -30C. This is not a problem with the Serial-USB converter, which seems to work perfectly well, as the silly temperature is recorded as a minimum temp in the station. In the past the station has also produced ridiculously high temp readings too, but none since I started using Cumulus about a month ago.
I'm away at work searching for NEOs for a bit over a week at a time, so cant always edit these out when they occur, so was wondering if there was any way to filter out this LaCrosse bug? Clearly the real temperature cant drop 50C in 1 minute! I'm recording data every minute, and the low temp reading only lasts a few minutes at a time. It also wipes out the real record minimum, which has to be searched out through the logs and re-entered. I suppose I should keep a separate manual list of records to save a lot of time and effort.

Today's new "record low temp" can be seen here:

http://members.ozemail.com.au/~loombera ... trends.htm

Great software Steve, it is so far ahead of that waste of HD space called Heavy Weather!

I'll be making a donation as soon as I can figure out how to make one via paypal from my credit card... that just isnt offered as an option when I click on the donate button.

thanks, Gordon

[steve]

About twice per week my WS2350 produces a temperature reading of -30C. This is not a problem with the Serial-USB converter, which seems to work perfectly well, as the silly temperature is recorded as a minimum temp in the station. In the past the station has also produced ridiculously high temp readings too, but none since I started using Cumulus about a month ago.

Gordon, Cumulus currently filters out high extremes from La Crosse (anything over 60C), so that probably explains why you're not seeing the high ones with Cumulus. The 'low' filter operates at -100C, so that's why it's not helping in your case. In Cumulus 2, I plan (eventually) to have user-configurable filters, and also hopefully some more 'intelligent' filtering, i.e. based on the recent readings.

I'm away at work searching for NEOs for a bit over a week at a time, so cant always edit these out when they occur, so was wondering if there was any way to filter out this LaCrosse bug? Clearly the real temperature cant drop 50C in 1 minute!

There's an undocumented facility that may help. Stop Cumulus and edit the cumulus.ini file. In the [Station] section add a line:

EWTempdiff=x.y

Where x.y is the maximum difference in degrees that you are prepared to accept between successive data readings.

I'll be making a donation as soon as I can figure out how to make one via paypal from my credit card... that just isnt offered as an option when I click on the donate button.

On the left hand side of the screen, where it says "Don't have a PayPal account? Use your credit card or bank account (where available). Continue". You should be able to enter an amount, then click on the "continue" link to take you to a page where you can enter your card details.

Steve

PS Your job sounds much more interesting than mine! I spend most of my time searching for bugs in somebody else's code.

[Gordon-Loomberah]

I'll add that line when I get home tomorrow. Another -30 this morning to mess up my stats...

On the left hand side of the screen, where it says "Don't have a PayPal account? Use your credit card or bank account (where available). Continue". You should be able to enter an amount, then click on the "continue" link to take you to a page where you can enter your card details.

Thanks, it worked fine. Previously I have tried to pay for things without logging in, but paypal would not let me do it.

PS Your job sounds much more interesting than mine! I spend most of my time searching for bugs in somebody else's code.

Yes its good, although the hours suck sometimes, especially in winter!

Gordon

[Gordon-Loomberah]

I added that line to the bottom of cumulus.ini a couple of days ago, setting it to 3.0, but yesterday I had 4 minutes of -29.3, so it didnt filter them out.
I've since found there are a couple of other EW... lines, so moved it to just under them. I hope that works!

Gordon

[steve]

It needs to be somewhere (anywhere) in the [Station] section, not at the bottom of the file. You've probably now put it after EWInterval and EWFile, which just happen to be in the [Station] section, so it should work now.

[Gordon-Loomberah]

Ok thanks Steve, thats where I had moved it to. Somehow I managed to miss that vital piece of info, I guess I should have consulted your post immediately before going ahead and doing it from memory!

Gordon

[Gordon-Loomberah]

Steve, is there a similar line I can use to filer out ridiculous humidity readings? Not as frequent as the -30 temps, but occasionally there is a 0% humidity reading for a few minutes. Strangely, the dew point doesn't seem to drop at the same time!

thanks, Gordon

[steve]

No, I'm afraid not. I should really filter out humidity readings of zero (I've found my Fine Offset produces them). When you have Cumulus set to calculate the dewpoint, it checks for zero (as otherwise the calculation would fail), but perversely I still allow zero as the humidity reading. In Cumulus 2 I'm filtering them out - I'll put something in the next build to filter them out in the current version.

[Eugen-Oradea]

There's an undocumented facility that may help. Stop Cumulus and edit the cumulus.ini file. In the [Station] section add a line:

EWTempdiff=x.y

Where x.y is the maximum difference in degrees that you are prepared to accept between successive data readings.


I would really like to try this to see if it works.
Is it OK if I write it like this:
[Station]
EWTempdiff=5
Type=6
Model=WS2300
Port=7
Latitude=47,0666666666667
Longitude=21,9333333333333

I'd be happy if it'd work.
Thanks, Eugen, Oradea

[steve]

I would really like to try this to see if it works.
Is it OK if I write it like this:
[Station]
EWTempdiff=5

Yes, that should work, and should ignore anything more than 5 degrees different from the last 'good' value.

[Eugen-Oradea]

So far it seems to work, in the recent hours I didn't notice any abnormal data.
I only hope it will keep that way.
Thanks again, I was really impressed by your fast response to all of my questions.
Eugen, Oradea, Romania.