Extrange average temp

[yv1hx]

Hi folks!

I'm getting this strange graph of temperature. The blue line corresponds to the average temperature and I have no idea of the origin of such behavior. (I live in a tropical coastal area, the lowest recorded temperature was +16C a few years ago).

The trends for minimum and maximum temp behave within the observed by the weather station..

Graphs can be observed at:

http://www.yv5fih.org.ve/red/Cabimas/trends.html

BTW, My station is LaCrosse WS-2310 in wired mode, records captured every minute, Cumulus version 1.8.8 Build (881).

[steve]

The min, max and average temperatures are plotted from the corresponding values in dayfile.txt, so that's the first place to start. If you want to attach it here, I'll have a look. It's probably a good idea to also attach your april and may data files.

The daily average temperature is calculated by sampling the temperature once a minute, calculating the sum of all of the samples and dividing by the number of samples. If you stop Cumulus, the current total and number of samples are written to today.ini, and read back in again when you start Cumulus up again. What is your pattern of usage of Cumulus? Do you leave it running all the time, or do you stop it every night and start it in the morning, or stop it just occasionally, or...?

You're using an old version of Cumulus, but I don't remember any problems with the average temperature calculation in 1.8.8. What is the average temperature on your Cumulus screen showing right now?

Your graph also stops at 13th May, which is strange, and probably related to this problem.

[yv1hx]

Steve, Sorry for my late reply,

The min, max and average temperatures are plotted from the corresponding values in dayfile.txt, so that's the first place to start. If you want to attach it here, I'll have a look. It's probably a good idea to also attach your april and may data files.

The requested files are attached!

The daily average temperature is calculated by sampling the temperature once a minute, calculating the sum of all of the samples and dividing by the number of samples. If you stop Cumulus, the current total and number of samples are written to today.ini, and read back in again when you start Cumulus up again. What is your pattern of usage of Cumulus? Do you leave it running all the time, or do you stop it every night and start it in the morning, or stop it just occasionally, or...?

The usage pattern is fulltime, except when I have power outages at home…

That is the answer to the mystery. Since January in our country we are suffering scheduled power cuts. We have a scheduled power outages Monday to Friday from 12:00 to 14:00 and the weekends usually from 15:00 to 17:00.

Obviously, when the power goes off, Cumulus don’t have time to write the last values in the. Ini file.

However, I have other station under test, running the same Cumulus version and doesn’t exhibit the same estrange graphs.

http://www.yv5fih.org.ve/red/Bomberos_C ... rends.html

Both stations are at the same location by now. Although they show different locations on the web pages.

You're using an old version of Cumulus, but I don't remember any problems with the average temperature calculation in 1.8.8. What is the average temperature on your Cumulus screen showing right now?

I should move on :)

Your graph also stops at 13th May, which is strange, and probably related to this problem.

It was a old print. The updated ones always are at the following link:
http://www.yv5fih.org.ve/red/Cabimas/trends.html

[steve]

Your average temperature is actually OK - it's the green line, not the blue line. The blue line is the minimum temperature, and you're getting some 'bad' values from somewhere, e.g:

11/05/10 0,4 at 11:28
12/05/10 0,8 at 11:43
13/05/10 0,8 at 11:11
17/05/10 1,0 at 19:20
19/05/10 0,2 at 08:15
24/05/10 1,0 at 14:29

Most of these don't appear in your May log, but this one does:

11/05/10;11:29;0,4;65;-5,5;4,3;12,2;228;0,0;0,0;1012,1;45,2;32,4;46;10,1;0,4;0,4

This suggests that you are getting brief data corruptions which usually correct themselves straight away so don't often appear in the log, but do affect the daily low value.

[yv1hx]

Thanks for your help, Steve.

What will be the source of these readings? Maybe the weird data protocol of the LaCrosse WS23xx??

[steve]

It looks like you are also getting wind speed 'spikes'. Are you using a serial/USB adapter?

[yv1hx]

Nope!

All of my PC have RS-232 serial ports built-in.

So I have the WS-23xx directly connected, without any USB converter boxes.

[steve]

The WS23xx is apparently very susceptible to RF interference, so perhaps that is the explanation?