Trends Page - Min. Temp, Max. Temp, Avg. Temp graph...

[Polaris]

Hello,

Nothing serious, just a curious question. On my Trends page, the graph that shows the Min, Max, and Avg Temps always shows the Min. Temp starting off with 0 deg. C, even though I haven't had the temperature reach that low yet. (although I'm certain its coming soon, unfortunately! )

Is that normal, or does my page have a glitch? If so, how can I fix it, as I would prefer it to show the actual Min as recorded during the past 12 hours....

[steve]

This is from the first day you used Cumulus. Sometimes odd values can be written to dayfile.txt on the first day, and since typically it's not a complete day so the values aren't so useful anyway, it's probably best to delete that day.

[MrBodil]

I went on a vacation last week of august. I unplugged the weatherstation and shut down my pc. When i got back a week later, i started it up again and cumulus downloaded the data. Something got really wrong at the trend-page. It tells me that it had been 0 degrees from august 10 to august 20 when something really weird happend. 1000 degrees celsius min temp and -500 as max...?
August 25 it's back on 0 degrees, and still is... I cant remember if it looked like that before the vacation 21 august, but something is really wrong here. Any suggestion how to solve it?

http://www.samone.se/cumulus/trends.htm



//Magnus

[steve]

You will have some bad entries in dayfile.txt, which can happen for various reasons. If you edit your dayfile.txt with Cumulus stopped, you should be able to spot the faulty entries and remove or correct them. If you zip up your diags folder and attach it, I may be able to see what happened to cause this.

Those entries that look like zero aren't actually zero, it's just the scale of the graph caused by the very high and very low figures.

[gemini06720]

Steve.

Writing about trends...

I was about to send you a personal message asking you to include more trends in version 2 of Cumulus - so far, and please correct me if I am wrong, the only trends produced by Cumulus are a temperature trend and a pressure trend.

I would really appreciate if trends for things such as the dew point, the wind chill, the humidity, the humidex or the heat index were added ... please...

[steve]

Magnus - I've looked at the diags files you sent me, and Cumulus was actually running for the period with the bad data - 20th-25th August. Unfortunately the station disconnected from the USB at about 1640 on the 20th, so for the whole of the rest of that run, Cumulus read no data; this is why those days have the 'default' values in the dayfile. If you'd noticed, you could have recovered by using the backup from the 20th. Cumulus would then have downloaded the missing data. If that backup hasn't been overwritten yet, you could still use it to recover.

[steve]

Writing about trends...

I was about to send you a personal message asking you to include more trends in version 2 of Cumulus - so far, and please correct me if I am wrong, the only trends produced by Cumulus are a temperature trend and a pressure trend.

This thread isn't actually about those sort of trends, if I understand you correctly. The confusion comes because Mrs Sandaysoft chose to call the graphs page 'trends' when she produced the web pages. So we're actually talking about graphs here, but anyway...

I would really appreciate if trends for things such as the dew point, the wind chill, the humidity, the humidex or the heat index were added ... please...

Pressure and temperature are the ones usually quoted as being useful, but I don't see why most of those couldn't be done as well. I'm not sure about how you produce a trend for wind chill as it is likely to be fluctuating all the time, and the simple calculation used for pressure and temperature won't really be suitable.

[MrBodil]

Magnus - I've looked at the diags files you sent me, and Cumulus was actually running for the period with the bad data - 20th-25th August. Unfortunately the station disconnected from the USB at about 1640 on the 20th, so for the whole of the rest of that run, Cumulus read no data; this is why those days have the 'default' values in the dayfile. If you'd noticed, you could have recovered by using the backup from the 20th. Cumulus would then have downloaded the missing data. If that backup hasn't been overwritten yet, you could still use it to recover.

Thanks for your replys in this!

In the backup-folder, there is backup from 20090829104846. There is several folders but no one seems to be from any earlier date. Strange... It also seems that they all have these wrong data for the current days. Can i "just" erase data between 20-24, or adjust them? What causes the graphs malfunction? It does not work properly since then.

//Magnus

[steve]

As you don't have a backup from before the problem, it's a bit harder to 'rewind', but it's possible. You would have to delete the September log, and edit the August log to take out all of the data back to where the problem started on the 20th. Then edit today.ini to change the timestamp back to the corresponding date and time. Then start Cumulus. This won't be perfect, but will be close. It also relies on you having all the data still in the data logger, and it's about 23 days ago now, so that would need you to have your station logger at 10 minutes or more.

Or you could just delete the bad data and carry on.

The graphs look strange because the high and low values are changing the scale so the real values are not significant.

[MrBodil]

As you can see by the pic above, i have just deleted the days with the weird data. I do have the data left in EasyWeather so it means that i am able to "restore" it manually (i guess?)

Thanks for your help!

//Magnus