1.8.8 b855 Cloudbase and dew point going weird.

[Alcedema]

Just in case this hasn't been pointed out already, I was seeing straight lines for dew point on my graphs, so investigated and found that cloudbase was a negative number.

Heres a portion of the log file where it just started to create a straight graph, which then went on for a few hours.

Code: Select all

05/10/09,18:59,14.8,99,14.6,0.0,0.0,0,0.0,9.3,1008.1,198.9,22.2,55,0.0,14.8,14.8
05/10/09,19:00,14.8,99,14.6,0.0,0.0,0,0.0,9.3,1008.2,198.9,22.2,55,0.0,14.8,14.8
05/10/09,19:01,14.8,99,14.6,0.0,0.0,0,0.0,9.3,1008.2,198.9,22.2,55,0.0,14.8,14.8
05/10/09,19:02,14.8,99,14.6,0.0,0.0,0,0.0,9.3,1008.2,198.9,22.2,55,0.0,14.8,14.8
05/10/09,19:03,14.8,99,14.6,0.0,0.0,0,0.0,9.3,1008.2,198.9,22.2,55,0.0,14.8,14.8
05/10/09,19:04,14.8,99,14.6,0.0,0.0,0,0.0,9.3,1008.3,198.9,22.2,55,0.0,14.8,14.8
05/10/09,19:05,14.8,100,14.6,0.0,0.0,0,0.0,9.3,1008.3,198.9,22.2,55,0.0,14.8,14.8
05/10/09,19:06,14.8,99,14.6,0.0,0.0,0,0.0,9.3,1008.2,198.9,22.2,56,0.0,14.8,14.8
05/10/09,19:07,14.8,100,14.6,0.0,0.0,0,0.0,9.3,1008.2,198.9,22.2,55,0.0,14.8,14.8
05/10/09,19:08,14.8,100,14.6,0.0,0.0,0,0.0,9.3,1008.2,198.9,22.2,56,0.0,14.8,14.8
05/10/09,19:09,14.8,100,14.6,0.0,0.0,0,0.0,9.3,1008.2,198.9,22.2,55,0.0,14.8,14.8
05/10/09,19:10,14.8,100,14.6,0.0,0.0,0,0.0,9.3,1008.2,198.9,22.2,56,0.0,14.8,14.8
05/10/09,19:11,14.8,100,14.6,0.0,0.0,0,0.0,9.3,1008.2,198.9,22.2,56,0.0,14.8,14.8
05/10/09,19:12,14.8,100,14.6,0.0,0.0,0,0.0,9.3,1008.3,198.9,22.2,55,0.0,14.8,14.8
05/10/09,19:13,14.8,100,14.6,0.0,0.0,0,0.0,9.3,1008.4,198.9,22.2,56,0.0,14.8,14.8
05/10/09,19:14,14.8,100,14.6,0.0,0.0,0,0.0,9.3,1008.4,198.9,22.2,56,0.0,14.8,14.8
05/10/09,19:15,14.8,100,14.6,0.0,0.0,0,0.0,9.3,1008.4,198.9,22.2,56,0.0,14.8,14.8
05/10/09,19:16,14.8,100,14.6,0.0,0.0,0,0.0,9.3,1008.3,198.9,22.2,56,0.0,14.8,14.8
05/10/09,19:17,14.8,100,14.6,0.0,0.0,0,0.0,9.3,1008.4,198.9,22.2,56,0.0,14.8,14.8
05/10/09,19:18,14.8,100,14.6,0.0,0.0,0,0.0,9.3,1008.4,198.9,22.2,56,0.0,14.8,14.8
05/10/09,19:19,14.8,100,14.6,0.0,0.0,0,0.0,9.3,1008.4,198.9,22.2,56,0.0,14.8,14.8
05/10/09,19:20,14.8,100,14.6,0.0,0.0,0,0.0,9.3,1008.4,198.9,22.2,56,0.0,14.8,14.8
05/10/09,19:21,14.8,100,14.6,0.0,0.3,90,0.0,9.3,1008.4,198.9,22.2,56,0.3,14.8,14.8
05/10/09,19:22,14.8,100,14.6,0.0,0.3,111,0.0,9.3,1008.5,198.9,22.2,56,0.3,14.8,14.8
05/10/09,19:23,14.8,100,14.6,0.1,0.3,187,0.0,9.3,1008.4,198.9,22.2,56,0.3,14.8,14.8
05/10/09,19:24,14.8,100,14.6,0.1,0.3,193,0.0,9.3,1008.5,198.9,22.2,56,0.3,14.8,14.8
05/10/09,19:25,14.8,100,14.6,0.1,0.3,205,0.0,9.3,1008.5,198.9,22.3,56,0.0,14.8,14.8
05/10/09,19:26,14.8,100,14.6,0.1,0.3,205,0.0,9.3,1008.5,198.9,22.3,56,0.0,14.8,14.8
05/10/09,19:27,14.8,100,14.6,0.1,0.3,205,0.0,9.3,1008.5,198.9,22.2,56,0.0,14.8,14.8
05/10/09,19:28,14.8,100,14.6,0.1,0.3,205,0.0,9.3,1008.6,198.9,22.3,56,0.0,14.8,14.8
05/10/09,19:29,14.8,100,14.6,0.1,0.7,197,0.0,9.3,1008.5,198.9,22.3,56,0.7,14.8,14.8
05/10/09,19:30,14.9,100,14.6,0.2,0.7,190,0.0,9.3,1008.6,198.9,22.3,56,0.3,14.9,14.9
05/10/09,19:31,14.8,100,14.6,0.2,0.7,204,0.0,9.3,1008.7,198.9,22.3,56,0.7,14.8,14.8
05/10/09,19:32,14.8,100,14.6,0.2,0.7,210,0.0,9.3,1008.7,198.9,22.3,56,0.3,14.8,14.8
05/10/09,19:33,14.9,100,14.6,0.2,0.7,211,0.0,9.3,1008.6,198.9,22.3,56,0.7,14.9,14.9
05/10/09,19:34,14.8,100,14.6,0.3,0.7,211,0.0,9.3,1008.7,198.9,22.3,56,0.3,14.8,14.8

[steve]

Your humidity is 100%, so your dew point will be approximately equal to your temperature, which it appears to be, and in that case you ought to see a cloudbase close to zero. For the cloudbase to go negative, the dew point would need to be higher than the temperature, which it isn't in the sample you've shown. It could be a rounding error - was it a very small negative number? Does the log file show the dew point to be higher than the temperature at any point? Are you using any calibration settings in Cumulus?

It's actually quite reasonable for the dew point to stay constant for some time, particularly when you have no wind, as is the case here. It's the same air you're sampling, so you would expect it to have the same dew point.

[steve]

I've realised why your dew point is getting stuck. Your humidity is going over 100%, either because the station is giving that value, or you have a calibration value set. Cumulus does a temporary calibration of the humidity to work out the dew point; if the result is over 100 it doesn't try to work out the dew point (it continues to use the last good value). Eventually the temperature falls below the last good dew point value and the cloudbase goes negative.

I'll make sure the temporary calibration can't result in a humidity > 100 (like the 'main' calibration code does) and think about what to do in the other case, where the station supplies a value > 100.

[steve]

Further investigation reveals that the bug is much simpler than I thought - it stops calculating the dew point when the humidity gets to 100% (as it did in your case), when it should really be setting the dew point to the same as the temperature. I'll fix this when I get back next weekend.

[Alcedema]

That's great. I would have helped sooner but I had the wrong email address set in the account.

Even though it appears you have found the reason from it, I'll still post the graph I saw which made me realise something was wrong.



Hope you had a good holiday.

-Tim

[Alcedema]

Also from memory, the previous time I noticed it the cloudbase was about -60m. The calibration of humidity was set to 7, which was roughly correct when I set up the station, so I'm assuming it must change throughout the year.

[steve]

I was only thinking about this just now, while out with the dogs. I know I've done the changes for this, so just need to package it up, but I had a feeling there was another fix to be included, but I can't remember what it was!

If anyone is expecting a fix for something, please let me know!

[steve]

I've uploaded the build with this fix now:

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

[Alcedema]

That's great, thanks Steve.

[Alcedema]

Using that build, I am still seeing flat lines with dew point, this time they are staying above current temperature..

http://weather.a3n.co.uk/images/temp.png?1257028530608

[steve]

It's stopped calculating the dew point at 100% humidity again. I changed the code which handles logger data but somehow forgot to change the 'live' code. Or, I did change it but the changes got lost. Anyway, apologies for that, I'll have a new build available soon.