Cumulus Support

If I close Cumulus 1.9.3 1055, wait about 10 seconds or so, and reopen it, the wind run jumps ahead about 3 miles. Here is a video of this issue. Jump to 1:13 for the updated wind run total to appear...

http://www.youtube.com/watch?v=ywyiN2QFeSk

It's because the latest logger entry is read and processed at start up (note the number of processed entries in the task bar), an amount is added to the wind run corresponding to the speed in that entry (which is a huge approximation to begin with). I'm afraid there's nothing I can do about this as Fine Offset stations don't timestamp their logger entries. The only option would be to ignore the latest entry, and then in some circumstances the wind run figure would under-read rather than over-read, and other data would be missed.

steve wrote:Fine Offset stations don't timestamp their logger entries

Actually, this bit is probably irrelevant. It's this bit:

an amount is added to the wind run corresponding to the speed in that entry (which is a huge approximation to begin with)

which is the real cause, because it doesn't use any logger entries that are calculated to be from before the time the today.ini file was saved (with the wind run in it). The extra 3 miles are due to the addition of the amount corresponding to the timespan of the logger entry based on the wind speed in that entry. But the actual figure used is the average speed, so if you have Cumulus set to calculate that, it complicates things somewhat.

Note that the windrun figure doesn't continue from the point where you closed it down, it continues from the last time it was written to the log file (as with all data). The figure for the period since then comes from the logger data.

If you want to zip up the diags folder and attach it, with the backup today.ini which was saved after you restarted Cumulus, I can look for the entry corresponding to your video to see where the figure came from.

Files attached.

Maybe I am missing something, but why not have Cumulus read the last backup file before it tries to update from the station?

Buford T. Justice wrote:Maybe I am missing something, but why not have Cumulus read the last backup file before it tries to update from the station?

I'm not sure I understand you. It reads the today.ini file, which contains the state of the data at the time of the last entry in the Cumulus logs before it was closed down previously. It knows from the data in today.ini how far back it needs to go in the station's logger. It makes sure to go back far enough to cover the entire missing period.

Yours had last logged at 05:11 with a wind run of 83.9. It then looked for the logger entries to cover the period that it missed while it wasn't running. Your station's logger is set to 15 minutes, so it used that entry and added in an amount of wind run corresponding to 15 minutes at 11.4 mph = 2.8 miles. At 05:14 it then added in an amount for the previous minute at a speed of 7.6 mph = 0.12 miles, so your total is now 86.9.

So, because your logger is set to 15 minutes, but Cumulus had itself logged only a minute ago at 05:11 (because you had restarted it several times; ordinarily it would have logged at 05:00), you were unfortunate in effectively getting 14 minutes of duplication on the wind run.

I suppose it could just add in a pro-rata amount for the wind run, but given that this is a slightly unusual situation, and the wind run calculation from logger data isn't accurate in the first place, that would be rather like putting lipstick on a pig.

I noticed it happened during a normal closing of the program and reopening it. I had to repeat it a few times to get a video of it for you. I probably ought to set the logger on the station to something small. I'd have to dig around to find out what the smallest possible limit is I can set it to.

You always stand the chance of getting slightly odd results if you close Cumulus and then start it up again immediately, unfortunately, particularly with Fine Offset stations; and the chances of getting the odd results are increased the bigger the logger interval. It's all something of a compromise.

If you use my utility (on the download page) you can set the logger interval to one minute, which gives you just under three days worth of data. I'm not sure I'd recommend that interval, though, as Easyweather only allows you to set it down to 5 minutes, and there may be a reason for that.

I just ran your utility and sat it to 5 minutes. We'll see how it does.

We are in the middle of a little blizzard here in Illinois. I have my unit on a tripod roof mount on a thick poll about 7' long with the smaller pole for the sensors clamped to that. That wind gets to blowing and it shakes it just enough to trigger a bucket dump on the rain gauge every once in awhile.

I've modified the logger data wind run calculation in the latest build. There is still likely to be a difference between the wind run figure shown when last closed down compared to when it starts up again (because it is recalculated for the period since the last time Cumulus logged an entry), but it shouldn't show such a large change as the one you were seeing.

I just installed it. Thanks again for all your effort

UPDATE: The wind run jumps are nowhere near as bad as before.

I saw this on page 29 and am now using 1 minute intervals...

http://site.ambientweatherstore.com/Manuals/ws1090.pdf

Seems safe as far as I can tell.