Error in Cumulus Highest Hourly Rain Record

[dids04]

Good Evening

It appears an error may exist in how cumulus calcs the highest hourly rain rate.

heavyweather2.png

cumulus.png

As per the images attached, cumulus is taking 3 readings to work out the hourly rate. (perhaps because my setting is for 20 min. readings??)

Problem is that it takes the record at 1.40,2.00 & 2.20 then works out the difference from last to first to get a figure. (which is only a 40min reading 1.40am-2.20am)

Shouldn't it calculate from 1.40am reading to 2.40am reading to get the true hourly reading (thereby taking into account 4 records)

Hope this makes sense. Records in question relate to 1.40am - 2.40am on 1 December 2012.


Any help appreciated.


Regards


Steve

[steve]

It doesn't work it out in terms of logger entries, it uses the same algorithm as it does when it's running 'live'. It keeps a list of all of the data readings in the last hour and works out the difference between the latest and earliest rain counters in that list. The check that it does is that the time of the first entry is less than or equal to an hour before the time of the last entry. So, in theory, at 2.20 am it should be doing the calculation based on the difference between the counter at 2.20 am and the counter at 1.20 am, so your high figure should 12.4 at 2.20 am.

For some reason, the system routines which do the time comparison aren't including the 1.20 am reading at 2.20 am. I think this is probably down to rounding errors. I'll investigate, but in the meantime could you please zip up the diags folder and attach it, so I can check the timestamps that Cumulus was seeing. When running live, this isn't so much of an issue, as differences of only seconds are involved, but with logger data, and in particular with your relatively long logging interval, it becomes more significant.

Edit: I've just checked the date and time comparisons, and they appear to work correctly; i.e. when the 2.20am reading is added to the 'last hour' list, the 1.20am should become the first entry in the list. So I don't know why it apparently didn't give the correct result for you. Hopefully the diags logs will throw some light on it. Could you also attach your December log file, please. You haven't got 'ignore LaCrosse clock' set, by any chance?

[dids04]

Hi Steve

December Log File & Diags Folder attached.

Also, I checked Cumulus and ignore LaCrosse Clock is NOT checked


Cheers

Dec12log.txt

Diags.zip

[steve]

I'm baffled, I'm afraid. I've looked at the code and it looks OK. I've looked at the results on two different types of weather station and it appears to work OK. But I don't have a La Crosse station. It's possible that there's a bug in the code for that station, or it's something to do with the calculated timestamps for that station, but I can't see the problem in the code. I suppose it's possible that I've changed something since the build that you're using.

I'm going to put some extra logging to the diags files in the next build of 1.9.3; all I can suggest is that you try that and we can see if the extra logging gives an insight into what's going on.

[steve]

Sorry, I meant to post here that I had added the extra diagnostics, but forgot. The download link is in this thread - https://cumulus.hosiene.co.uk/viewtopic.php?f=2&t=7341 - if you want to try it. The extra information is logged to the diags file during logger data processing.