Negative value data, Evap Year Roll over

[nossis]

A question...
I believe that this would be for our master guru, illustrious legendary programer... Steve (with great applause!)

I have found that on the last day of 2013, the Evap data in the log files at the turn of midnight went from a positive value to a big negative value. The one thought I have is that the station roll over time is at 9am, so this is what may have confused it. Last day of the year being midnight but station roll over on the next day at 9am 1st of Jan.

To be more specific this is what happened.... (in file Dec13log.txt)
(Column 00 Date, Column 01 Time, Column 19 Evap, Column 20 Ann Evap)
31-12-13 23:58 Evap 3.68 Ann Evap 1537.9
31-12-13 23:59 Evap 3.68 Ann Evap 1537.9
01-01-14 00:00 Evap 3.68 Ann Evap 1537.9
01-01-14 00:00 Evap -1534.2 Ann Evap 0.0
01-01-14 00:02 Evap -1534.2 Ann Evap 0.0
and so on till...
At 8:59, just before roll over, the last entry in the file is
01-01-14 08:59 Evap -1534.2 Ann Evap 0.28

So now to the station roll over time at 9:00am Jan 2014 (in file Jan14log.txt)
The first line entries are.
01-01-14 09:00 Evap -1534.2 Ann Evap 0.28
01-01-14 09:00 Evap 0.28 Ann Evap 0.56
01-01-14 09:00 Evap 0.28 Ann Evap 0.56
and so on...

The reason why I picked this up was when looking at the day file. Noted a big negative value and thought that cant be right.

Files attached for perusal
Dec13log
Jan14log
and dayfile

I suspect that this may be a conundrum, as it appears tricky having a hole heap of things resetting to start the new day, month and year. And at different times.

Barry

[steve]

Yes, this comes up quite often, and you're correct that it's because of Cumulus having to work around the fact that the station has no 0900 rollover setting, so the annual figure resets at midnight. There is code in there to cater for this, but it relies on the annual ET figure doing certain things at certain times, and it clearly isn't so obliging. It's complicated by the fact the ET figure that Cumulus receives is buggy and unreliable anyway. I notice my reading doing odd things quite often.

I think that the real solution is for Cumulus to ignore the ET figure from the station and do the calculation itself. But the calculation appears to be very complicated.

[nossis]

Hi Steve,

Ah right O
Will do some fudging.. I mean adjustments manually
Thanks for your insights... Yep figured it would be "not worth getting excited about it" basket.

Barry