Cumulus Support

Noticed evapotranspiration started for the new day on the first day of 2015 at -122.81mm!

Last entry for 31/12/14 was 3.73 recorded in the first entry 00.00 on 01/01/15

Subsequent records at each logger period were -122.81. Console showing 0.0
Diags showed *** ET Reset every ~ 2.5 secs and continuing until restart suggesting ET was not resetting.
Restarted Cumulus and ET reverted to 0.0
After Cumulus restart, new diag file showed a single *** ET Reset in the start up phase.

Been noted before but a bit hidden away ....

Davis EVT ... There is a known problem for Jan 1st data.
See https://cumulus.hosiene.co.uk/viewtopic.php?f=4&t=11390 and https://cumulus.hosiene.co.uk/viewtopic.php?f=4&t=11158
I would suggest copying the day before or after value if you have these negative figures.

I'm pretty sure this is documented somewhere, but as I can't find it myself I guess it doesn't count. I've edited the FAQ which talks about rainfall on January 1st for Davis owners so that it also mentions ET.

Did a search before posting and did not find anything relevant - obviously my search terms were not good!

A negative value can be a valid value - just not -122.81!
I have found +122.81 was the annual value as at 30/12/14 12:55 and 31/12/14 00:00
Annual value at 31/12/14 23:55 was 126.54.

Have seen values as low as -0.08 just after midnight some days this month. Research on this indicates these are are valid values.

Reading between (various) lines suggests that a negative hourly value may be valid in some circumstances, but the daily summation must be > or = zero.
Generally the very small percentage of negative values do not significantly affect this and thus are 'left in' rather than ignored.

Does that make any sense ?

Well.. yes - does not say anything I was not already reporting as far as I can see?

Research indicates negative values are valid.
We have been getting some (small) negative values early in the morning which are soon overtaken by later ongoing positive values.

OK, plot thickens.

I have been looking at the Monthly files.
Looking at the first record for the year, 01/01/15 00:00,
I now find all the other daily total numbers were not zeroed until 00:05, the second record for the day.
At daily rollover, all but ET are zeroed at 00:00.

This did not seem to be the case for any of the previous month or year rollovers, all being zeroed as expected in the first record for the new month/year.

Perhaps your console clock is/was not quite in sync with your PC. What happens at the start of the year depends on when the console resets its annual counters relative to midnight on the PC.

How far out would the clocks have to be? Right now the clocks are within a second of each other.
I rely on the windows time sync to maintain the PC time (re-syncs every 3 hrs). I rely on Cumulus to keep synching the clock on the VP2 console to the PC.

There seemed to be a number of interrelated aspects to this issue.
1. The day before's total ET for the year turned up as a - val for the new first current ET value for the first day in the new year.
2. The '*** ET Reset' was not working as it repeated every ~2.5 secs until Cumulus was restarted.
3. Evapotranspiration does not reset in sync with the other daily total data items at midnight but after midnight.

I don't know how any of this works in detail so can only report what I observe in the hope that it may help to improve the future experience.

2. The '*** ET Reset' was not working as it repeated every ~2.5 secs until Cumulus was restarted.

It does look like that reset code isn't working as intended, it does seem to handle the reset eventually without restarting. When I get chance I'll look at it to see if I can see what the problem is.

3. Evapotranspiration does not reset in sync with the other daily total data items at midnight but after midnight.

Yes, this was mentioned in another thread. I've since changed the code (as yet unreleased) so that they all reset at midnight (before the midnight log entry).

Great - I'm sure the reset at the same time as the others will help.

I restarted Cumulus so often on the 1st sorting out this and other issues (of my own making) that I no longer have the diag file which showed the number of ' *** ET Reset ' events but I think it went on for at least 20 mins so I did not expect it to correct itself though it may well have done.