Possible Wind Run Bug?

[colinpb]

Either I'm misinterpreting, I have a problem with my records or I may have found a bug. In Highs and Lows the reports “all time”, “monthly records” “this month highs and Lows “and “this year’s highs and Lows” the highest wind run value shown is not for the date recorded but for the day before. These shows up now we have moved into September, highest wind run has todays date not yesterday as it should do. I checked dates with "View - This Period". I'm running 1.9.3 1043

Colin

[steve]

Can you show me the data and explain what you mean, please? You seem to first be saying that the value is from the wrong day, and then saying that the date is wrong, so I must be misunderstanding something. Do you run Cumulus 24/7, or do you use the station's logger data? Do you use a midnight rollover, or 0900/1000? Is this with your Davis station, or your Fine Offset, or both?

If you zip up the diags folder and attach it, and the same with your data folder, I'll see if I can see where the problem is.

[colinpb]

I use midnight roll over and I use the stations logger. I have now checked both my WH3080 and VP2 versions of cumulus. The 3080 version is working correctly, the VP2 is incorrect. See screen shot of VP2 records showing two sample dates, the “view this period” dates of 11/08/2012 and 01/09/2012 are the correct dates not the Monthly Records Highs of 12 August 2012 and 02 September 2012.

The incorrect dates appear in the Highs and Lows reports “all time”, “monthly records” “this month highs and Lows “and “this year’s highs and Lows”

The only difference is that the 3080 started with a 1.9.2 version of cumulus in November 2011 before moving to 1.9.3 beta and my VP2 started with a 1.9.3 beta in June 2012.

VP2 Data and Diags folders attached

Colin

[steve]

Here's the problem:

02/09/2012 12:17:20.271 : VP2: processing logger data entry 117 for 02/09/2012 [ Note: there's an implied time of 00:00 here which doesn't get printed]
02/09/2012 12:17:20.271 : ih=61.0 oh=87.0 it=21.8 ot=13.6 wc=13.6 dp=10.4 pr=1024.589 wg=5.0 ws=1.0 wd=9 rt=0.00 rr=0.00
02/09/2012 12:17:20.271 : hi=13.5 thwi=13.5 thswi=-3.28E4 idp=10.4 ihi=21.3 uv=-3.28E4 sol=-32768 et=0.0
02/09/2012 12:17:20.359 : New monthly record, month = 9: 2012-09-02 00:00 21.967 "High daily windrun" 2012-09-01 23:55 21.883
02/09/2012 12:17:20.373 : Rain counter = 224.99885559082
02/09/2012 12:17:20.488 : Writing today.ini, LastUpdateTime = 02/09/2012 raindaystart = 224.99885559082 rain counter = 224.99885559082
02/09/2012 12:17:20.489 : VP2: Day rollover 02/09/2012
02/09/2012 12:17:20.489 : === Day reset, today = 2
02/09/2012 12:17:20.489 : === Day reset for 01/09/2012

It processes the midnight entry from the logger before doing the rollover (since the data in that entry belongs to the current day) but as it's midnight, the date is (just) the following day. I'll have to think about how to fix that without breaking something else...

[colinpb]

Curiosity begs me to ask, is the reason the WH3080 produces the correct date down to the fact it doesnt log 00:00, the closest it gets is 23.59?

Colin

[steve]

With a Davis, there will always be a midnight logger entry, and it is clear that the midnight entry contains data for the previous day, so the entry is processed first before the day reset is done. Doing it the other way around would make this particular problem go away but then the logged data would be slightly less accurate. I think originally it did do it the other way around, before this discrepancy was pointed out to me.

With the Fine Offset, there will occasionally be a midnight entry, but more often there won't be - unless a 1-minute interval is in use, which is something I didn't realise you could do until recently. So it does the day reset first, and then processes the data, on the assumption that most of the time at least part of the data will be for the next day, and it's easier and safer to do the reset first.