Weird - Time seems to alter data loaded after a flat batter?

[Keyz]

Hi there,
I left this mrning for work after checking the weather on my station, so I know it was running about 8am.

I got home and the batteries were flat on the receiver/screen, so I quickly plugged it into the pc, and it loaded the data. I had to reset the clock on the screen. It was on 12 hours, and I left it on that and reset the date.

Anyway, when I opened cumulus, I ended up with some extreme rain rates, so I checked the data.

As you can see from the below, the time steps in the 7 minutes I desired, but the earlier lines (where the temp is 10deg C) was from 3-4am this morning, even though time on there is saying 3-4pm.

Then, it steps up to just after the actual 5pm when I plugged the unit back in, and reads correctly from there on out. It appears my upload has written over the data from 5am in the morning till after 8am when the batteries went flat? Could this be so, and due to the time?

Because it appears that previously I had the incorrect time set somewhere (Dunno how!), so it was recording in Cumulus in reverse AM/PM to what the time actually is?????

I've attached my Sept log, and you can see that according to the prior data, since 14/09/10 it is getting colder during the day (12:00) and warmer at night (24:00)!! And then resetting it has put it back to normal?

Please help, been searching and can't find an answer to what I might have done and how I might get my missing data back from this morning

Check out the weird cumulative rain figures too

14/09/10,17:18,15.5,80,12.0,7.2,14.8,304,0.0,1.2,1010.0,686.7,21.0,62,14.8,15.5,15.5
14/09/10,17:37,15.5,80,12.0,7.2,14.8,304,0.0,1.2,1010.0,686.7,21.0,62,14.8,15.5,15.5
14/09/10,17:38,15.2,82,12.1,7.2,23.4,309,0.0,1.2,1010.0,686.7,21.3,63,15.8,15.2,15.2
14/09/10,17:39,15.2,82,12.1,11.2,23.4,310,0.0,1.2,1009.9,686.7,21.4,63,20.9,14.7,15.2
14/09/10,17:40,15.2,82,12.1,14.8,23.4,301,0.0,1.2,1010.0,686.7,21.4,63,22.0,14.3,15.2
14/09/10,17:43,12.7,96,12.1,0.0,1.1,181,0.0,1.2,1017.5,664.8,20.1,63,1.1,12.7,12.7
14/09/10,17:50,12.7,96,12.1,0.0,1.1,203,0.0,1.2,1017.2,664.8,20.0,63,1.1,12.7,12.7
14/09/10,17:57,12.7,96,12.1,0.0,3.6,188,0.0,1.2,1017.2,664.8,20.0,63,3.6,12.7,12.7


16/09/10,17:00,9.9,99,9.8,2.5,5.0,200,0.0,0.0,1018.9,669.9,22.4,60,5.0,9.9,9.9
16/09/10,17:07,10.0,99,9.9,0.0,5.0,200,0.0,0.0,1019.0,669.9,22.4,60,0.0,10.0,10.0
16/09/10,17:14,10.1,99,10.0,0.0,1.1,200,0.0,0.0,1019.1,669.9,22.3,61,1.1,10.1,10.1
16/09/10,17:21,9.9,99,9.8,0.0,1.1,200,0.0,0.0,1019.1,669.9,22.2,61,0.0,9.9,9.9
16/09/10,17:28,10.0,99,9.9,0.0,0.0,200,0.0,0.0,1018.9,669.9,22.2,61,0.0,10.0,10.0
16/09/10,17:35,15.7,85,13.2,3.6,8.6,217,262.3,30.6,998.3,700.5,21.9,60,8.6,15.7,15.7
16/09/10,17:42,15.5,84,12.8,8.6,18.4,249,0.0,30.6,998.5,700.5,21.5,60,18.4,15.4,15.5
16/09/10,17:49,15.3,85,12.8,6.1,18.4,257,0.0,30.6,998.7,700.5,21.4,61,17.3,15.3,15.3
16/09/10,17:56,15.1,85,12.6,6.1,17.3,251,0.0,30.6,998.8,700.5,21.2,61,12.2,15.1,15.1

[steve]

With Fine Offset stations, Cumulus always uses the PC clock. The entries in the Fine Offset logger aren't timestamped, so Cumulus assumes when it starts up that the latest entries in the logger are those that it needs to fill in the missing data from the time when it wasn't running. It has no way of confirming that the entries do in fact correspond to the timestamps that it generates - it starts with the current time and works backwards through the entries; each entry has a field which gives the interval for that entry.

If your data has got out of sync, then I suspect that your logger stopped logging, so when Cumulus started, the latest data wasn't actually for that moment in time.

If you zip up and attach the diags folder, I might be able to tell you a bit more about what happened.

[Gina]

If the batteries go flat in the console and there is no USB power from the computer then the logger memory will be cleared as I recall (not sure). But even if the memory is held, the console will no longer be logging data and there will be a gap and (as Steve says) the timings will be wrong.

[Keyz]

Oh, I see, so it just back filled the data with that from when it wasn't running, up to the poinjt that it had already loaded, which I assume was the 14th?

Weird, as the screen definately was still running over the last few days, I saw the low of 6 odd degrees on the night of the 15th myself, hence why I am so confused.

In the time the logger was down, then, we happened to have have a lot of rain, whcih I'm not too concerned about recording right.

But what I do want to get right is that high rain rate showing on the 16th (260 odd mm) and to correct the cumulative rain amount. Can you help with what I should do with that?

Thanks so much again, glad I have donated now

[Keyz]

I tried to look for the lines of data in the start and end of the inserted period (assume 14/09 to 16/10) but I can't find the same temp,humidity,pressure,wind direction all in the same line with the same figures for the log files for July, August or September. The data is set to 7 minutes, so would be written over in 18 days or so, hence not needing to go further back than August really.

It also seems strange that the temp recorded is coincidentally the same at 11am on 15th data as was at 11pm 'in real life'.

If you can't find an easy solution Steve, should I just delete the misnomer data?
How would I correct the cumulative rain etc. A bugger, as I like to know how much rain we've had! My fault for not charging the batteries, but I never saw a battery warning though

Would the 30.2mm of rain recorded in the first correct line on the 16th be the correct figure for that day? Given what we had it seems plausible there was that much rain, but not sure if it's plausible that Cumulus has pulled the cumulative figure for the day from nowhere?

And where would the rain rate (263mm) come from? A cumulus calculation?

Thanks again for your help

[steve]

The station's 'current location' pointer got reset to zero (i.e. the start of the logger). So Cumulus started from that point, assuming that the previous data (wrapping around to the end of the logger) was what it needed to backfill from 1807 on 16th Sept to when you last closed it down at 1743 on 14th Sept. So the data it used was actually data from some time ago; from the values in the entries it looks like the first entry was actually from the morning of the 9th Sept:

This the first entry from the logger after start up
14/09/10,17:43,12.7,96,12.1,0.0,1.1,181,0.0,1.2,1017.5,664.8,20.1,63,1.1,12.7,12.7
This is where it came from originally
09/09/10,06:22,12.7,96,12.1,1.1,5.0,157,0.0,0.0,1017.5,664.8,20.1,63,5.0,12.7,12.7

The entries match apart from those values which Cumulus works out based on the preceding data (averages, totals etc). So this explains why the temperatures appear 12 hours out of phase; they're actually several days plus 12 hours out.

All I can tell you about the rain totals is that when you closed Cumulus on 14th Sept the counter was 686.7, and when Cumulus started reading the current data on 16th Sept at 1810 after downloading all of that old data, it was 700.5:

14/09/10,17:40,15.2,82,12.1,14.8,23.4,301,0.0,1.2,1010.0,686.7,21.4,63,22.0,14.3,15.2

16/09/10,18:10,14.9,85,12.4,6.1,12.2,254,0.0,30.6,999.1,700.5,21.1,61,11.2,14.9,14.9

So between those two points you had 13.8 mm of rain. It's not possible to say when that fell.

In summary, all of the data in your logs between 1743 on 14th Sept and about 1735 on 16th Sept are invalid. All I can suggest is that you delete that data, and then adjust the entries in dayfile.txt correspondingly. You'll need to decide how to apportion the rainfall; you had already had 1.2 mm of rain when you closed down on 14th Sept, so if you make sure that the totals for 14th/15th/16th Sept add up to that 1.2 plus the 13.8 plus any rain you had on the 16th after the end of the log at 1904, then your cumulative totals should be correct.

You can correct your all-time high rain rate using the editor on the edit menu; the previous record will have been logged in alltimelog.txt. The daily highs are in dayfile.txt.

[Ned]

Your'e not alone! There's at least 2 other Aucklanders here (I'm one) who could supply missing data from their records - won't be perfect, but maybe better than guessing. Just say....

[Keyz]

Thanks so much Steve! That is awesome that you helped me with that, thank you so much!

Just want to reiterate I love using this software, and your assistance is top notch!

Ned, I PM'd you for the data, if you could copy and paste it in a reply that would be awesome!

Sure has been a crazy few days of weather here over the weekend.

[Ned]

Keyz - PM sent