Cumulus Select-a-graph not workng

[aaardvaark]

I had a phantom rain event last week which I thought I solved by suitably modifying various files following instructions on this site.

I don't know if it's related, but since then my Select-a-graph doesn't work. I just get a blank graph no matter what I select - see attached screen capture.

As far as I know everything else is working ok - does anyone have any thoughts?

Also - is there any other way (other than select-a-graph) of seeing what happened a few days ago, say rainfall or temperatures?

Thanks
Julian

[Gordon-Loomberah]

I've seen this problem before, I believe it was caused by a corrupt line in the monthly log file. Look back to when you did your edititing and find and remove the offending line(s) or edit them to the correct format and all should be good again.

[steve]

Also - is there any other way (other than select-a-graph) of seeing what happened a few days ago, say rainfall or temperatures?

Look at the data files directly, either in a text viewer/editor (but this can be tricky, depending on what you want to look at) or in something like Excel (make sure you don't save any changes done in Excel), or - best of all - in the toolbox

[aaardvaark]

something like Excel (make sure you don't save any changes done in Excel), or - best of all - in the toolbox

This might be my problem, I thought that editing in excel by exporting result as csv would be ok. Is this satisfactory or is csv also a problem?

Thanks Gordon I'll look more carefully at the monthly file, maybe something doesn't make logical sense, or maybe it's my csv method as above.

[aaardvaark]

Thanks for help I've fixed the problem by some more careful and mildly tricky editing of the monthly file.

I now guess that excel can't be used even if you open as text and save as csv because it might save incorrect numbers of digits after decimal points, or mess up date formats etc.

For the benefit of others:

THe toolbox edit is great if you haven't messed up your file in any way other than some wrong values. It would have fixed one of my two initial problems but could not fix another of the initial problems or one of the problems I introduced:

- some extra lines with odd times were inserted when cumulus restarted after my changing some values. Toolbox won't remove a whole line, and it won't help you to identify this problem. For this I used excel to identify the extra lines by looking for those that weren't 5 minutes apart (but not saving from excel, just analysing) and then a text editor to fix.

- possibly due to using excel all of my dates prior to the problem area were in /yyyy format instead of /yy. For this I had to use a text editor to fix (other than manually changing 1000 entries in toolbox editor).

So if I'd used toolbox in the first place, I would have had a hard time identifying the extra lines, but for replacing the corrupt values it would have been very handy.

[steve]

- some extra lines with odd times were inserted when cumulus restarted after my changing some values. Toolbox won't remove a whole line, and it won't help you to identify this problem. For this I used excel to identify the extra lines by looking for those that weren't 5 minutes apart (but not saving from excel, just analysing) and then a text editor to fix.

This is the expected behaviour. It is valid to have lines that aren't all the same number of minutes apart, there's no easy way around it. Data logged by Cumulus which was downloaded from the logger will be timestamped differently to that logged while Cumulus is running; there's no synchronisation between the station logger and Cumulus in this respect. As long as the timestamps and data appear valid, it's not a problem.

possibly due to using excel all of my dates prior to the problem area were in /yyyy format instead of /yy.

Yes, the date format is the main reason for not using Excel to edit the log files.

[aaardvaark]

- some extra lines with odd times were inserted when cumulus restarted after my changing some values.

This is the expected behaviour. It is valid to have lines that aren't all the same number of minutes apart, there's no easy way around it. Data logged by Cumulus which was downloaded from the logger will be timestamped differently to that logged while Cumulus is running; there's no synchronisation between the station logger and Cumulus in this respect. As long as the timestamps and data appear valid, it's not a problem.

OK thanks - I thought the select-a-graph may have been expecting a set number of regular spaced entries, I wasn't otherwise concerned about the extra entries so now I know I can leave them next time.

Cheers
Julian