Re-setting the rain sensor

[Blackmyre]

In dealing with frozen wind sensors I caused my station to report rain when there was none. I've edited today.ini and set its "start" value to the latest value in the monthly log file, and that seems to have resulted in correct (i.e. zero) values being reported in the Rainfall section of Cumulus. The Rain Rate is still reporting something silly though. Is there something else I need to edit to correct that?

I also tried changing the rain values in the monthly log file to the values prior to the disturbance, but I'm not at all sure if that's necessary or even sensible. I don't yet have a clear idea of what sort of values are reported by the hardware and what are calculated by Cumulus, which doesn't help. I get the impression that the rain figures are the most vulnerable to such disturbance. Is there (or should there be) a FAQ addressing the recommended way to recover from such situations?

[steve]

The data structure that is used to calculate the rain rate is set up from a combination of the 'total' values in the log file (i.e. Jan09log.txt) and the data downloaded from logger in the station (if any has been downloaded this time). So if those are showing the false rain, then the rain rate will be correspondingly wrong.

The rain figures are always the most problematic because the station only supplies a total, and everything is calculated from that, and once that goes wrong, everything subsequent to that is wrong, and you have to correct the bad data and get back into step with the new total. With transient things like temperature, it's just a case of removing any incorrect log entries.

Steve

[Blackmyre]

Thanks Steve. Does that mean that the "false rain" reporting will eventually correct itself as the data in the station is overwritten? Is it possible to re-set some/all figures in the station?

I noticed that the historical rainfall views (from 'select-a-graph') now seem to be showing zero, presumably something I've broken in my efforts to correct the log files. Where does this feature get its data from?

[steve]

Thanks Steve. Does that mean that the "false rain" reporting will eventually correct itself as the data in the station is overwritten?

Once Cumulus has downloaded the data and logged it, then it will be using the data from its own log files. Are you still seeing the effects somewhere?

Is it possible to re-set some/all figures in the station?

I would imagine it's possible to reset all the data, but maybe not individual ones - I don't know.

I noticed that the historical rainfall views (from 'select-a-graph') now seem to be showing zero, presumably something I've broken in my efforts to correct the log files. Where does this feature get its data from?

From the 'monthly' log files. The 'rainfall rate' graph plots the eighth item from the file(s) and the 'rain' graph plots the ninth.

Steve

[Blackmyre]

Once Cumulus has downloaded the data and logged it, then it will be using the data from its own log files. Are you still seeing the effects somewhere?

Sorry for the delay in replying. I had to put this down for a while for various reasons, and I don't recall exactly what I was seeing when I originally posted. In trying to retrace my steps, as it were, I think perhaps I was confused by some aspects of the graphing behaviour. It looks like several of the graphs show data from the time Cumulus was last started, whereas others go further back and represent data from the logs. That might have led me to believe I had messed something up in resetting the rain readings, because the rain graphs were showing blank on restarting.

Secondly, if I select rainfall (either total or rate) using the "select-a-graph" feature, I just seem to get a flat line of zero mm. Does that indicate a data error, or does "select-a-graph" not support rain graphing in the current version?


Steve.

[steve]

It looks like several of the graphs show data from the time Cumulus was last started, whereas others go further back and represent data from the logs

The graphs are all populated from the logs at startup.

Secondly, if I select rainfall (either total or rate) using the "select-a-graph" feature, I just seem to get a flat line of zero mm. Does that indicate a data error, or does "select-a-graph" not support rain graphing in the current version?

It should work. It reads the rainfall rate and total figures from the monthly log files.


Steve.

[Blackmyre]

The graphs are all populated from the logs at startup.

Strange - they don't seem to show that here. I restarted Cumulus to check just before posting, and the only ones that appeared to show historic data were the daily rainfall and the min/max/avg temperature. Having just re-checked, all of the others are only displaying from just before 14:35, whch would have been when I restarted it.

[Rainall in select-a-graph] should work. It reads the rainfall rate and total figures from the monthly log files.

I've tried it several times and it always gives a flatline. The logs seem to contain credible figures. I realise that it's always possible that I've introduced an error though, given that I had manually removed some spurious rain readings.

[steve]

It sounds as though you may have some blank lines or other errors in your log files, Cumulus gives up looking for data if anything unexpected happens.

Steve

[Blackmyre]

OK, I'll take a look - thanks for the pointer. Dunno why the min/max and daily rainfall graphs are unaffected in the main graphing option, presumably a different branch of the code (or perhaps their data comes from the daily log file). Similarly everything but the rainfall is OK with select-a-graph.

Is a terminal CR/LF at the end of a file required/allowed/forbidden in the monthly logfiles, by the way?

[steve]

or perhaps their data comes from the daily log file

Yes

Is a terminal CR/LF at the end of a file required/allowed/forbidden in the monthly logfiles, by the way?

There has to be exactly one CR/LF at the end of each line.

Steve

[Blackmyre]

OK, thanks.

[Blackmyre]

Well, I thought I had found it - when I removed the spurious rain details from the jan09 logfile I had managed to leave a block of original rows in there too. That means that the record times suddenly jumped back several minutes, so it's hardly surprising that Cumulus had trouble! I also noticed a step in the total rainfall counter figures. I had thought that these were calculated by Cumulus from other rain data reported by the station, but I now see that it's the other way round. I've now "back-filled" so that all previous totals are adjusted accordingly. I've also made sure that the "start" value in today.ini agrees with the latest total in jan09.

I thought that would fix it, but unfortunately not. Graphs (apart from min/max and daily rainfall) still only show from startup time (nothing from previously logged values). Select-a-graph show any values at all now (flat-line on everything, and just "00:00" in the centre of the horizontal axis), so I've actually made things worse.

Eyes are sore now, going to stop staring at datafiles for a bit!


Steve.

[steve]

If you want to send me your jan09log I can have a look at it for you?

Steve

[Blackmyre]

That's very kind of you, thanks - I'll try to attach it to a PM.

I'm sure you must have better things to do than pick over my errors though - please don't feel obliged to spend time time on it.

Steve.