Incorrect Rain Today values

[mlj1]

Hi Steve,

For some days now, Cumulus sometimes reports rain today values ending with ,1 ,3, 5, 7 ,9 ! Until now, all rain values, according to Davis protocol, were allways multiples of 0,2.



So, I must edit rain values to have the same values as reported by weatherlink. Not sure if this started with the last Cumulus update (1089).

[steve]

Please zip up the diags folder and attach it.

[mlj1]

Here it is

http://www.meteo-saint-brevin.fr/Divers/diags.zip

Marcel

[mlj1]

Ooops ! it is better like this, sorry !

Marcel

[steve]

The communication protocol doesn't contain the actual amount of rain, it contains the number of tips. Cumulus doesn't actually get involved in converting tips to actual amount, the Davis DLL does it and just gives Cumulus the actual amount. So Cumulus is just using the amounts it has been given. The DLL converts the number of tips to an actual amount by reading the rain gauge type which is configured in the console. Cumulus reads the setting from the DLL when it starts up; it doesn't actually use the value, it just logs it to the diags file for diagnostic purposes.

In your diags files, I can see that the DLL is sometimes unable to read the rain gauge type from the console. I guess that what it does in that case is default to a 0.01" collector, with the result that you are seeing the odd values due to rounding of the slight difference between 0.01" and 0.2mm. Here's what happened the last nine times Cumulus started:

15/03/2014 19:09:18.105 : Rain collector type = METRIC 0.02 mm
15/03/2014 19:13:12.269 : Rain collector type = METRIC 0.02 mm
15/03/2014 20:43:17.341 : Rain collector type = METRIC 0.02 mm
15/03/2014 20:49:46.952 : Rain collector type = METRIC 0.02 mm
15/03/2014 23:34:10.306 : *** Unknown rain collector type
16/03/2014 00:08:15.623 : Rain collector type = METRIC 0.02 mm
16/03/2014 00:42:50.529 : Rain collector type = METRIC 0.02 mm
16/03/2014 10:58:43.119 : Rain collector type = METRIC 0.02 mm
16/03/2014 11:23:57.568 : *** Unknown rain collector type

That last one is from the current run, you haven't restarted Cumulus since then. So the DLL is presumably currently assuming a collector type of 0.01" and passing incorrect values to Cumulus.

If you restart Cumulus, the DLL may be able to read the collector type successfully, so will presumably then start passing correct rain totals to Cumulus. I'm not quite sure what will happen to today's count when it does that, I assume there will be a discrepancy. You can check near the beginning of the diags file (around line 75 in cumulus.009) whether it was able to read the rain collector type.

I have no idea why the DLL is sometimes unable to read the rain collector type from your station; that's between the DLL and the console. But I do note that you are using a very old version of the console firmware - 1.80.

[mlj1]

Thank you very much Steve for your investigations and time.

I will try to check everything. For the console firmware version, I was told this one (1.80) was quite stable and I was a little bit afraid to have everything down after updating to the last version, assuming I also use Virtual VP.

Best regards,

Marcel

[steve]

Ah, it would have been useful if you had said that you were using Virtual VP - the Davis DLL and Virtual VP don't play nicely together. That may be the cause of the DLL being unable to read the rain collector type sometimes.

[mlj1]

I understand that, Steve. But I was feeling happy because I have been using Virtual VP for 15 months now without any problem.

In fact, I first saw the problem after restarting Cumulus many time during the same evening. After a new restart, I was surprised Cumulus got a rain of 87,7 mm during the catchup and it was not raining. I gave a correction to 0,00 and I think the problem related here started at this moment.

I just restarted everything here now, made the correction for today's rain in order to have the same values than given by Weatherlink for my stats.

Thank you again for your help.

Marcel

[steve]

Does the current cumulus.009 file show that the rain collector type was read successfully this time?

[mlj1]

And also I omitted to say the problem related in this thread does not happen in weatherlink. All rain values are multiples of 0,2. I keep on checking everything.

[steve]

That's to be expected, because Weatherlink doesn't use the Davis DLL which Cumulus uses.

Does the current cumulus.009 file show that the rain collector type was read successfully this time?

[mlj1]

It seems to be ok for me.

[steve]

Yes, looks good. You'll just need to do that check each time you restart Cumulus.