1.8.3 beta hourly rainfall

[steve]

There seems to be a problem with my new hourly rainfall mechanism where occasionally the hourly rain (and hence, for Fine Offset users, the rate) will suddenly become the total.

Anyone else seeing this? I think I'll have to put some tracing in.

[dane]

Agree: rainfall rate looks a little strange.
Have been away for several hours and on return I find that I had 0.3 mm of rain, starting at about 20:45 (which I think is correct). Shows up correctly on total rain graph, but shows a rainfall rate of 0.3 mm lasting a full hour on the rainfall rate graph.

[steve]

Yes, that's how the highly sophisticated rainfall rate calculation works in Cumulus during normal running (for Fine Offset stations anyway). Rain rate = Rain in last hour.

I didn't have anything better to use, but now I've found a description of how the Davis stations work out rain rate, so at some point I'll probably implement that for those stations that don't supply rates.

[dane]

so what used to be a spike (like I had at 4 am) in 1.8.2 will now be a 1-hour block in 1.8.3?

[steve]

I haven't changed the rainfall rate calculation in 1.8.3. What you're seeing is the difference between the rate calculated when processing history data, and normal running. During normal running, it's as I described above. During history data processing, it uses the number of tips recorded between each entry in the history data.

So, say your station is logging at 5-minute intervals, and 0.3 mm of rain (1 tip) fell between two successive entries in the history data. This is a rate of 3.6 mm/hour for that 5-minute period. If no other rain fell, during normal running that would have been recorded as 0.3 mm/hour for a 1-hour period. Both are valid approximations, and they're just done differently because of the different data that is available during normal running and history processing.

I hope to remove or reduce this inconsistency between the two different calculations when I implement a more sophisticated algorithm.

[dane]

Steve, thanks for your explanation (and patience...
I think I understand what you're telling me

[steve]

No problem. At some point, I should write some documentation on how Cumulus does things, so people understand why they see what they see. It will help people point out potential problems with the way it works, also.