precipitation signals... actual rainfall vs delayed snowmelt

[jasonebox]

I am starting to contend with is separating '~accurately recorded rainfall' from 'delayed melting of snow that accumulated in the instrument at an earlier (ambiguous) period'.

The approach can include using backward time-lagged air temperature to know if abrupt warmups are part of a spurious 'rainfall' signal.

Ultimately, we can flag quite a lot of the cases as suspicious (in an accompanying quality identifier code column) and flag others as cases passing a test of extended period of above freezing temperatures followed by a legitimate rainfall signal. :-)