Handling of NAs

[mpvginde]

Hello,

I would like to use this package to validate some precipitation forecasts against radar observations.
Can the code handle NAs in the radar observation fields?

Kind regards,
Michiel

I really liked reading the paper btw ;-).

[wolfgruber]

Hey,

the code does accept NANs an input and treats them like zeros. This is because the input fields are compared with a threshold like field >= threshold and NANs always give False in such comparisons.

I don't know if this is the desired behavior for your case, but otherwise we could try to work something out.

cheers, Ludwig

[mpvginde]

Thanks for your answer, that is indeed what I did for now (also applying the same NAN-mask to the forecasts).
But if I understand correct, this will probably inflate the scores a bit, as these NANs are now included as correct forecasts instead of just being ignored.
(And I guess deflated when you don't apply the NAN-mask to the forecast fields)

I'll have a deeper look into the code to verify if this indeed the effect of using NANs in the fields.