Dewpoint when RH=0

[sgdecker]

Right now, dewpoint_from_relative_humidity returns nan when the relative humidity is 0 (as can happen with NWP output). Should it return 0 K instead?

[dopplershift]

Physically speaking, 0% RH implies that there isn't a single molecule of water around, so there is no temperature you can cool to that will result in the condensation of water in the air. In my mind, that makes nan a reasonable return for that case.

What problems is the current behavior causing you?

[sgdecker]

I agree this is a pathological case in the sense that the dewpoint isn't really defined when the RH is 0%, and even at absolute zero, the RH becomes 0/0. I thought this was triggering an error downstream, but it turns out it was something else. Thus, nan seems as good as anything.