In NFC, you can add the bool flip_hemisphere which will invert the Y coordinate (thereby essentially converting it into a "poleward/equatorward" framework).
For rotating systems like TCs, this results in the U component of the wind looking like if it was in the NH, but the V component of the wind is not correctly flipped to preserve "cyclonic/anticyclonic" motions. When compositing an even number of NH and SH stors, this essentially "washes out" the V component average.
Not sure what the best option is here. Since the predominant flow in the mid-lats of both hemispheres is westerly, I don't think we want flipping both X/Y in SH to be default if user requests it (for example, it would make ARs look weird).
Maybe we want another bool like --flip_cyclonic that flips both the X and Y while --flip_hemisphere only flips Y (and if a user sets both, it just collapses to --flip_cyclonic or throws an error).