How to interpolate windspeed timeseries for power: nearest neighbor ?

[denis-bz]

Hello MetPy people,
can you help me with a question on interpolating windspeed timeseries from a scattered grid to wgs84 ?

Power from windmills is highly sensitive to peak windspeeds.
Interpolating timeseries smooths, lowers peaks, which would bias windpower estimates:
wind power is 0 below ~ 3 m/s, so interpolating 0 6 0 6 ... to 3 3 3 ... would give 0 power, hmm.
Don't want that, so I just do nearest-neighbor --

wgs84 gridpoint ⟶   the nearest staggered gridpoint ⟶   the windspeed timeseries there, unchanged.

This has pros and cons.
Pros: we see exact power at the known, nearby staggered gridpoints
Cons: nearest-neighbor is blocky
Do you have any experience with nearest-neighbor vs. say IDW ?

Thanks -- appreciate your comments
cheers
-- denis

(Some notes and plots on what I'm doing are under https://gist.github.com/denis-bz .)

[dopplershift]

If peaks are of chief concern, then nearest is probably your best bet. ESMF may have some other options in their regridding tools for this; if that's the case I think xESMF might be helpful as an xarray-based Python library using htis.