Alternatives to resampling with land observational data

[ph-kev]

Resampling don't work as well as it should be for observational data. The first plot is of observational data for gpp. The second plot is after resampling onto simulation data. Because there are NaNs in the data, resampling don't work quite as well as the NaNs poison the values near the boarder. One potential solution is to resample on the union of the two grids so that no information is lost when going to the new grid.