The tools provided by NCL (NCAR’s Command Language) are indispensable when it comes to analyzing output from CAM. Take, for instance, functions that convert vertical profiles of wind (U and V) to a vertical profile of vorticity. The reliability of the output depends on which function you use though, and we’ve found that uv2vrF results in strange banded patterns that uv2vr_cfd does not.
Below is a plot of relative vorticity for a single day simulated by CAM5 with pre-industrial forcing, which has been computed by uv2vrF. Note that there are multiple zones of seemingly banded structures, particularly prevalent at mid-to-high latitudes (focusing in the southern hemisphere, they seem particularly noticeable).
If we move to using uv2vr_cfd, however, that banded structure has been smoothed, giving a more cohesive vorticity field.
Banded patterns in the relative vorticity field continued when assessing absolute vorticity, and even carried through to calculations of genesis potential index (which relies on not only absolute vorticity, but also RH, mixing ratio, and potential intensity in m/s). Achieving reasonable results for relative vorticity therefore became a first order problem in our results, where switching functions was necessary before we could see a smoothing of the vorticity field.