Add script to mask off frozen regions of hydro domain

[alexolinhager]

Creates a new script, floodFillThawedIce.py, that deactivates frozen regions of the hydro domain by using a flood fill of thawed basal ice originating at the grounding line. Thickness of frozen ice is set to zero and a no-flow waterFluxMask is defined surrounding the frozen area.

[xylar]

@alexolinhager, if you want this to be available in the mpas_tools conda package, it will also need to be added here:
https://github.com/MPAS-Dev/MPAS-Tools/blob/master/conda_package/setup.py#L70
and ideally you would add some sort of test to make sure it exists somewhere around here:
https://github.com/MPAS-Dev/MPAS-Tools/blob/master/conda_package/recipe/meta.yaml#L121

A final note that we typically use script names that are lowercase with underscores but I'll leave it up to @matthewhoffman how strict things are with the land ice scripts.

[alexolinhager]

Thanks @xylar, right now we just want this to be a script we can refer to and will use occasionally, but that doesn't need to be part of the conda package. I'll change the name though to make sure it's consistent with the rest of MPAS-Tools

[xylar]

Okay, sounds good!

[matthewhoffman]

@alexolinhager , thanks for opening a PR for this script. It will be good to have this saved for future use. I have requested some changes. A few will change the result, so take a careful look at things after making the adjustments and make sure those suggestions make sense.

[matthewhoffman]

@alexolinhager , thanks for the changes. I re-reviewed and just noticed one minor typo I made a suggestion for. Once you fix that and rename the script I can go ahead and merge this.

One other thing - can you comment on what impact the changes to the logic had? (changing growMask to use an or of the two conditions; considering thawed regions only where BMB is negative rather than nonzero)

[alexolinhager]

@matthewhoffman I made a few more changes to the script, mostly just minor debugging to allow the script to work with different input files that may have various variables in it that we want to get rid of. I did change back the calculation of basalMeltInput to abs(groundedBasalMassBal) though. For some reason I don't understand the previous implementation was creating all zeros throughout the domain. It seems like groundedBasalMassBal should almost always be negative, so it doesn't seem like this would have too much of an detrimental effect.

The change from and to or in the growMask definition was certainly necessary. Judging by the figure below (thawed areas are outlined in black), using an and requirement would have limited the active hydro domain to the small regions where both thawed ice and basalSlidingSpeed > 25 m/yr overlap, which would drastically cut down on the active hydro domain.

Based on the plots below though, I am confident the current script is doing what we want it to do. Non-zero thickness only exists in areas where basalMeltInput > 0, basalSlidingSpeed > 25 m/yr, and that are connected to the grounding line. Using the additional basalSlidingSpeed criteria allows us to capture the Whillans Ice Stream while having a minimal effect on the rest of the domain

[matthewhoffman]

@alexolinhager, looks good, thanks for making the changes