To set up different resolutions:

Copy the desired resolution over the top of 'periodic_hex.namelist.input.sqrt'.

Note that some resolutions get closest to the 20km width with an odd or even number of rows.
Thus, depending on resolution, you also need to modify:
  experiment_A/setup_sqrt_mesh_step.xml 
to either include or remove this line:
  <argument flag="--remove_extra_y"></argument>

2km: do NOT remove extra y; effective width=20207.2m
1km: remove extra y; effective width=19629.9m
500m: remove extra y ; effective width=20207.2 m
250m: remove extra y; effective width=20062.9 m

Note that the best way to measure model domain width is between the second to final yVertex coordinates 
on both north and south ends of the domain.  In other words, we want to exlude the hexagon's 'tents' on
the north and south from the effective width of the domain.  This is because there is no flux north or
south of those positions and those 'tents' are effectively dead space.  Note that the setup script
adjusts the externalWaterInput magnitude for cells in the north and south so that these 'tents' 
effectively receive no input.
