CAM-SE works with data stored at Gauss-Lobatto-Legendre (GLL) nodes.  These nodes can be interpreted in several ways shown in the three grids below.  In all three grids, the GLL nodes are the same (the green dots in the figure).   Internally, the dycore uses the CAM-SE native grid connectivity, but the rest of CAM and ACME has no knowledge of the grid connectivity and only sees data at GLL nodes and their associated area weight.  The other other types of connectivity are not used within the model, but are sometimes used by pre- and post-processing tools.

...

1. For quasi-uniform cubed-sphere grids, ACME currently uses the "pentagons" option to construct the dual grid for making ESMF mapping files.  To create these meta data grid files,  we use a utility distributed with the ACME code - see acme/components/homme/test/templates.
2. For RRM grids, ACME currently uses the "cheveron" option to construct the dual grid for making ESMF mapping files.  To create these meta data grid files we use a Matlab program that performs a Newton interation.  This matlab code is in the ACME PreAndPostProcessingScripts repo under regridding/spectral_elements_grid_utilities/

Future Plans

We note that the TempestRemap package uses an algorithm that understands both SE and FV grids (Ullrich & Taylor, MWR 2015) and can produce map files directly from the CAM-SE Native grid.  It only needs the SE Native grid (the first grid shown on this page) and does not need any type of dual grid.  These mapping files are more accurate since they take advantage of the SE discretization.  If ACME atmosphere adopted TempestRemap mapping files, we will not need to use the SE dual grid.    

Also, the "physics grid" work, when completed, will allow us to run CAM physics at cell centers of the subcell grid (instead of vertices).  It will also allow us to run the physics on a subcell grid with equally spaced subcell centers.  Either of these options will mean we would no longer need the SE dual grid.  

CAM-FV Grids

We also show some plots of grids used by the Lin-Rood CAM-FV model:  

The first plot below shows the grid used by CAM-FV for all scalar fields.  NCL refers to this as a "fixed" gridThe red dots show where the data are located, and the blue lines give the cells associated with each data point.  Data is interpreted as cell averages at the red dots. Because of the lat/lon representation, data at the poles are actually duplicated for every longitude.   A grid with equal spacing in both latitude and longitude will have dimensions (N+1) x 2N.   Data at 180E = 180W are not duplicated in the file, and to convey this, we only plot red dots at 180W and not 180E.

Image Removed

The next plot shows what NCL refers to as the fixed offset grid  (aka Uniform).  This is the internal CAM-FV velocity grid.  CAM-FV can output these velocities if you ask for 'US' or 'VS' in the history.  The 'U' and 'V' data in CAM-FV history has been interpolated to the fixed grid shown above.    Gaussian grids (used by CAM-EUL) look nearly identical to these offset grids, except Gaussian grids use a spacing in latitude which is not quite equi-angular.  A grid with the same equal spacing in both latitude and longitude will have dimensions N x 2N. 

Image RemovedSee also CAM-FV Grid Overview