|
Hi Ryo, EDGES was implemented long ago in Ferret, and is a 1-D version of BOUNDS, listing the low and high edges of the grid cells of a coordinate axis in one list. So if there are N coordinates, the list of edges is of length N+1. BOUNDS was then chosen for the CF convention and we implemented them in Ferret as well in order to be compliant with CF. Bounds list the low and high coordinate of each coordinate location in an Nx2 array. They both describe the boundaries between grid cells, and so they allow for the coordinate values to be anywhere inside the grid cell. If cell edges or bounds are not defined, then the bounds are taken by Ferret to be midway between coordinate locations. Ferret handles grids only where adjacent grid cells share boundaries, so that grid cells are contiguous, so as far as Ferret is concerned they always describe the same set of numbers. Take a look at the entries under "BOUNDS" in the Users Guide index, and also "DEFINE AXIS, BOUNDS" and "DEFINE AXIS, EDGES" in the Ferret Users Guide for more description. Essentially all the capabilities needed are now included in the BOUNDS syntax, but EDGES are in Ferret for backwards compatibility. I don't know myself what GrADS or Matlab does -- maybe other users can address that. Because BOUNDS are in the CF convention, they are probably more likely to be recognized by other software than EDGES would be. Ansley Ryo Furue wrote: Hi Ansley and David, | Z:units = "meters" ; | Z:positive = "up" ; | Z:edges = "Zp1" ; I'm wondering how the "edges" attribute is related to the "bounds" attribute. (The latter is defined in the CF convention. See, for example, http://cf-pcmdi.llnl.gov/documents/cf-conventions/1.4/ch07.html#cell-boundaries ) Is "edges" Ferret-specific? Do you recommend switching to "bounds"? Do other packages (like GrADS and Matlab) support either of them? Regards, Ryo |