[CF-metadata] Handling data from land surface models with sub grid scale tiling

From: Martin Dix <Martin.Dix>
Date: Thu, 13 Mar 2008 12:29:46 +1100

Many land surface models include sub-grid scale tiling to represent
surface heterogeneity. Each grid box is divided into a number of tiles
of different surface characteristics. The relative areas of each type
vary from point to point and they have no definite location within the
grid box.

CF doesn't seem to say anything about how best to specify such variables
and I couldn't find any other useful conventions either. We've been
using something like

dimensions:
  lat = 192 ;
  lon = 96 ;
  tile = 9
  time = UNLIMITED ;
variables:
  float time(time);
    time:standard_name = "time" ;
    time:units = "days since 2000-01-01 00:00" ;
  float tile(tile)
    tile:long_name = "tile index"
  float lon(lon) ;
    lon:standard_name = "longitude";
    lon:units = "degrees_east";
  float lat(lat) ;
    lat:standard_name = "latitude" ;
    lat:units = "degrees_north" ;
  float tile_frac(tile,lat,lon)
    tile_frac:long = "tile fractional area" ;
  float ts(time,tile,lat,lon)
    ts:standard_name = "surface_temperature" ;
    ts:units = "K" ;

though we've sometimes been inconsistent and used (lat,lon,tile) too. A
further complication is that one of my colleagues is contemplating using
tiling for soil temperature and moisture, so we'll have something like
 tsoil(time,depth,tile,lat,lon) (or some not quite random permutation).

Does anyone have any advice on this? Is it something that should be
included in CF?

Regards

Martin Dix
Received on Wed Mar 12 2008 - 19:29:46 GMT