Custom Query (116 matches)

Create a new optional "area" attribute for domain. It must be a double array with the dimension of the numbers of cell of the domain (ie same dimensionnality of the domain mask attribute).

If present, it must be writen in the netcdf file according to the CF convention : ​http://cfconventions.org/Data/cf-conventions/cf-conventions-1.5/build/cf-conventions.html#cell-measures.

A new variable must be created related to the name of the domain similarely to the lon or lat coordinate (ie a simplified name when a single horizontal domain is included) array that contains the area of the cell :

[ float areas(cell) or float areas(x,y) or float areas_grid_T(x,y) ] cell_area:long_name = "area of grid cell" ; cell_area:standard_name="area"; cell_area:units = "m2"

For all field living on a grid composed of the domain the metadata attribute "cell_measures" must be added in the netcdf file with the area specification : ex : float sosstsst(time_counter, y, x) ;

sosstsst:long_name = "sea surface temperature" ; sosstsst:units = "degC" ; sosstsst:online_operation = "average" ; sosstsst:interval_operation = "5760s" ; sosstsst:interval_write = "1mo" ; sosstsst:_FillValue = 1.e+20f ; sosstsst:missing_value = 1.e+20f ; sosstsst:cell_measures = "area: areas_grid_T"

1) time units must be conform to UDUnits specification ( ​http://www.unidata.ucar.edu/software/udunits/udunits-2.1.24/udunits2.html#Database )

2) Add cell methods for time operation in field metadata cell_methods = “time_variable_name: online_operation (interval: interval_operation) ex :

float sosstsst(time_counter, y, x) ;

sosstsst:long_name = "sea surface temperature" ; sosstsst:units = "degC" ; sosstsst:online_operation = "average" ; sosstsst:interval_operation = "5760s" ; sosstsst:interval_write = "1mo" ; sosstsst:_FillValue = 1.e+20f ; sosstsst:missing_value = 1.e+20f ; sosstsst:coordinates = "time_centered nav_lon nav_lat" ; sosstsst:cell_methods = "time_centered: mean (interval: 5760s)"

We have a name mapping between xios and Cf compliance : average -> mean instant -> point accumulate -> sum

For better symetry with domains (i_index and j_index) and in order to be able to describe sparse distribution on axis.