Hi Ansley,
thanks - including the station ID and 'cf_role' attribute did shift the variables that are a function of 'station' to the E axis; updated simple files in the attachment.
FYI: The other thing that had been unclear to me: On the ferret and cfconventions.org website mentioned below, 'temp' is only a function of 'obs':
yes? show grid temp
GRID GQA2
name axis # pts start end
OBS X 20787 r 1 20787
normal Y
normal Z
normal T
normal E
normal F
So I assume that all measurements from all stations are concatenated in temp (to save space and dimensions as opposed to having temp be a function of 'obs' AND 'station'), and can be assigned to the respective stations by means of the 'rowsize' variable.
The SEPARATE function works well as described on the ferret website:
yes? let/units="`temp,return=units`"/title="`temp,return=title`" separate_temp = SEPARATE(temp, rowsize, 0)
yes? list separate_temp
VARIABLE : temperature (Celsius)
FILENAME : ragged.nc
SUBSET : 13 points (X)
1 / 1: 1.00
2 / 2: 2.00
3 / 3: 3.00
4 / 4: 4.00
5 / 5: ....
6 / 6: 5.00
7 / 7: 6.00
8 / 8: 7.00
9 / 9: 8.00
10 / 10: 9.00
11 / 11: ....
12 / 12: 10.00
13 / 13: 11.00
... and expanding the variable into 2-D also works well as described by ferret, e.g.,
yes? let longest = `rowsize[m=@max]`
yes? let temp2d = EXPNDI_BY_M_COUNTS(temp, rowsize, longest)
yes? shade temp2d
Nice!
Best wishes,
Hella
On 08/03/17 21:13, Ansley C. Manke wrote:
Hi Hella, The other attribute that's needed is "cf_role", as an attribute on a variable which will identify the station (or trajectory, or whatever the feature type is, in general). Your file doesn't seem to have a variable that is a station id. This is described in the CF conventions but it looks like we need to add this information to the Ferret documentation about Discrete Sampling Geometries files. -Ansley On 3/8/2017 8:07 AM, Hella Riede wrote:Hello ferreters, I checked out the ferret documentation website about discrete geometries http://ferret.pmel.noaa.gov/Ferret/documentation/users-guide/data-set-basics/2-8-discrete-geometries-datasets-1, especially about the attribute 'sample_dimension'. Based on http://cfconventions.org/cf-conventions/v1.6.0/cf-conventions.html#_contiguous_ragged_array_representation_of_time_series, I made a simple dataset, see attachments, containing the 'sample_dimension' attribute. When I load this simple example into ferret, I don't get the observations on the X axis and the instance axis ('station') in E direction as described on the ferret website above, but rather two X grids instead, see output below. Any idea why? Is the 'dsg_file.nc' dataset used in the ferret example available somewhere to better understand the structure? I am not sure if 'obs' should be the sum of all observations of all stations which 'rowsize' will then divide up per station, or if 'obs' should be the maximum length per 'station'. Thanks in advance! Best wishes, Hella on FERRET v7.022 (beta/debug) yes? use ragged.nc yes? sh da currently SET data sets: 1> ./ragged.nc (default) name title I J K L LON station longitude 1:3 ... ... ... LAT station latitude 1:3 ... ... ... ROWSIZE number of observations for this 1:3 ... ... ... TEMP temperature 1:11 ... ... ... yes? sh grid temp GRID GNN2 name axis # pts start end subset OBS X 11 r 1 11 full normal Y normal Z normal T yes? sh grid rowsize GRID GNN1 name axis # pts start end subset STATION X 3 r 1 3 full normal Y normal Z normal T
Attachment:
ragged.nc
Description: Cdf file
netcdf ragged {
dimensions:
station = 3 ;
obs = 11 ;
name_strlen = 4 ;
variables:
float lon(station) ;
lon:long_name = "station longitude";
lon:units = "degrees_east";
float lat(station) ;
lat:long_name = "station latitude" ;
lat:units = "degrees_north" ;
int rowsize(station) ;
rowsize:long_name = "number of observations for this station" ;
rowsize:sample_dimension = "obs" ;
char station_name(station,name_strlen) ;
station_name:long_name = "station name" ;
station_name:cf_role = "timeseries_id";
float temp(obs) ;
temp:long_name = "temperature" ;
temp:units = "Celsius" ;
temp:coordinates = "lat lon" ;
temp:_FillValue = -999.9;
data:
lon = 10., 20., 30. ;
lat = 40., 50., 60. ;
rowsize = 4, 5, 2 ;
station_name = "A", "B", "C" ;
temp = 1., 2., 3., 4., 5., 6., 7., 8., 9., 10., 11. ;
}