Reading in NEMO data

[mollyjames2]

i just noticed something funky going on with the shape when trying to read in NEMO data. my NEMO file dimensions are

        axis_nbounds = 2 ;
        x = 324 ; 
        y = 424 ; 
        depthu = 31 ; 
        time_counter = UNLIMITED ; // (744 currently)

velcoty fields in these files have the dimensions (U data as an example):
uoce(time_counter, depthu, y, x)

and my coordinates file looks like

dimensions:
        x = 324 ;
        y = 424 ;
        z = 1 ;
        time = 1 ;
variables:
        float nav_lon(y, x) ;
                nav_lon:axis = "Longitude" ;
                nav_lon:short_name = "nav_lon" ;
                nav_lon:units = "degrees_east" ;
                nav_lon:long_name = "Longitude" ;
        float nav_lat(y, x) ;
                nav_lat:axis = "Latitude" ;
                nav_lat:short_name = "nav_lat" ;
                nav_lat:units = "degrees_east" ;
                nav_lat:long_name = "Latitude" ;
        float nav_lev(z) ;
                nav_lev:axis = "Model Level" ;
                nav_lev:short_name = "nav_lev" ;
                nav_lev:units = "unitless" ;
                nav_lev:long_name = "Model Level" ;
        float e1t(y, x) ;
        float e2t(y, x) ;
        float glamt(y, x) ;
        float gphit(y, x) ;
        float e1u(y, x) ;
        float e2u(y, x) ;
        float glamu(y, x) ;
        float gphiu(y, x) ;
        float e1v(y, x) ;
        float e2v(y, x) ;
        float glamv(y, x) ;
        float gphiv(y, x) ;
        float e1f(y, x) ;
        float e2f(y, x) ;
        float glamf(y, x) ;
        float gphif(y, x) ;
}

I am reading in the data using the following code:

ufiles = sorted(glob(f"{dataset_folder}/SRIL34_1HL_*1h_grid_U.nc"))  # the name of the file containing the zonal velocity
vfiles = sorted(glob(f"{dataset_folder}/SRIL34_1HL_*1h_grid_V.nc"))  # the name of the file containing the meridional velocity
wfiles = sorted(glob(f"{dataset_folder}/SRIL34_1HL_*1h_grid_W.nc"))  # the name of the file containing the vertical velocity
mesh_mask = f"{dataset_folder}/coordinates.nc" # the name of the file containing the grid specifications for the hydrodynamic data      
 
# 2.
filenames = {
    "U": {"lon": mesh_mask, "lat": mesh_mask, "depth": wfiles[0], "data": ufiles}, # get coordinate data from the grid specification file, depth from the vertical velocity file, and everything else from the zonal velocity file 
    "V": {"lon": mesh_mask, "lat": mesh_mask, "depth": wfiles[0], "data": vfiles}, # get coordinate data from the grid specification file, depth from the vertical velocity file, and everything else from the meridonal velocity file 
    "W": {"lon": mesh_mask, "lat": mesh_mask, "depth": wfiles[0], "data": wfiles}} # get coordinate data from the grid specification file, and everything else from the zonal vertical velocity file - this ensures that the depth is consistent across the variables 

# 3.
variables = {"U": "uoce", # the name of the zonal velocity within the files SRIL34_1HL_*1h_grid_U.nc
             "V": "voce", # the name of the meridional velocity within the files SRIL34_1HL_*1h_grid_V.nc
             "W": "woce"} # the name of the vertical velocity within the files SRIL34_1HL_*1h_grid_W.nc


# 4.
c_grid_dimensions = {"lon": "glamf", #define the names of lat and lon from the coordinate file (C-grid)
                     "lat": "gphif",
                     "depth": "depthw",
                     "time": "time_counter"}




dimensions = {"U": c_grid_dimensions,
              "V": c_grid_dimensions,
              "W": c_grid_dimensions}

however when looking at my data i can see that I am missing the time dimension and instead the shape suggests i have the axis_nbounds dimension instead.

fieldset.U.data.shape
(2, 31, 424, 324)

any advice on how to rectify this would be greatly apprectiated. Thank you in advance

[erikvansebille]

Hi @mollyjames2, thanks for asking. I suspect that the 2 in the fieldset.U.data.shape[0] is not the axis_nbounds, but is the fact that Parcels always holds two time slices in memory, in order to interpolate between them. That is also the difference between fieldset.U.grid.time (which always has length 2) and fieldset.U.grid.time_full (which in your case should have length 744).

This is the deferred_load=True option in feildset creation (the default), and is there to minimise the memory footprint of Parcels. On most computer systems, there's not sufficient memory to load in entire hydrodynamic fieldsets.