Error reading FVCOM file

[AndresSepulveda]

Hi,

I want to use an FVCOM file do lagrangian analysis. My .py reads it with the following commands:

from opendrift .readers import reader_basemap_landmask from opendrift .readers import reader_netCDF_CF_generic from opendrift .models .oceandrift import OceanDrift

o =OceanDrift (loglevel =0 )# Set loglevel to 0 for debug information

filename_nc ='../.././FVCOM/seleccion_fecha_de_datos/MESO0001/Mesoescala_0001.nc'

loco_native =reader_netCDF_CF_generic .Reader (filename_nc )

but it crashes with the following errors. Any clue on what is missing? Attached is the full log and the ncdump from the file being used.

Thanks...

16:33:23 DEBUG: Parsing variable: wet_cells_prev_ext /home/mosa/new_opendrift/opendrift/opendrift/readers/reader_netCDF_CF_generic.py:199: RuntimeWarning: invalid value encountered i n float_scalars rel_delta_x.min())/self.delta_x) /home/mosa/new_opendrift/opendrift/opendrift/readers/reader_netCDF_CF_generic.py:202: RuntimeWarning: invalid value encountered in float_scalars rel_delta_y.min())/self.delta_y) 16:33:23 DEBUG: Lon and lat are 1D arrays, assuming latong projection /home/mosa/new_opendrift/opendrift/opendrift/readers/basereader.py:194: RuntimeWarning: invalid value encountered in float_scalars numx = (self.xmax - self.xmin)/self.delta_x + 1 /home/mosa/new_opendrift/opendrift/opendrift/readers/basereader.py:195: RuntimeWarning: invalid value encountered in float_scalars numy = (self.ymax - self.ymin)/self.delta_y + 1 /home/mosa/new_opendrift/opendrift/opendrift/readers/basereader.py:231: RuntimeWarning: divide by zero encountered in double_scalars pixelsize)) + 2 Traceback (most recent call last): File "/var/www/html/FVCOM/www_data/simulations/2019-08-08_16-33-11_964118/2019-08-08_16-33-11_964118.py", line 26, in loco_native =reader_netCDF_CF_generic .Reader (filename_nc )#,gridfile='/var/www/html/Loco/3SE/croco_grd.nc') File "/home/mosa/new_opendrift/opendrift/opendrift/readers/reader_netCDF_CF_generic.py", line 254, in init super(Reader, self).init() File "/home/mosa/new_opendrift/opendrift/opendrift/readers/basereader.py", line 200, in init self.set_buffer_size(max_speed=5) # To be overriden by user/model File "/home/mosa/new_opendrift/opendrift/opendrift/readers/basereader.py", line 231, in set_buffer_size pixelsize)) + 2 OverflowError: cannot convert float infinity to integer

fvcom.ncdump.txt 2019-08-08_16-33-11_964118.txt

[knutfrode]

This file is not on a regular grid (with fixed deltax and deltay), but rather on a triangular, unstructured grid. You have to use another reader for this kind of files, try with either: https://github.com/OpenDrift/opendrift/blob/master/opendrift/readers/reader_FVCOM.py or https://github.com/OpenDrift/opendrift/blob/master/opendrift/readers/reader_netCDF_CF_unstructured.py

These readers (two slightly different variants of the same) are less developed and robst than the regular netCDF-reader, so you might need to do some minor edits to get it working.

[AndresSepulveda]

Sorry, my bad. I didn't saved the .py file and it used the reader for ROMS.

Now, the reader_netCDF_CF_unstructured.py works fine, but the FVCOM results are in UTM coordinates (easting, norting) and as such, OpenDrift finds that all particles are outside the domain.

How can I configure it so it can convert (using proj?) between lat/lon and proj?

[knutfrode]

But according to your ncdump above, the file contains arrays of lon ant lat, and then these should used as coordinates? I don't find anything about UTM in the ncdump output? Can you post the output of >>> print <reader>

[AndresSepulveda]

Indeed, there are lan and lon variables in the file but they are empty (all values equal 0)! Seems the file description is inconsistent, as in the global attributes it says

           :CoordinateSystem = "Cartesian" ;
            :CoordinateProjection = "none" ;

Could it be possible to tell the reader to read x and y, and convert with pyproj to lat and lon?

Here is the output from print

Reader: /var/www/html/FVCOM/FVCOM/seleccion_fecha_de_datos/MESO0001/Mesoescala_0001.nc Projection: +proj=latlong Coverage: [degrees] xmin: 0.000000 xmax: 0.000000 ymin: 0.000000 ymax: 0.000000 Corners (lon, lat): ( 0.00, 0.00) ( 0.00, 0.00) ( 0.00, 0.00) ( 0.00, 0.00) Vertical levels [m]: Not specified Available time range: start: 2009-12-01 00:00:00 end: 2009-12-30 23:00:00.054932 step: 1:00:00.000107 719 times (0 missing) Variables: ocean_sigma/general_coordinate y_sea_water_velocity longitude sea_floor_depth_below_geoid sea_surface_height_above_geoid sea_water_salinity x_sea_water_velocity latitude sea_water_temperature

[AndresSepulveda]

Well, we have a patch for the problem. Thanks to the work of @DanielBrieva who created this script to fill the lat/lon, latc/lonc arrays in the file

[AndresSepulveda]

Sorry, I forgot to attach the file

fill_latlon_FVCOM.py.txt

[gauteh]

Think this will be handled in #243