Tutorial on how to use Parcels on NEMO curvilinear grids¶

Parcels also supports curvilinear grids such as those used in the NEMO models.

We will be using the example data in the NemoCurvilinear_data/ directory. These fiels are a purely zonal flow on an aqua-planet (so zonal-velocity is 1 m/s and meridional-velocity is 0 m/s everywhere, and no land). However, because of the curvilinear grid, the U and V fields vary north of 26N.

In [1]:
from parcels import FieldSet, ParticleSet, JITParticle, ParticleFile, plotTrajectoriesFile
import numpy as np
from datetime import timedelta as delta


We can create a FieldSet just like we do for normal grids. Note that we need to provide an extra filename for the mesh_mask file that holds information on the Curvilinear grid.

In [2]:
filenames = {'U': 'NemoCurvilinear_data/U_purely_zonal-ORCA025_grid_U.nc4',
'V': 'NemoCurvilinear_data/V_purely_zonal-ORCA025_grid_V.nc4',
variables = {'U': 'U',
'V': 'V'}
dimensions = {'lon': 'glamf', 'lat': 'gphif'}
field_set = FieldSet.from_nemo(filenames, variables, dimensions, mesh='spherical', allow_time_extrapolation=True)

/Users/delandmeter/anaconda2/envs/py2_parcels/lib/python2.7/site-packages/xarray/conventions.py:416: SerializationWarning: Unable to decode time axis into full numpy.datetime64 objects, continuing using dummy netCDF4.datetime objects instead, reason: dates out of range
result = decode_cf_datetime(example_value, units, calendar)
/Users/delandmeter/anaconda2/envs/py2_parcels/lib/python2.7/site-packages/xarray/conventions.py:435: SerializationWarning: Unable to decode time axis into full numpy.datetime64 objects, continuing using dummy netCDF4.datetime objects instead, reason: dates out of range
calendar=self.calendar)
WARNING: Casting lon data to np.float32
WARNING: Casting lat data to np.float32
WARNING: Casting time data to np.float64
WARNING: Casting depth data to np.float32


And we can plot the U field. Note that we need to rescale the grid.lon so that it is monotonic (needed for plotting purposes only; Parcels itself can handle non-monotonic grids)

In [4]:
# Remap the longitudes because field.show() requires monotonic longitudes
field_set.U.grid.lon[field_set.U.grid.lon<73] += 360
field_set.U.show()

# Now reverse the remapping above
field_set.U.grid.lon[field_set.U.grid.lon > (72+360)] -=360


As you see above, the U field indeed is 1 m/s south of 50N, but varies with longitude and latitude north of that

Now we can run particles as normal. Parcels will take care to rotate the U and V fields

In [5]:
# Start 20 particles on a meridional line at 180W
npart = 20
lonp = -180 * np.ones(npart)
latp = [i for i in np.linspace(-70, 88, npart)]

# Create a periodic boundary condition kernel
def periodicBC(particle, pieldSet, time, dt):
if particle.lon > 180:
particle.lon -= 360

pset = ParticleSet.from_list(field_set, JITParticle, lon=lonp, lat=latp)
pfile = ParticleFile("nemo_particles", pset, outputdt=delta(days=1))

INFO: Compiled JITParticleAdvectionRK4periodicBC ==> /var/folders/h0/01fvrmn11qb62yjw7v1kn62r0000gq/T/parcels-503/1d71ca6ee352199d6fc370d8a50fff55.so

plotTrajectoriesFile("nemo_particles.nc")

<module 'matplotlib.pyplot' from '/Users/delandmeter/anaconda2/envs/py2_parcels/lib/python2.7/site-packages/matplotlib/pyplot.pyc'>