Drifter trajectories (NEMO outputs on ORCA36 fields)

[clementbricaud]

Parcels version

parcels 3.1.2

Description

Hello
I am trying to launch parcels with NEMO outputs on ORCA36 fields.

I want to compare trajectories with drifters comoing from https://data.marine.copernicus.eu/product/INSITU_GLO_PHY_UV_DISCRETE_NRT_013_048/description
The trajectories are initialised at 15m depth.

As ORCA36 is huge for memomry, I made an extraction of the first 10 levels of the model:
depthv = 0.50576, 1.555855, 2.667682, 3.85628, 5.140361, 6.543034, 8.092519,
9.82275, 11.77368, 13.99104, 16.52532, 19.4298, 22.75762

I have this errors and I don't understant where it comes from:
out_of_bound_error
raise FieldOutOfBoundError(f"Field sampled out-of-bound, at (depth={z}, lat={y}, lon={x})")
parcels.tools.statuscodes.FieldOutOfBoundError: Field sampled out-of-bound, at (depth=15.0, lat=27.415000915527344, lon=-61.00699996948242)

Code sample

# Paste your code within this block

# Paste your error message within this block

[erikvansebille]

Hi @clementbricaud, what is the horizontal extent of your fieldaet? What is the output of print(fieldset)? Could it be that -61 lon is outside the domain? What if you add 360 degrees to all your practice longitudes?

[clementbricaud]

Hello
thank you very much for your help !

I am trying that:
fieldset.U.grid.lon = np.where(fieldset.U.grid.lon > 180, fieldset.U.grid.lon-360, fieldset.U.grid.lon)

but I have an error:

fieldset.U.grid.lon = np.where(fieldset.U.grid.lon > 180, fieldset.U.grid.lon-360, fieldset.U.grid.lon)
^^^^^^^^^^^
AttributeError: property 'lon' of 'CurvilinearZGrid' object has no setter

print(fieldset) gives:

fields: name : 'U' grid : CurvilinearZGrid(lon=array([[ 73.00, 73.03, 73.06, ..., 72.92, 72.94, 72.97], [ 73.00, 73.03, 73.06, ..., 72.92, 72.94, 72.97], [ 73.00, 73.03, 73.06, ..., 72.92, 72.94, 72.97], ..., [ 73.00, 73.00, 73.00, ..., 73.00, 73.00, 73.00], [ 73.00, 73.00, 73.00, ..., 73.00, 73.00, 73.00], [ 73.00, 73.00, 73.00, ..., 73.00, 73.00, 73.00]], shape=(10842, 12960), dtype=float32), lat=array([[-84.20, -84.20, -84.20, ..., -84.20, -84.20, -84.20], [-84.19, -84.19, -84.19, ..., -84.19, -84.19, -84.19], [-84.20, -84.20, -84.20, ..., -84.20, -84.20, -84.20], ..., [ 50.00, 50.00, 50.00, ..., 50.00, 50.00, 50.00], [ 50.00, 50.00, 50.00, ..., 50.00, 50.00, 50.00], [ 50.00, 50.00, 50.00, ..., 50.00, 50.00, 50.00]], shape=(10842, 12960), dtype=float32), time=array([ 0.00, 86400.00, 172800.00, 259200.00]), time_origin=2023-02-01T12:00:00.000000000, mesh='spherical') extrapolate time: True time_periodic : False gridindexingtype: 'nemo' to_write : False name : 'V' grid : CurvilinearZGrid(lon=array([[ 73.00, 73.03, 73.06, ..., 72.92, 72.94, 72.97], [ 73.00, 73.03, 73.06, ..., 72.92, 72.94, 72.97], [ 73.00, 73.03, 73.06, ..., 72.92, 72.94, 72.97], ..., [ 73.00, 73.00, 73.00, ..., 73.00, 73.00, 73.00], [ 73.00, 73.00, 73.00, ..., 73.00, 73.00, 73.00], [ 73.00, 73.00, 73.00, ..., 73.00, 73.00, 73.00]], shape=(10842, 12960), dtype=float32), lat=array([[-84.20, -84.20, -84.20, ..., -84.20, -84.20, -84.20], [-84.19, -84.19, -84.19, ..., -84.19, -84.19, -84.19], [-84.20, -84.20, -84.20, ..., -84.20, -84.20, -84.20], ..., [ 50.00, 50.00, 50.00, ..., 50.00, 50.00, 50.00], [ 50.00, 50.00, 50.00, ..., 50.00, 50.00, 50.00], [ 50.00, 50.00, 50.00, ..., 50.00, 50.00, 50.00]], shape=(10842, 12960), dtype=float32), time=array([ 0.00, 86400.00, 172800.00, 259200.00]), time_origin=2023-02-01T12:00:00.000000000, mesh='spherical') extrapolate time: True time_periodic : False gridindexingtype: 'nemo' to_write : False name: 'UV' U: V: W: None

[VeckoTheGecko]

@clementbricaud it would be really helpful if we could see the code that you are running. Would you be able to copy/paste your full script here so that we can have a look?

Are you following our tutorials listed on our documentation website? Perhaps if you let us know which tutorial you're following/code you're adapting we can point you in the right direction?

[clementbricaud]

if (grid_regularity == 'native'):
  fieldset_current = FieldSet.from_nemo(filenames, variables, dimensions, allow_time_extrapolation=True)

fieldset = fieldset_current

    if ('GLO36V1_' in model):
        print(fieldset)
        print(fieldset.U)
        print("-------------- fieldset.U.grid")
        print(fieldset.U.grid)
        print("-------------- fieldset.U.grid.lon")
        print(fieldset.U.grid.lon)
        print(fieldset.U.grid.lon[5000,0])

        #fieldset.U.grid.lon = np.where(fieldset.U.grid.lon > 180, fieldset.U.grid.lon-360, fieldset.U.grid.lon)
        fieldset.U.lon = np.where(fieldset.U.lon > 180, fieldset.U.lon-360, fieldset.U.lon)

[VeckoTheGecko]

Hi @clementbricaud , this is not the full script (i.e., the script from top to bottom, or in the case of a notebook - the code from all the cells). Its really difficult to see from the partial snippet here what you're trying to do, or what you've attempted so far. There just isn't enough information here.

Also fieldset.U.grid.lon = is invalid code for Parcels, which is why I was asking whether you were following a tutorial. Which code did you start working with? Was it an example from our documentation? Was it someone else's code?

Also, please use code blocks so that your formatting is preserved. You can write like the following

```python
print("your code goes here")
```

which results in the following being rendered

print("your code goes here")

[clementbricaud]

test_parcel.py.txt

Hello,
thank you for your help

I have tried the example_nemo_curvilinear.py example from https://docs.oceanparcels.org/en/latest/documentation/additional_examples.html with my model fields from ORCA36.
my python script is attached to this message

I have the same error: File "/perm/ar5/conda/envs/parcels/lib/python3.13/site-packages/parcels/tools/statuscodes.py", line 61, in _raise_field_out_of_bound_error
raise FieldOutOfBoundError(f"Field sampled out-of-bound, at (depth={z}, lat={y}, lon={x})")
parcels.tools.statuscodes.FieldOutOfBoundError: Field sampled out-of-bound, at (depth=0.0, lat=47.0, lon=-17.0)

[clementbricaud]

it works well with ORCA12 model outputs

[clementbricaud]

do you have an idea of where the test is realized ?
I don't the function that checks that the particle in inside the domain and where it change the particule error status

[clementbricaud]

and when I search ErrorOutOfBounds in the code,
I don't sea where this value is used

./field.py: than ErrorOutOfBounds is thrown, the function is stopped. Otherwise, next field is interpolated.
./field.py: NestedField returns an ErrorOutOfBounds only if last field is as well out of boundaries.
./kernel.py: elif p.state == StatusCode.ErrorOutOfBounds:
Fichier binaire ./pycache/field.cpython-313.pyc correspondant
Fichier binaire ./pycache/kernel.cpython-313.pyc correspondant
./tools/statuscodes.py: ErrorOutOfBounds = 60
./tools/statuscodes.py: FieldOutOfBoundError: StatusCode.ErrorOutOfBounds,
Fichier binaire ./tools/pycache/statuscodes.cpython-313.pyc correspondant

[clementbricaud]

Hello
is it necassary for parcels that the NEMO model grid contains the east/west and north overlapping bands in the model grid ?
because it works with ORCA12 ( it is based on nemo 3.6 and the grid contains overlapping bands)
but for ORCA36, it is based on NEMO 4.2 and it not cotains overlapping bands in the grid

[clementbricaud]

with ORCA36, in _search_indices_curvilinear, I have eta > 0 with ORCA36

[clementbricaud]

sorry ...eta <0