Units (m or degrees) in DiffusionUniformKh

[sbravo10]

Question

Question

Hello all,
After going through the code, documentation, and forum, I was unable to determine whether I need to convert the diffusion coefficient when using DiffusionUniformKh. I am using Copernicus current data (Mediterranean Sea Physics Analysis and Forecast dataset) and ERA5 wind data as forcing for my fieldset. My experimental area is the whole Mediterranean. Number of particles near 900k from different line sources.
Since my mesh is spherical, I also loaded the Kh values with the mesh set to spherical. I conducted four tests. The goal was to apply the same Kh value for both zonal and meridional directions—first using 2 m²/s and then 10 m²/s—to compare the results.

Tests:

1. 2 m²/s to degrees:
   Meridional: 1.613 × 10⁻¹⁰ deg²/s
   Zonal: 2.598 × 10⁻¹⁰ deg²/s

2. 10 m²/s to degrees:
   Meridional: 8.066 × 10⁻¹⁰ deg²/s
   Zonal: 1.299 × 10⁻⁹ deg²/s

3. Used 2 m²/s, Left it without converting (code example below)

4. Used 10 m²/s, Left it without converting

Kh_meridional_deg = 2
Kh_zonal_deg = 2  

fieldset.add_constant_field("Kh_zonal",       Kh_zonal_deg,      mesh="spherical")
fieldset.add_constant_field("Kh_meridional",  Kh_meridional_deg, mesh="spherical")

Results

(1) and (2) were I used deg²/s there were not any noticeble difference compared to the results obtained when not applying diffusion.
(3) and (4) had the results I would expect.

However since I did not find any conversion on the code from the kernel I was not able to determine if the conversion is being done somewhere else.
Unfortunatelly I was not able to look into the code and documentation previously used to Brownian kernel, which was discontinued, from a topic related to mine - #515.

I guess my question is quite straightfoward to developers and users. I appreciate a lot for a response. Thank you!

[erikvansebille]

Thanks for your question, @sbravo10. When you provide fields with names Kh_zonal and Kh_meridional and spherical mesh, then Parcels automatically attaches a UnitConverter object to them, and all the conversion from meters to degrees happens under the hood (taking into account also the sphericity of the earth).
See https://docs.oceanparcels.org/en/latest/examples/tutorial_unitconverters.html for more information

So approach 3 and 4 are correct

[sbravo10]

Wow, just amazed on how spot on the answer is and how I managed to not see it before. Thank you very much.