Correct cell not found

[norafried]

I'm using POP and want to release particles at different depths for about two years:

When I'm executing my code I get the following message:

Correct cell not found for (-7.563470, 63.168785) after 1000000 iterations
Debug info: old particle indices: (yi, xi) 207 399
new particle indices: (yi, xi) 207 0
Mesh 2d shape: 401 401
Relative particle position: (xsi, eta) 1.00264628104553

What exactly does it mean and can I do something to stop it? It only happens after a certain amount of time.

[erikvansebille]

Hi @norafried. This error occurs when Parcels can't find in which cell a particle is located. This sometimes happens when grids are highly curved. In your case, it seems to happen on the zonal boundary (zonal cell was 399 first, then 0). So my hunch would be that this happens when your particles cross your zonal reentrant boundary.

You could have a look at this discussion (for the Hycom grid): #1240. A solution that worked in that case is at #1240 (comment). Perhaps it works in your case too?

[norafried]

How can I figure out it's on the domain boundary? For sure my data is limited to the subpolar North Atlantic. So in my head it makes sense that there are particles trying to leave the area. Would make sense to just end their path? I don;t really care where they leave to I would say?

[erikvansebille]

You can plot the trajectories?

[norafried]

Yes I can. I get a normal output and 'normal' trajectories. Therefore I was wondering whether it is an error or more a warning.

[erikvansebille]

OK, I didn't realise this error only happens very incidentally; I thought that you got a lot of these and that it was hampering your analysis...

As I said above, the error occurs when Parcels can't find in which cell a particle is located. This sometimes happens when grids are highly curved (e.g. close to a pole). See https://github.com/OceanParcels/parcels/blob/master/parcels/include/index_search.h#L318-L452 for the full code, and then in particular the code below for the error message https://github.com/OceanParcels/parcels/blob/56e988ae0f627ec8994c656f5f4bab71154a2e47/parcels/include/index_search.h#L406-L412
You could try to increase the maxIterSearch parameter, but that typically doesn't help in my experience.

The problem likely is that the grid cells are so skewed and the particle is so close to an edge, that within numerical precision, it's correct grid cell can't be found. So there's not much you can do about it, the particle will be deleted. If it occurs only a handful of times, I would not worry about it and simply ignore the particles (they will be shorter in the output file)

[norafried]

Okay, thanks. I will see how many time it occurs with a longer run. It did occur more than a handful of times. But I guess depending on how many particles you release that number might scale up.