How to deal with vectors when separating large scales?

[jbusecke]

In yesterdays meeting this issue was brought up, and I will just note it down here to keep track of it.

There were two main issues:

• How can we coarsen/remap/filter velocities
• Which velocities should we consider? We can either use the absolute wind or the wind relative to the ocean surface velocities.

We might be able to get the relative wind from the model output, so we wouldnt have to deal with computing it ourselves.

[paigem]

My understanding is that, if we only focus on the heat fluxes, we do not need to coarsen velocities, just the scalar wind field. @jbusecke is my thinking correct here? If so, we can close this issue for now (or maybe we can rename it if we want an issue for relative winds).

[jbusecke]

I am not sure this is entirely true, especially given the consideration that we will calculate the relative wind to use as input, we might need to give some additional consideration to this question.

My current approach would be:

• interpolate oceanic velocities onto ocean tracer grid position
• interpolate atmospheric wind components onto the same tracer position
• We could then either compute a velocity magnitude and coarsen (we would then feed that as either u or v component with a 0 in the other direction), or coarsen each component separately and feed to the air-sea algorithm.

Given that we might spend a considerable amount of time on this process (including the 'sensitivity' experiments where only wind or temp fields are degraded in resolution) we should perhaps discuss what the 'right' way of doing this should look like.

I would love some more discussion from @ocean-transport/collab_team here.