Approaches to nest high resolution subset of a domain (meshing, smoothing)

felicitymccormack

Hello forum!

I have some very high resolution bed topography realisations in a subset of a domain and I'd like to conduct some sensitivity analyses using these data. The data are not consistent with the bed topography for the rest of the domain (i.e. they have fine-scale topographic roughness that is not in the underlying BedMachine Antarctica data I'm using).

I have generated a finer mesh (consistent with the resolution of the dataset ₅₀₀ m) in the subset region of interest, but I'm wondering if anyone has insight on approaches to deal with blending the edges between the high res and background datasets. I thought I could smooth the edges of the high res data in the subset, and then reduce the mesh resolution in a buffer zone (~ 20 km) from the 500 m in the subset to the ₁₅ km in the interior the domain.

Has anyone done anything similar? Any thoughts on whether this is a good / terrible approach??!

Cheers,
Felicity

mathieumorlighem

Hi Felicity

I was hoping someone else would reply because this is not an easy question... In short, there is no recipe, and it all depends on how abrupt the transition is between the two datasets. If you have high cliffs because the high-res and low-res beds are not compatible, you will need to do some smoothing at the transition. If the high-res bed is similar to the low-res bed but with more roughness, then you probably don't need to do anything.
You can let bamg deal with the model resolution, forcing 200 m where you have a high resolution bed, and a more standard resolution outside, with a smooth transition in mesh resolution between then two (using the option 'gradation').
I hope this helps!
Mathieu

felicitymccormack

Thanks heaps Mathieu,

I tried the meshing with gradation, but there are sharp cliffs that remain between the high res subset and the background data. I'll try incrementally increasing the buffer on the subset region over which the gradation is applied - hopefully I'll find something that works there. Will get back to you!

felicitymccormack

The promised update: I tried a very quick and dirty approach to smooth in the region of the cliffs in addition to a 20 km buffer within which I applied mesh gradation in bamg. The final result was not ideal, but it fixed the main issue of sharp discontinuities.

However, a better approach is to make the high resolution bed topography consistent with the underlying smoothed bed at the edges, so we'll follow up on that.

Thanks for your help Mathieu!