Closed alessandrofelder closed 3 years ago
Source grid resolution or final output grid resolution? Doing it before remove-restore would certainly help incorporate more of the better resolution data.
@Devaraj-G any thoughts?
We should be able to specify the grid resolution of the output dataset (i.e. merged bathymetry) as an input parameter. The remove-restore can do all the procedures on the grid resolution of the output.
I think after discussions today that this should be an input parameter (maybe optional, defaulting to source grid resolution).
The base grid (and the source grid, if resolution is non-default) should be subsampled to this resolution using grdsample
, as described on page 156 of the GEBCO cookbook (final step of "Gridding - workflow steps B and C").
This is already implemented in the branch feature/upscale-basegrid
using pygmt's surface
instead of grdsample
which appears to work down to grid resolutions of around 100m. At resolutions less than this the output grid starts showing what looks like aliasing errors (however I don't know if this is due to surface
or other parts of the algorithm).
I'll try running it through grdsample too.
Upsampling is now implemented with grdsample
...if I understand the GEBCO cookbook correctly.