parkermac
commented
1 year ago Aurora, you've got the problem cornered, we just don't know yet which corner it is!
Regarding the "constant offset" hypothesis, have a look at w in the final history file at the source location. Do the numbers match what your plot says?
Tiling: the 2x2 tiling is likely to put the source right at the corners of 4 tiles. You could try 3x3 to avoid this, or even better 1x1 (you might have to run oceanS on one core in that case, or maybe you can just set the tiles to 1 and 1 in the .in).
LwSrc Tests in Even Simpler Grid | Weekly Research Update
This week I have been running LwSrc tests in an even simpler grid, courtesy of John Wilkin. I am able to reproduce strange w values in my runs. However, I am unable to reproduce John’s w results even when I run his model without any modification. Maybe something is uniquely problematic with the way I am running ROMS. More details below. Note on all videos: The N-S and E-W axis are all labeled “km.” Unfortunately, after I had already created and processed the videos, I realized that the axis are actually in units of “grid cells.” The domain is 5 x 5 km large, or 41 x 41 grid cells. Note on Volume Conservation Before we dive into the details of the new grid, I’d like to bring closure to the issue of vol
https://ajleeson.github.io/research_blog/2023/04/17/LwSrc-in-simpler-grid.html