search

NorESMhub / BLOM

Bergen Layered Ocean Model
GNU Lesser General Public License v3.0
16 stars 25 forks source link

Fix bug in cgs to mks conversion for tracer field #334

Closed JorgSchwinger closed 5 months ago

JorgSchwinger commented 6 months ago

This PR fixes a deviation from expected results when switching the BLOM_UNIT option to mks. With this fix global integrated time series remain close to the previous results, and spatial variation are reduced much. There are some remaining differences, but these are relatively small and not larger (and in the same places) than in physical variables (SST and SSS). As shown below for alkalinity, there were huge differences after 100 years in simulations without this fix (labeled 34), which are greatly reduced in simulations including this fix (labelled 35).

The treatment of virtual fluxes is still not (and was not) correct, I believe, but I suggest to fix this later, and first establish the expected behavior when moving to mks.

NorDiag_2d_T62_tn21_31-T62_tn21_34_kms_old_Global_lev01_diff_0100_srftalk NorDiag_2d_T62_tn21_31-T62_tn21_35_kms_new_Global_lev01_diff_0100_srftalk NorDiag_gts_T62_tn21_31-T62_tn21_34-T62_tn21_35_mkstest_bgc_0001-0100.pdf

JorgSchwinger commented 5 months ago

For the record, I also checked that this results in bit-identical results if cgs is used.