Lauderdale and Cael "Impact of remineralization profile shape on the air-sea carbon balance" manuscript repository.

Model code, output, and analysis routines looking at the impact of remineralization profile shape (i.e. functional form) on the air-sea carbon balance, and characterizing structural uncertainty in the ocean's biological pump. Read the paper in Geophysical Research Letters or find the pre-print in the Earth and Space Science Open Archive.

Fitting the remineralization profiles:

Generate all the coefficients for the six functional forms statistically fit to the reference power-law curve in three different ways using the MATLAB routine profile_coefficients.m. Coefficients are stored in export_profile_coefficients.csv, with the coefficients used in the manuscript supplied here.

Numerical model code and configuration:

Simulations were run with the "Checkpoint63m" version of MITgcm, substituting files in the code_mods directory. pkg DIC files DIC_VARS.h, dic_readparms.F, and phos_flux.F supply the different remineralization functions (with an added function for exponential integral, or upper incomplete gamma function expint.F). The functional forms can each be activated by setting selectExportRemin between 1--7 in data.dic, where:

• selectExportRemin=1 for the simple exponential profile,

• selectExportRemin=2 for the power-law curve,

• selectExportRemin=3 for the ballast profile,

• selectExportRemin=4 for the rational profile,

• selectExportRemin=5 for the double exponential profile,

• selectExportRemin=6 for the stretched exponential profile,

• selectExportRemin=7 for the gamma function profile.

Preformed tracers, which aid in partitioning carbon and nutrients into "physically-preformed" and "biologically-regenerated" nutrients are included in the code_mods directory, by substituting the default pkg GCHEM files GCHEM_OPTIONS.h and gchem_forcing_sep.F, and adding gchem_preformed_tracers.F. Activate the tracers by increasing the number of tracers associated with pkg PTRACERS, and include definitions in data.ptracers file (see experiment input files for details).

The same executable was used for each simulation. Coefficients for each remineralization function are passed using the four-value KRemin array in data.dic (filled with zeroes tot he right where needed).

Numerical model simulations:

Each simulation is associated with its own folder in this repository, which includes input files and steady-state output as an average of the last 100 years of the siumulation. The naming convention is:

• mar0.70, mar0.84, or mar0.98 for the reference simulations using the Martin Curve with b values of 0.70, 0.84 (reference), and 0.98.

• abs, rel, or efd for fitting to the reference profile by minimizing absolute error, relative error, or e-folding depth of remineralization.

• exp, bal, dbl, str, rat, or gam for the remineralization profile used (simple exponential, ballast, double exponential, stretched exponential, rational, and gamma functions).

• noflux for the simulation where no particulate organic carbon is produced at the surface, and 100% of production is channeled instead to dissolved organic carbon that degrades with a timescale of 6 months.

Analysis:

The Jupyter Notebook export_flux_analysis.ipynb contains the analysis routines used to generate the figures. Dependencies include xarray, dask, and mitgcm_tools (get from github).

Any questions or comments, please get in contact!