**Appendix A**

**Table A1.** Max Root-mean-squared (RMS) error per variable when comparing Serial GCCOM vs. Parallel GCCOM Model from 1 to 240 processors across 12 nodes.


**Table A2.** Max RMS per variable when comparing Parallel GCCOM Model outputs from 20 to 240 processors across 12 nodes vs. Parallel GCCOM output in a single processor.



**Table A3.** Parallel performance of the high resolution stratified seamount experiment (3000 × 200 × 100) in the COD system.

**Table A4.** Parallel performance of the medium-sized stratified seamount experiment (2000 × 100 × 100) in the COD system.


**Table A5.** Parallel performance of the low resolution stratified seamount experiment (1500 × 100 × 50) in the COD system.

