Sounds like a great idea, thanks @ChrisC28. How coastal is the obs data? Bear in mind that the model has no coastal source of nutrients, ie runoff carries no BGC tracers…
The data extends from over the continental shelf (very coastal) to ~100km off shore (not coastal). The region might see a run-off signal in very wet year, but is normally dominated by blue-water/shelf interaction.
Hi all,
Recently, I conducted perturbation tests on ADIC based in IAF-Cycle4 outputs. Specifically, I made localized changes to the ADIC concentration of a certain magnitude, and then compared these changes with the ADIC in the control group (cycle4) to explore how the perturbations spread. Due to storage limitations, most of the BGC variables in cycle4 have reduced precision (to 3 or 4 significant figures). I found that for specific purposes requiring high-precision comparisons (such as comparing perturbation experiments), this reduction in precision has a significant impact. I am posting the issue I encountered here:
The figure shows the impact range (control - perturbation) of the initial ADIC perturbations created at three locations (numbered 4, 7, and 12) after 4 years. It is evident that compared to the restart file, the output file loses a significant amount of low perturbation region information. Given the same significant figures, the magnitude of the variables determines the magnitude of the remaining perturbation information. Consequently, the horizontal sum also has a considerable error.
Additionally, if I can get some SUs at the end of the quarter, I am preparing to rerun several years IAF-Cycle 4 starting from 2004 for perturbation experiment comparisons. The high-precision outputs currently planned to be saved include ADIC and PCO2.
If this rerun would be helpful for anyone’s research, please let me know, and we can discuss the high-precision or high-frequency variables that need to be saved.
Thank you