Compilation of the dataset and estimation of Ω, as described in the Methods (Section 2.1 and 2.2). Contains a sparate README for the workflow.
Validation of our method to estimate Ω, described and shown in SI.
-
glodap: Validation of method using recent GLODAPv2.2022 data
-
cdisk4: Comparison to measured values from the CDisK-IV cruise
Analysis of the compiled dataset (Section 3).
-
dissolution-above-saturation-horizon: Investigating dissolution above the saturation horizon (Section 3.3.1)
-
omega-crit: Investigating the existence of two dissolution regimes (Section 3.3.2)
Machine learning part (Feature Importance) as described in the Methods (Section 2.3) and discussed in Section 3.2.
Note that there is a separate environment.yml file for all jupyter notebooks in this folder.
For part of the code, data from World Ocean Atlas 2018 is required. To run the code, download the files that are listed in 'required-woa18-files.txt' and save them locally. Copy the path to the folder with those files into the textfile 'path-to-woa18-files.txt'.
-
Figure 3: analysis/map-experiments.ipynb
-
Figures 4, 5, 6, S12: analysis/overview-compilation.ipynb
-
Figures 7, 9, S14, S15: xgboost/interpret-model-pfi-pdp.ipynb
-
Figure 8: xgboost/plot-model-prediction.ipynb
-
Figures 10, 11, S16: analysis/dissolution-above-saturation-horizon/m77-dissolution.ipynb
-
Figures 12, 13, S17: analysis/omega-crit/Omega-crit-linear-regression-erf.ipynb
-
Figure 14: analysis/omega-crit/different-omega-crit.ipynb
-
Figures S1, S2, S3, S4, S5, S6: validation/glodap/glodap22-recent-woa-canyonb-comparison.ipynb
-
Figure S7, S8: validation/nutrient-comparison-WOA-CANYONB.ipynb
-
Figures S9, S10: validation/cdisk4/cdisk4-ctd-woa-canyonb-data-comparison.ipynb
-
Figure S11: validation/cdisk4/cdisk4-ctd-woa-canyonb-data-comparison.ipynb
-
Figure S12: analysis/experiment-conditions-representative.ipynb
-
Figures S13: validation/uncertainty-calculations.ipynb