MedClim

Mediterranean hydrographic processing and visualization workflows used for the analysis of accelerating Mediterranean warming and salinification (Terzić & Vilibić, Geophysical Research Letters).

Repository scope

This repository is organized around the tracked analysis scripts used to preprocess source datasets, build basin-based climatologies and time series, fit trends, and generate manuscript figures. Large source data files and most generated products are intentionally excluded from version control.

Running the code

• environment.yml lists the Python packages, pinned to the versions the manuscript products were made with (conda env create -f environment.yml). It also installs the two packages that are not on PyPI, Basins (basin polygons) and MapPlotter (satellite basemaps), from GitHub.
• Every script resolves its input and output locations through Scripts/paths.py. The repository root is taken from that file's own location, and the data trees Data/, Stats/, Plots/, Results/ and Diagnostics/ live under the repository unless the environment variable MEDCLIM_DATA points somewhere else. The Basins package's manca.py and the JADMON 2024 delivery are the only inputs outside the repository; Scripts/paths.py says where they are expected and which variables (MEDCLIM_BASINS, MEDCLIM_JADMON_2024) override that.
• run_all.sh runs the stages of the main workflow below in order, or a named selection of them (./run_all.sh --list).
• Each script starts with two comment lines, Reads and Writes, naming its inputs and outputs relative to the repository.
• CITATION.cff gives the citation for the code (Zenodo concept DOI 10.5281/zenodo.20410480) and for the article.

Main workflow

1. Scripts/Seadatanet/preprocessing/clean_seadatanet_odv_export.py cleans the SeaDataNet export and prepares basin-tagged profile files.
2. Scripts/Argo/acquisition/ and Scripts/Argo/preprocessing/ download, quality-control, and basin-tag Mediterranean Argo profiles.
3. Scripts/Adriatic/preprocessing/ ingests the Adriatic regional datasets and exports harmonized long-format tables. These four readers are the provenance for the cleaned Adriatic datasets deposited in the accompanying Zenodo data record. Scripts/Adriatic/diagnostics/fix_adriatic_outliers_per_basin.py then reassigns the casts the readers left outside the three Adriatic basins (BASIN_3ADR == 'out') to DJ1/DJ2/DJ3 by nearest centroid and writes the *_long_fixed.csv tables in Data/Adriatic/Processed/ that the later stages read.
4. Scripts/MultiPlatform/filtering_qc/remove_coastline_data.py applies the coastal and estuarine exclusion masks used before pooled analysis.
5. Scripts/Seadatanet/preprocessing/identify_argo_copies_in_seadatanet.py lists the SeaDataNet casts that are Argo float profiles already held by the Argo set (the SeaDataNet aggregation archives float cycles as casts of its own), and Scripts/Seadatanet/preprocessing/remove_argo_copies_from_seadatanet.py writes the SeaDataNet master file without them (*_no_argo_copies.csv), which every downstream script reads. The list of the 21,926 affected casts is tracked as Scripts/Seadatanet/preprocessing/sdn_argo_copies_pids.txt. See CHANGELOG.md.
6. Scripts/MultiPlatform/core/multi_source_seasonal_climatology.py builds the seasonal climatology products.
7. Scripts/MultiPlatform/core/build_monthly_timeseries_all_bins.py and Scripts/MultiPlatform/core/build_monthly_timeseries_composed_from_atomic.py create monthly basin-by-depth series.
8. Scripts/MultiPlatform/core/fit_trends_all_bins.py fits the linear and quadratic trend products used in the manuscript.
9. Scripts/MultiPlatform/core/save_metadata.py rebuilds map-support metadata tables.

Manuscript figures

• Figure 1: Scripts/Publication_Papers/dataset_coverage.py
• Figure 2: Scripts/Publication_Papers/century_differences_composite.py
• Figure 3: Scripts/Publication_Papers/maps_trends.py
• Figure 4: Scripts/Publication_Papers/vertical_trends_heatmaps.py
• Supplementary vertical-structure figures: Scripts/Publication_Papers/supplement_vertical_trend_bars_other_basins.py
• Robustness checks: Scripts/Additional/robustness_model_comparison.py and Scripts/Additional/sensitivity_test_trends.py

Notes

• The tracked repository is script-focused. Large raw inputs and most derived data files are ignored to keep the repository lightweight.
• The data layout is the one used during manuscript production, described in Scripts/paths.py. The workflow documentation in Scripts/*/README.md identifies which scripts were used for each stage.
• The mapping scripts parse basin rectangles from the Basins package's own manca.py, located by Scripts/paths.py (installed package first, or a checkout named by MEDCLIM_BASINS).

Data availability

Argo and SeaDataNet source data are publicly available from their original repositories and are cited in the manuscript; they are not redistributed here. The cleaned regional Adriatic in situ datasets are deposited in a separate Zenodo record, produced by the readers in Scripts/Adriatic/preprocessing/.

License

Distributed under the Creative Commons Attribution 4.0 International license (CC BY 4.0).