Global chlorophyll-a and environmental driver datasets for regime-dependent variability analysis (1998–2022)

Ocean Productivity Data Set (1.Chl_a.nc)

• Global Ocean Chlorophyll-a from Copernicus Marine Biogeochemical Reanalysis (1998–2022)
• Variable: Chlorophyll-a concentration (Chl-a)
• Area: Global Ocean (60°S ~ 60°N, -180°W ~ 180°E)
• Period: 1998-01-01 ~ 2022-12-31
• Horizontal resolution: 0.25° × 0.25°
• Temporal resolution: Monthly
• Source: https://data.marine.copernicus.eu/product/GLOBAL_MULTIYEAR_BGC_001_029/description

Ocean Environmental Data Set (2.SST.nc, 3.SSHF.nc, 4.MLD.nc, 5.WS.nc, 6.NO3.nc, 7.PO4.nc, 8.Fe.nc, 9.Si.nc)

• Global ocean physical and biogeochemical environmental variables (1998–2022)
• Area: Global Ocean (60°S ~ 60°N, -180°W ~ 180°E)
• Period: 1998-01-01 ~ 2022-12-31
• Horizontal resolution: 0.25° × 0.25°
• Temporal resolution: Monthly
• All datasets were regridded to a common spatial resolution and synchronized with the chlorophyll-a dataset.
• Variables:
  SST – Sea surface temperature (°C)
  SSHF – Surface sensible heat flux (W m⁻²)
  MLD – Mixed layer depth (m)
  WS – 10 m wind speed (m s⁻¹)
  NO₃ – Nitrate concentration (µmol L⁻¹)
  PO₄ – Phosphate concentration (µmol L⁻¹)
  Fe – Dissolved iron concentration (nmol L⁻¹)
  Si – Silicate concentration (µmol L⁻¹)
• Sources:
  ERA5 Reanalysis : https://cds.climate.copernicus.eu/datasets/reanalysis-era5-single-levels-monthly-means?tab=download
  ORAS5 Ocean Reanalysis : https://cds.climate.copernicus.eu/datasets/reanalysis-oras5?tab=download
  Copernicus Marine Biogeochemical Reanalysis : https://data.marine.copernicus.eu/product/GLOBAL_MULTIYEAR_BGC_001_029/description

Pre-processing code (Set_data_1~3)

• Set_data_1 : Merging downloaded monthly raw datasets into a single NetCDF file for each variable over the full study period.
• Set_data_2 : Cropping the merged datasets to the study domain used in this study (60°S–60°N, 180°W–180°E).
• Set_data_3 :  Remapping all cropped variables to the chlorophyll-a reference grid so that all datasets share the same latitude, longitude, and spatial resolution.

Analysis code (code1~4)

• code_1 : This script calculates linear trends for each variable at every grid point using the de-seasonalized anomaly datasets. For each variable, linear regression is performed along the time dimension, and the slope field is saved as a NetCDF file. The variables processed include chl-a, SST, SSHF, MLD, WS, NO₃, PO₄, Fe, and Si.
• code_2 : This script classifies grid cells into three quantile-based trend regimes using the chl-a slope field. Grid cells at or below the 20th percentile are defined as the lower 20%, grid cells between the 20th and 80th percentiles are defined as the middle 60%, and grid cells at or above the 80th percentile are defined as the upper 20%. The resulting regime map is saved for later analyses.
• code_3 : This script uses the quantile classification map to save anomaly datasets separately for each regime. For each variable, anomaly fields are masked so that only pixels belonging to the selected regime are retained. Separate NetCDF files are saved for the lower 20%, middle 60%, and upper 20% regions.
• code_4 : This script performs pointwise multivariate EOF analysis for the lower and upper chl-a trend regimes. At each valid grid point, the anomaly time series of chl-a, SST, SSHF, MLD, WS, NO₃, PO₄, Fe, and Si are standardized, a covariance matrix is constructed, and eigen-decomposition is applied to obtain EOF loadings, principal components, and explained variance ratios.

Acknowledgments

This work was also supported by Global - Learning & Academic research institution for Master’s·PhD students, and Postdocs (LAMP) Program of the National Research Foundation of Korea (NRF) grant funded by the Ministry of Education (No. RS-2023-00301914).