Combining BGC-Argo floats and satellite observations for water column estimations of the particulate backscattering coefficient

The study examines how satellite-derived bio-optical properties of the ocean surface , when combined with BGC-Argo profiles, can be used to reconstruct the vertical structure of the particulate backscattering coefficient (bbp) throughout the upper 250 meters of the water column. It builds on the SOCA2016 method developed by Sauzède et al. (2016).

Sauzède et al., 2016 — https://doi.org/10.1002/2015JC011408

Related Publication

Full description, methodology and results in:

Combining BGC-Argo floats and satellite observations for water column estimations of the particulate backscattering coefficient
Jorge García-Jiménez, Ana B. Ruescas, Julia Amorós-López, Raphaëlle Sauzède
EGUsphere (2025)
🔗 Read the article

Regions of Interest

• North Atlantic
• Subtropical Gyres

Modeling Approach

• Model type: Multi-output Random Forest Regressor
• Input features:
  • Sentinel-3 OLCI: reflectances at 12 wavelengths, C2RCC IOPs (apig, adet, agelb, bpart, bwit, atot, btot)
  • GlobColour: reflectances at 5 wavelengths
  • GlobalOcean: Sea Level Anomaly (SLA)
  • BGC-Argo Float profiles: temperature, salinity, density, spiciness (via PCA)
  • Mixed Layer Depth (MLD)
  • Spatial-temporal: latitude, longitude, day of year
• Target: Particulate Backscattering Coefficient (Bbp):
  • 26 depths for the 0–50 m model
  • 126 depths for the 0–250 m model

Model Performance – Deep Profiles (0–250 m)

Notebooks Overview

Notebook	Purpose
0_data_analysis.ipynb	Quick exploration of datasets (Argo profiles, temporal distribution...)
1_run_single_experiment.ipynb	Run and validate a model for one region–depth–experiment
2_experiments_analysis_plots.ipynb	Load and compare performance metrics across multiple experiments

How to use

1. Install the environment with Pixi (see below)
2. Launch Jupyter

pixi run jupyter lab

How to Cite

APA

García-Jiménez, J., Ruescas, A. B., Amorós-López, J., & Sauzède, R. (2025). Combining BGC-Argo floats and satellite observations for water column estimations of the particulate backscattering coefficient. EGUsphere. https://doi.org/10.5194/os-21-1677-2025

BibTeX

@article{garcia-jimenez2025bbp,
  author    = {García-Jiménez, Jorge and Ruescas, Ana B. and Amorós-López, Julia and Sauzède, Raphaëlle},
  title     = {Combining BGC-Argo floats and satellite observations for water column estimations of the particulate backscattering coefficient},
  journal   = {EGUsphere},
  year      = {2025},
  doi       = {10.5194/os-21-1677-2025}
}

Project structure

SatArgoBbp/
├── src/
│   ├── data_loader.py          # Load and preprocess dataset
│   ├── feature_selector.py     # Feature selection by experiment type
│   ├── model/
│   │   ├── train.py            # Training pipeline
│   │   └── model_metrics.py    # Evaluation metrics (R², MAE, Bias, etc.)
│   ├── plotting/
│   │   └── plot_utils.py       # Plotting functions for model performance and comparisons
│   ├── evaluation/
│   │   └── metrics_loader.py   # Load and aggregate metric CSVs
│   └── utils.py                # Experiment setup, config classes, I/O helpers
│
├── notebooks/                  # Jupyter notebooks for running and analyzing experiments
│   ├── 0_data_analysis.ypnb
│   ├── 1_run_single_experiment.ipynb
│   └── 2_experiments_analysis_plots.ipynb
│
├── datasets/                   # Processed input datasets (excluded from Git — contact us if interested)
├── results/                    # Saved metrics, plots, models and model outputs
│
├── scripts/ (to do)            
│   └── run_all_experiments.py  # Script to batch-run all experiments
│
├── docs/
│   └── img/                    # Figures for README and manuscript
│
├── .gitignore
├── README.md
├── pixie.toml                 # Project environment and dependencies (managed with Pixi)
└── pixi.lock                  # Pixi lockfile

How to clone the repository

This project uses Pixi for fully reproducible environment management.

Step 1 — Install Pixi (if not already installed)

curl -sSf https://pixi.sh/install.sh | bash

After installation, restart your terminal or reload your shell:

exec $SHELL

Step 2 — Clone the Repository

git clone git@github.com:IPL-UV/SatArgoBbp.git
cd SatArgoBbp

Step 3 — Set Up the Environment

pixi install

This command reads pixi.toml and creates the environment with all dependencies.

---

Step 4 — Activate the Environment

pixi shell

Step 5 — Run Notebooks or Scripts

To launch the Jupyter interface:

pixi run jupyter lab