OMZs are regions where O2 concentration drops below 60 μM, challenging the living environment for marine biota but enabling bacterial activity of pivotal importance for the marine nitrogen cycle and hence the carbon cycle.

There is a wide variety of ocean circulation and ocean biological regimes that lead to the formation of the OMZs. These regimes have been classified by Mikaloff-Fletcher et al., 2006, including well ventilated open ocean areas, upwelling regions or polar oceans.

Two O2 concentration boundaries have been widely accepted in the marine scientific community as the critical thresholds for biological stress. These are called hypoxia, with O2 lower than 60 μM, and suboxia, with almost a complete depletion of O2 below 5 μM.

Oceanic O2

Oceanic O2 is one of the most challenging biogeochemical variables to be represented in contemporary Ocean General Circulation and Biogeochemical Models (OGCBMs). The occurrence of very low dissolved O2 concentrations, known as Oxygen Minimum Zones, is a combination of stagnated or poorly ventilated areas with a prominent biological activity associated with high rates of O2 consumption.

Low oxygenated waters, with either hypoxic or suboxic regimes, have been summarized in the World Ocean Atlas* data product as presented here. Hypoxia is present in the three major oceanic basins. The largest hypoxic area is located in the Pacific and it expands from the Eastern Tropical Pacific (ETP) towards the northern hemispheric part of the basin. The hypoxic areas in the north Pacific are located around 600m deep. Hypoxia in the Indian ocean is concentrated in the Arabian Sea and the Bay of Bengal. These hypoxic areas are much shallower than those from the Pacific, around 100m deep. Two small areas of hypoxia appear in the Atlantic: the Benguela Upwelling System (BUS) and the area off-coast of Senegal. Suboxia is exclusively found at the ETP and the Arabian Sea.

* The interpolation techniques used in the World Ocean Atlas have been revised by Bianchi et al., 2012., particularly in regions where the low O2 regimes are found, i.e., hypoxia and suboxia. Bianchi et al., 2012. released a new data-based product with larger hypoxic and suboxic volumes than the ones from World Ocean Atlas 2005. In this data visualization project I have used the revised WOA2005 for greater accuracy when depicting the Oxygen Minimum Zones.

Earth System Models

Earth System Models (ESM) included into the Coupled Model Intercomparison Project 5 (CMIP5, Taylor et al., 2012) are used in this model evaluation and help to frame the envelope of uncertainties when estimating oceanic O2.

The set of ESMs who participated in the CMIP5 project share the same scope in terms of simulated time periods and applied forcings, with a standarized model output of relevant physical and biogeochemical variables, but they have different ocean circulation and ocean biogeochemical components. This capability allows the CMIP5 models to participate in model evaluations of the same kind and scientific assesments on global climate projections (ex., Intergovernmental Panel for Climate Change, IPCC reports 2007 and 2013).

In this project I have analysed the output of 8 models, namely GFDL-ESM2G, GFDL-ESM2M, Hadley-GEM2, IPSL-CM5-LR, IPSL-CM5-MR, MPI-ESM-LR, MPI-ESM-MR and NOR-ESM2. These models are a combination of an atmospheric, land surface and ocean components that can be fully coupled to analyse present and future projections of ocean biogeochemistry and global climate feedbacks. The oceanic component of these models have a wide range of complexity, from basic NPDZs to more complex biogeochemical modules with dynamical stoichiometry, various Plankton Functional Types (PFTs) or full Iron cycle representation. A detailed model performance and model intercomparisons in terms of biogeochemical variables can be found in analysis from Bopp et al, 2013, and Steinacher et al, 2010, as well as Cocco et al., 2013 for the previous model generation.

To evaluate modeled oceanic O2 for present-day conditions, the ten year average over the 1995 to 2005 time period was computed for historical simulations as defined by the IPCC AR5 protocol. For this period, the models respond to applied forcings in terms of wind stress, radiation or temperature, and therefore with consequent changes on the ocean circulation fields, on global biogeochemical cycles and hence on the O2 concentration.

Data Visualization

Data analysis of the original NetCDF model files from the OCMIP5 repository has been done using NOAA's Ferret and NCO. Regridding of the different CMIP5 model grids into the ORCA2 grid was done using CDO scripts.

Data postprocessing and data conversion into JSON format were done with Python scripts, also for calculations of the Mikaloff Fletcher et al., 2006 region magnitudes.

Data visualization has been done using D3 (Data Driven Documents) javascript library developed by Mike Bostock. The spatial distribution of the oxygen concentration is shown on a Robinson projection using TopoJSON and GeoJSON from Natural Earth data.


This project has been developed at the Laboratoire de l'Environnement Marin at the Institut Universitaire Europeen de la Mer in Brest, France. It has been funded by the Universite de Bretagne Occidentale and the Laboratoire d'Excellence LabexMER. This project is supported by the SATT Ouest Valorisation consortium and Technopole Brest-Iroise.

Thanks to Patricia Cadule for all the discussions and suggestions.

For other data visualization and graphic design projects in science, please visit or follow datavizcowboy on twitter.

Oceanic Oxygen

by DatavizCowboy