A Profile Classification Model from North-Atlantic Argo temperature data

A quantitative understanding of the integrated ocean heat content depends on our ability to determine how heat is distributed in the ocean and what are the associated coherent patterns. This dataset contains the results of the Maze et al., 2017 (Prog. Oce.) study demonstrating how this can be achieved using unsupervised classification of Argo temperature profiles.

The dataset contains:

  • A netcdf file with classification~results (labels and probabilities) and coordinates (lat/lon/time) of 100,684 Argo temperature profiles in North Atlantic.
  • A netcdf file with a Profile Classification Model (PCM) that can be used to classify new temperature profiles from observations or numerical models.

The classification method used is a Gaussian Mixture Model that decomposes the Probability Density Function of the dataset into a weighted sum of Gaussian modes. North Atlantic Argo temperature profiles between 0 and 1400m depth were interpolated onto a regular 5m grid, then compressed using Principal Component Analysis and finally classified using a Gaussian Mixture Model.

To use the netcdf PCM file to classify new data, you can checkout our PCM Matlab and Python toolbox here: https://github.com/obidam/pcm


Physical oceanography


heat content, classification, North Atlantic, stratification, water mass, argo, profile, pattern


70N, 0S, 0E, -80W


Alternative access to data

Profile Classification Model
56 KoNetCDFProcessed data
Classification results
16 MoNetCDFProcessed data
How to cite
Maze Guillaume (2017). A Profile Classification Model from North-Atlantic Argo temperature data. SEANOE. https://doi.org/10.17882/47106

Copy this text