To learn more about a novel statistical method that improves the Argo data error detection accuracy, check out this paper:

Improved Statistical Method for Quality Control of Hydrographic Observations.

Read the full publication


How to better detect data errors for a realistic ocean state prediction? Thanks to a novel statistical method that improves the error detection accuracy, a French and American team of scientists from OceanScope, Met Office, WHOI, LOCEAN and LOPS answers. Once the reference historical dataset manually quality checked, this approach estimates local min/max statistics from the reference dataset. A local validity interval is then derived from them, rather than from local mean and variance as with the classical method. So, it can be used for outlier detection with any other independent observation. The method accuracy is evaluated through statistics of good and bad detections, allowing a severe reduction of the percentage of erroneous detections. This “MinMax” method is eventually compared with the classical one.