Implementation

Coverage

• Coverage and Coverage Sum are two indicators used to assess the performance of Argo float observations.

• Coverage Sum represents the average monthly coverage across all boxes. It is computed by taking the mean of the monthly coverage values for each individual box.

• Coverage is also the average monthly coverage across all boxes but with an additional constraint: it is limited by the coverage target for each box. For example, if there are 4 observations in a box in a given month, but the target is 3, the Coverage for that box will only consider 3 observations, not the full 4.

• The difference between Coverage and Coverage Sum can be interpreted as the potential for optimizing the array while maintaining the same level of activity.

• Additionally, note that Coverage and Covervage Sum only includes data that is considered good for temperature and salinity (T/S) measurements.

The following paragraph is for the future implementation of Coverage

There are two coverage. The monthly coverage and the annual coverage:

• Annual coverage is the mean of each monthly coverage.
• Monthly coverage are computed every month. Monthly coverage are always computed in two steps:

Step 1: Compute the number of observations per design element (Procedure update_gis.argo_design33_obs_month() )

Step 2: Compute coverage metrics.

We compute 3 different coverage metrics:

• Coverage (“Well-sampled” elements): Only count elements where the number of observations per month/(3*target) >=1/total number of elements. Note that this coverage is the continuity of the "coverage" indicator before evolution of the coverage metrics in 2026.
• Coverage, capped (Summed metric):
  For each element, compute the number of observations per month/(3*target), capped at a maximum of 1. Then sum these values across all elements/total number of elements. Note that this coverage is the continuity of the "coverage, sum" indicator before the evolution of the coverage metrics in 2026.
• Coverage potential: For each element, compute the number of observations per month/(3*target), not capped at a maximum of 1. Then sum these values across all elements/total number of elements. This coverage provide a indication or the potential performance of Argo is the floats where perfectly distributed.

Data flow

Quality

Every month, the index file is checked and A quality profil are counted. The ratio between A quality profile and all profile is displayed in the quality KPI for all variables.

We need to check if the computation is re-done regularly (e.g. once a year) to take into account improvement in the overall data quality of the data base.