C. Heuzé, F. Vivier, J. Le Sommer, J.-M. Molines and T. Penduff (2015), Can we map the interannual variability of the whole upper Southern Ocean with the current database of hydrographic observations?, Journal of Geophysical Research Oceans, vol 120, pp 7960-7978, doi:10.1002/2015JC011115 

This is an extension of the work done during my Master thesis. To study the interannual variability of hydrographic properties, the first action is often to bring scattered observations onto a regular space-time grid. Yet there is no consensus in the literature regarding which method to use for this remapping, and authors do not really justify their choice of method.

We compared the biases induced solely by the remapping for the methods used in the literature, focusing on the Southern Ocean mixed layer depth. We used outputs from the 1/12° DRAKKAR simulation as our reference, that we colocated onto the location and date of actual hydrographic observations before remapping them.

For each method, performances are assessed using Taylor diagrams: the closer to the black star, the better the method. Percentages indicate the fraction of the Southern Ocean area for which each method returns values. After Fig. 5 of Heuzé et al. (2015).

We found that there is no best method, but rather compromises between map coverage and accuracy:

• an objective analysis with a low search radius induces little bias, but leaves large areas of the Southern Ocean empty;
• an objective analysis with a large seach radius induces the largest biases (up to 250 m in winter) since it smoothes any extreme value, but fills most of the Southern Ocean at each time step;
• no method could reproduce the trends correctly without also creating fake trend.

Download the full-text by clicking here.