Performance of Ice Sensing Algorithms for Argo Floats in the Southern Ocean

Abstract This study evaluated the performance of the various ice sensing algorithms (ISAs) used for Argo floats in the Southern Ocean with the hydrographic profiles observed by Argo and marine mammals and the sea ice concentration (SIC) measured by satellite. From our assessment, the ISAs with their original settings do not successfully prevent the float from surfacing in ice-covered regions. For example, Klatt et al.’s ISA with its original setting of −1.79°C for the critical temperature (T crit ) for decision-making, which was first introduced to profiling floats and enhanced their survivability in winter significantly, fails to predict sea ice with a probability of 38% and allows a float to ascend to the icy surface with an average SIC of up to about 80%. Therefore, we suggest that T crit be changed to −1.5°C “by default” for Argo floats for the global monitoring array in the seasonally ice-covered open ocean in the Southern Ocean. This setting can reduce the failure rate for sea ice prediction to less than 10% and the average SIC where the float is allowed to ascend to about 25% or lower. The reduction of ice collision risk extends the average lifetime of floats and, ultimately, enhances the cost-effectiveness of Argo array in the Southern Ocean. A warmer setting of T crit > −1.5°C, however, increases the local risk that the ISA may not allow a float to ascend to the sea surface throughout the year, especially in the Weddell and Ross Seas.