An investigation into the sampling bias of Argo profiling floats in the Southern Ocean

• Main Author: Tsei, Senam Kofi
• Other Authors: Backeberg, Björn
• Format: Master Thesis
• Language: English
• Published: Department of Oceanography 2015
• Subjects: Applied Marine Science; Southern Ocean
• Online Access: http://hdl.handle.net/11427/20017

Numerous Argo floats (approximately 3800 floats) have been deployed in the world's oceans to gather hydrographic and biogeochemical data from the upper 2000 m. However, limited research has been done on the spatial and temporal distribution and potential sampling bias of Argo profiling floats brought on by the effects of bathymetric steering of currents, as well as oceanic features, such as meanders and eddies, that affect their distribution over the global ocean. This study investigates the sampling distribution of profiling floats and assesses the mechanisms that impact their trajectories and distribution in the Atlantic sector of the Southern Ocean. The study reveals that Argo floats are influenced and steered towards frontal jets, which in turn are steered by the underlying bathymetry. Argo floats have a 30 % higher probability of sampling regions where depths range from 4000 - 5000 m, rather than shallow regions of the oceans. Using bootstrapping, this result was shown to be statistically significant at the 95 % confidence interval. The sampling bias is associated with floats becoming entrained into deep reaching frontal jets that occur in the Southern Ocean and dominate the deeper waters. This is shown by analyzing the Argo float positions in relation to mean geostrophic currents which shows that there is a 40 % higher probability (statistically significant at the 95 % confidence level) of finding Argo floats in regions where geostrophic currents range from 0.1 - 0.22 m.s-1 even though the majority of surface currents in the Southern Ocean are found below 0.05 m.s-1. This indicates a non-uniform distribution of Argo floats in the Southern Ocean, which leads to a spatial sampling bias in the float data. This has implications for how we characterize the oceanography or understand the distribution and variability of oceanographic processes and its relation to climate.