(1983) andHallock and Teague (1992),theXBTprobetypesinvestigated in this study are characterized by depth underestimation
in the lower part of the temperature profile. Within the upper
50–200 m, the XBTs tend to overestimate the sample depth at least
for a considerable number of years. The depth overestimation may be
explained by the constant near-surface depth offset inherent to the
XBT temperature profiles.
Second, application of corrections derived from the comparison of the GEBCO bathymetry to the global XBT dataset does not
lead to the elimination of the total temperature bias in the global
XBT dataset as illustrated by Fig. 7