The World Ocean Database 2005 as of May 2009 is used to estimate temperature and sample depth biases of expendable (XBT) and mechanical (MBT) bathythermographs by comparing bathythermograph temperature profiles with more accurate bottle and conductivity/temperature/depth (CTD) data. It is shown that the application of depth corrections estimated earlier from side-by-side XBT/CTD inter-comparisons, without accounting for a pure thermal bias, leads to even larger disagreement with the CTD and bottle reference temperatures. Our calculations give evidence for a depth-variable XBT fall-rate correction with the manufacturer-derived depth being underestimated in the upper 200. m and overestimated below this depth. These results are in agreement with side-by-side inter-comparisons and direct fall-rate estimates. Correcting XBT sample depths by a multiplicative factor which is constant with depth does not allow an effective elimination of the total temperature bias throughout the whole water column. The analysis further suggests a dependence of the fall rate on the water temperature which was reported earlier in the literature. Comparison among different correction schemes implies a significant impact of systematic biases on the estimates of the global ocean heat content anomaly. © 2010 Elsevier Ltd.

cited By 63