In Situ Data

A time-varying warm bias (systematically warmer temperature than the true value) has been found in the global XBT data, and a cold bias (systematically colder temperature than the true value) has been found in a small fraction of Argo float data (e.g., Gouretski and Koltermann, 2007; Wijffels et al., 2008; Ishii and Kimoto, 2009; Willis et al., 2009). XBT and MBT temperature observations are subject to instrument bias, such as depth bias. The depth of each temperature observation is calculated using a fall-rate equation and the time elapsed since the XBT entered the water. Inaccuracies in the fall rate affect the apparent depth at which the temperature profile is taken, which in turn causes a temperature bias that varies with depth. The MBT depth bias may have resulted from a delayed response by the diaphragm used to sense pressure and thus infer depth (Gouretski and Koltermann, 2007). The Argo biases were associated with a particular set of instruments deployed mainly in the Atlantic Ocean (Willis et al., 2009). The sensors on these instruments use pressure measurements to infer depth, but a flaw caused temperature and salinity values to be associated with incorrect pressure values, biasing the data.

Correcting for XBT depth bias reduced the magnitude of the interdecadal variability previously seen in the thermosteric sea-level signal during the 1970s (Domingues et al., 2008). An apparent sharp rise in thermosteric sea level during the 1970s was greatly decreased in the corrected data of Levitus et al. (2009), and essentially disappeared in the corrected data of Ishii and Kimoto (2009; compare the dotted and solid red and blue lines in Figure 3.1, top). Correcting for depth bias also changed the estimated rate of global thermosteric sea-level rise. For example, Ishii et al.’s (2006) original estimate of thermosteric sea-level rise for the upper 700 m was 0.26 ± 0.06 mm yr-1 from 1951 to 2005. After correcting XBT and MBT temperatures for depth bias and using an improved temperature climatology, Ishii and Kimoto (2009) found a slightly higher rate of 0.29 ± 0.06 mm yr-1 for the same time period (Table 3.1).

Discarding biased Argo profiles removed an apparent cooling trend from 2003 to 2006 (Willis et al., 2009). The linear trend from January 2005 to September 2011 in the newly analyzed data is 0.48 ± 0.15 mm yr-1 (Figure 3.2, Table 3.1).

Recent observational estimates of thermosteric sea-level rise have all been corrected for XBT and MBT depth bias (Table 3.1). The new estimates are based on updates of the Ishii and Kimoto (2009) data set (e.g., Ishii, personal communication; Kuo and Shum, personal communication), which corrects for depth bias, or the Ingleby and Huddleston (2007) data set (e.g., Domingues et al., 2008), which corrects for both XBT fall-rate bias and undersampling bias. Their estimates of the long-term thermosteric trend (beginning 1951–1961) in the upper 700 m of the ocean range from 0.29 ± 0.06 mm yr-1 to 0.52 ± 0.08 mm yr-1 (Table 3.1). The latter, by Domingues et al. (2008), is higher than the rates estimated by IPCC (2007) for the same period and by other investigators for similar periods. Estimates of the thermosteric trend in the upper ocean since 1993 range from 0.71 ± 0.31 mm yr-1 to 1.23 ± 0.30 mm yr-1 (Table 3.1). These rates are generally lower than those estimated by the IPCC (2007) for 1993 to 2003.

Observations for the deep ocean are sparse, so thermal expansion estimates for the full ocean depth are more uncertain than those for the upper ocean. The only recent estimates of the rate of thermosteric sea-level rise for the full ocean depth are by Domingues et al. (2008) and Church et al. (2011), who used a thermal expansion value of 0.2 ± 0.1 mm yr-1 and 0.17 mm yr-1, respectively, for the deep ocean. This deep-ocean value is comparable to a recent estimate of ~0.15 ± 0.08 mm yr-1 based on abyssal (below 4,000 m) and deep ocean (1,000–4,000 m) observations south of the SubAntarctic Front taken in the 1990s and 2000s (Purkey and Johnson, 2010). Kouketsu et al. (2011) estimated thermosteric sea-level change of ~0.11 mm yr-1 for the ocean below 3,000 m from the 1990s and to the 2000s based on observed data, and 0.12 mm yr-1 based on an ocean model data assimilation product. The IPCC (2007) assessment, based on work by Antonov et al. (2005), was 0.1 mm yr-1 from 700 m to 3,000 m (Bindoff et al., 2007). Given the scarcity of data, however, it is difficult to assess the uncertainty in deep ocean warming.

Domingues et al. (2008) estimated that thermosteric sea-level rise for the full ocean depth increased from 0.72 ± 0.13 mm yr-1 for 1961–2003 to 1.0 ± 0.4 mm yr-1 for 1993–2003 (Table 3.1). The Church et al. (2011) estimates for 1993–2008 are 0.88 ± 0.33 mm yr-1. For comparison, the committee calculated thermosteric sea-

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement