Effects of instrumentation changes on sea surface temperature measured in situ

Abstract

Measurements of sea surface temperature (SST) are an important climate record, complementing terrestrial air temperature observations, records of marine air temperature, ocean subsurface temperature, and ocean heat content. SST has been measured since the 18th century, although observations are sparse in the early period. Historically, marine observing systems relied on observations made by seafarers and necessary information on measurement methods is often not available. There are many historical descriptions of observing practice and instrumentation, some including quantification of biases between different methods. This documentation has been used, with the available observations, to develop models for the expected biases, which vary according to how the measurements were made, over time and with the environmental conditions. Adjustments have been developed for these biases and some gridded SST datasets adjust for these differences and provide uncertainty estimates, including uncertainties in the bias adjustments. The modern in situ SST-observing system continues to evolve and now includes many observations from moored and drifting buoys, which must be characterized relative to earlier observations to provide a consistent record of multi-decadal changes in SST. 

The reliability of global and hemispheric surface temperature records

“…A number of current projects are seeking to digitize much of the British logbook material available in archives. The potential size and importance of SST data, also requires enhancements to our knowledge of how SST and MAT measurements were taken in the past (Kent et al, 2010;Kennedy, 2014). More SST data are not only important for improving the reliability of hemispheric and global temperature series, but can help to improve infilled SST fields, which are vital for extended reanalyses.…”

“…The use of canvas buckets continued on most merchant and naval vessels up to about 1940. Bucket use continued after this, but designs were improved (see Kent et al, 2010).…”

“…When SST data were first examined in detail by climatologists in the 1970s (see references discussed by Folland and Parker, 1995;Kent et al, 2010;and Kennedy, 2014), it was soon realized that the method of measurement might influence the results. Between the wars there were a number of comparisons made of different measurement techniques on research vessels and on cruises, i.e., comparisons of different types of bucket, as well as with thermometers fitted into the engine cooling-water intake pipes of ships (see Kent et al, 2010;and Kennedy, 2014 for details).…”

“…Between the wars there were a number of comparisons made of different measurement techniques on research vessels and on cruises, i.e., comparisons of different types of bucket, as well as with thermometers fitted into the engine cooling-water intake pipes of ships (see Kent et al, 2010;and Kennedy, 2014 for details). More comparisons were undertaken in the 1960s and 1970s, and it was at this time that an extra code was added into both logbooks and transmitted data to indicate how the measurement was made (Woodruff et al, 2011).…”

“…These issues are being resolved with improved metadata and by attempting to relate individual measurements to the ships that took them (see discussion in Kennedy, 2014), but for about 30% of the SST observations during the period 1950-75, the measurement method is unknown. As stated earlier, buckets continued to be used after 1945, but designs were also much improved to minimize the bias (see Kent et al, 2010). Second, since the late 1970s, there have been major changes to the marine observing system, with the principal aim of improving weather forecasts and seasonal climate predictions.…”