気候：20世紀の海水温を再考する

Ocean temperatures in the early twentieth century were warmer than previously thought, according to new evidence presented in Nature.

Estimates of global temperatures are typically calculated using a combination of land surface air and ocean surface temperatures. However, problems can arise when using data from the first half of the 1900s because of improvements in measurement technologies in the years since, which call into question how we should interpret these values.

Sebastian Sippel and colleagues developed a method to reconstruct global temperatures using only land surface air or ocean surface temperature values. They found that in most years, both sets resulted in similar estimates of global temperature change, with the notable exception of those in 1900–1930. In this period, the estimates based on the ocean figures are on average 0.26 °C colder than the values generated by land data, a trend which is unsupported by other lines of evidence. The authors argue that this discrepancy could be due to an uncorrected bias in sea surface temperature data from this time, which stems from the use of buckets to collect water samples in the early twentieth century. The transition to other methods in the latter part of the century coincides with increased agreement in the temperature estimates. The authors find that this unresolved cold bias because of the collection method contributed to the lack of agreement between the data sources for the period 1900–1930.

The authors posit that correcting this bias would result in a more modest warming trend in the early twentieth century, although it is noted that this would not affect estimates of total global warming since the mid-nineteenth century. These results provide better agreement between simulated and observed temperature changes and highlight the importance of revisiting historical data to understand global temperature change.