A new study in the journal Nature shows that the oceans were less cold in the early 20th century (1900-1930) than previously thought. During this period the ocean appears too cold due to the way some measurements were taken. This makes global ocean surface temperature measurements during this period inconsistent with both land air temperatures and palaeoclimatic data and the differences between land and ocean are larger than shown in climate models.
This discovery has far-reaching implications for our understanding of past climate variability and future climate change. However, lead author and junior professor Dr Sebastian Sippel from Leipzig University stresses that the new findings do not affect the quantification of the global warming relative to 1850-1900 and the human contribution to that warming: the land and ocean temperatures of the 19th century (1850-1900), before the onset of the cold period, provide a physically very consistent picture of temperature changes up to the present day. Nevertheless, correcting this cold period could increase confidence in the amount of observed warming, changing what we know about historical climate variability and improve the quality of future climate models.
Understanding global temperature trends is crucial for climate research. Dr Sebastian Sippel, Junior Professor for Climate Attribution at Leipzig University, worked with international scientists to reconstruct the global mean temperature from historical climate data like a jigsaw puzzle -- including historical land and ocean measurements and palaeoclimatic analyses. When comparing land and ocean, Sippel noticed a systematic deviation: at the beginning of the 20th century, ocean temperatures were lower than in previous decades, while over land air temperatures remained relatively constant. This outcome is not consistent with physical theory and climate models.
New explanations for past phenomena
Using many different lines of evidence, the new study shows that reconstructions of the global mean temperature from ocean surface data for this period are too cold: on average about 0.26 degrees Celsius colder than seen in land-based reconstructions. This discrepancy is greater than what would be possible under natural climate variability. "Our latest findings do not change the long-term warming since 1850. However, we can now better understand historical climate change and climate variability," says junior professor Dr Sebastian Sippel. For example, the reasons for the early 20th century warming period between 1900 and 1950 have never been fully understood. If the ocean temperatures are corrected, the warming trend of the early 20th century is weaker. "The discrepancies between the climate models and the observed temperature trend at the beginning of the 20th century are mainly due to an incomplete understanding of the observations, rather than incomplete climate models or natural climate variability. There are well-established approaches to account for the effects of changing measurement methods on ocean surface temperature measurements. The new research shows that in the early 20th century these methods don't properly account for very rapidly changing differences in the way the observations were made. Our new understanding confirms the climate models and shows even more clearly the human impact since pre-industrial times," says co-author Professor Reto Knutti, Professor for Climate Physics at ETH Zurich.
A multidimensional approach
The study itself offers indications that the cause of the ocean cold anomaly could lie in insufficiently documented information about the measurement techniques used at that time. Before the Second World War, ocean temperatures were mainly measured with buckets on ships, but the method of measurement and the composition of ship fleets changed from decade to decade, making it much more difficult to correct for systematic measurement errors. The authors of the study therefore recommend a variety of approaches to data processing and analysis: "Our methodological approach emphasises the need to continuously rescue and digitise historical climate data and compare it with independent data. At the same time, very different assumptions regarding systematic adjustments of early climate data should be tested, as the observational data are of central importance as a basis for climate understanding and modelling," says Sippel.
