Identifying and correcting the World War II warm anomaly in sea surface temperature measurements

Abstract:

The most recent sea surface temperature (SST) estimates from both the U.S. (ERSST) and U.K. (HadSST) indicate anomalously warm conditions during WWII that greatly exceed what statistical and dynamical models can reproduce given current knowledge of climatic forcing and internal variability. A long-standing hypothesis relates this warm anomaly to rapid changes in measurement instruments and protocols during WWII, but specific biases have not been identified nor corrected. Using a recently developed linear-mixed-effect methodology, we quantify and trace the WWII anomaly to warm offsets in specific U.S. and U.K. Naval measurements and anomalously warm night-time bucket SSTs during WWII. We remove identified SST biases at both global and regional scales. Corrected estimates show no significant WWII anomaly and a much more homogeneous SST evolution across the middle 20th century. Our correction also leads to a tighter error estimate, with the uncertainty of WWII SSTs decreasing to only one-fourth of that in the most up-to-date observational estimates from the U.K. Met. Office. Furthermore, our correction brings global-mean SSTs into consistency with internal variability in state-of-art model simulations and reconciles the largest existing model-data discrepancy in the global-mean surface temperature. These results are highly statistically significant, robust to a wide range of alternate analyzing scenarios, and supported by historical documents and dynamical considerations. Our work also buttresses the community's current knowledge of the forcing history and internal variability and provides confidence to climate projections made by climate models.