Multidecadal Trends in Instrumental SST and Coral Proxy Sr/Ca Records

Abstract

Historical ship observations of sea surface temperature (SST) from 1850 to present were used to compute linear 40-yr trends for all 5° latitude by 5° longitude grid cells with sufficient data. Trends from throughout the twentieth century were centered about a 7 mK yr?1 warming trend with standard deviation 14 mK yr?1. Although different with high statistical significance from a distribution with zero mean, the warming trends were not unusual in amplitude compared to the available nineteenth-century trends. Trends at the same grid points from the latter half of the nineteenth century were distributed about near-zero mean with standard deviation 17 mK yr?1. The shift in mean is robust to accounting for the biases arising from differing observational methods prior to 1942. The 40-yr trends from the latter half of the twentieth century were centered about 10 ± 4 mK yr?1 and more clearly distinct from earlier trends. Linear 40-yr trends were computed at different locations and times from all publicly available coral skeleton records of the concentration ratio of Sr to Ca. The pre-twentieth-century distribution of 40-yr trends in the Sr/Ca ratio was significantly different from the twentieth-century trends, consistent with the warming found in the instrumental SST. The interpretation of the coral Sr/Ca 40-yr trends cannot yet be reduced to a single factor. Major uncertainties were due to (i) the correlation of modern Sr/Ca records with instrumental SST being dominated by seasonal effects, with correlations on time scales longer than the annual cycle much lower, and (ii) the poor quality instrumental SST on long time scales in remote locations. Based on the NOAA extended reconstructed instrumental SST dataset since 1870 and 499 yr of Sr/Ca data from 13 different coral records, the authors found a Pearson correlation coefficient r = ?0.77 for 40-yr low-pass-filtered times series. Interpreting the change in distribution of trends in Sr/Ca will require further study of the factors affecting Sr/Ca on time scales longer than seasonal.