Temperature Variability on Coral Reefs Versus Gridded SST – The Long and the Short of It

Coral-based temperature reconstructions and gridded sea-surface-temperature (gSST) data sets both provide valuable insights into tropical climate variability. However, coral records often exhibit greater interannual to decadal variability than is observed in gSST products or Earth System Models (ESMs). This discrepancy is often attributed to large differences in spatial scale: coral records reflect conditions over areas of only a few square centimeters, while gSST and ESM grid cells span 1 to 10,000 km². In situ temperature loggers on coral reefs allow us to isolate the effects of spatial scale from other non-climatic influences on coral temperature records. Many logger studies focus on hourly to monthly timescales, temperature biases, and whether gSST can capture temperature extremes associated with coral bleaching and mortality; however, paleoclimate reconstructions provide an understanding of variability on longer timescales. Here, we compare the power spectral density and coherence of logger temperature and gSST on daily to decadal timescales using logger data from 42 sites on the Great Barrier Reef. We find that temperature variations recorded by loggers on reefs are well correlated with and have the same amplitude as gSST variations at decadal to annual timescales. Therefore, the excess decadal variability commonly seen in coral-based temperature reconstructions cannot be attributed to a general effect of spatial scale.