摘要:On global and hemispheric scales, sea-surface temperature (SST) anomalies are assumed to be good surrogates for near-surface marine air temperature (MAT) anomalies. In fact, global gridded temperature datasets commonly blend SST and near-surface air temperature anomalies to overcome the lack of geographically homogeneous and reliable MAT observations. Here, we show that SST and MAT anomalies differ regarding crucial statistical properties such as multiannual trends and probabilistic distributions of daily and monthly averages. We provide evidence of the lack of interchangeability from an array of moored buoys in the tropical Pacific Ocean. We identify statistically significant discrepancies between SST and MAT anomalies for single as well as groups of such buoys. Thus, caution is required when characterizing and interpreting MAT variability through SST observations, especially at shorter than decadal timescale.