Temporal Multiscaling Behaviors of Sea Surface Temperature in Observations and CMIP5 Historical Simulations

Sea surface temperature (SST) is one of the most important parameters for the understanding of climate dynamics and climate change. While intense studies have focused on the low-frequency trends and components, only a few have approached the problem from the perspective of high-frequency components, e.g., the fluctuation pattern and long-range correlation. These scaling behaviors in SST series attract attentions of the community recently. However, the current studies generally focus on the whole time scale and the scaling behavior may vary at different time scales (i.e., multiscaling behavior). Such multiscaling behaviors need to be investigated in order to understand the mechanisms of SST variations and to improve the predictability of SST and relevant climatic phenomena.

The study investigates the long-range correlation and multiscaling behaviors of the global SST anomaly (SSTA) variations in observations (e.g., Met Office HadISST, Kaplan SST, NOAA/ERSST and COBE SST2) and historical simulations in Coupled Model Intercomparison Project Phase 5 (CMIP5). Detrended fluctuation analysis is employed in this study to detect the fractal scaling properties and long-range correlations of SSTA, since it can deal with not only stationary time series but also nonstationary time series with noise and trend. Monthly time series of SSTA in observational datasets and 26 historical simulations in CMIP5 are studied and compared.

Analysis results show that scaling behaviors of the SSTA in most ocean basins are separated into two distinct regimes by a crossover time scale of 2-6 years), indicating different scaling properties at different time scales. It is suggested that this crossover is modulated by the El Niño/La Niña–Southern Oscillation (ENSO). The scaling property of SSTA at small-scale (< crossover) is nonstationary and anti-persistent. It is, however, a stationary process and is long range correlated at large scale (> crossover), which is suggested to be modulated by oscillations like Pacific Decadal Oscillation (PDO) and Atlantic Multidecadal Oscillation (AMO). In addition, the scaling performance of the state-of-the-art global climate models in CMIP5 are also tested from the perspective of multiscaling property identified in the SSTA series, and plausible geographical dependence of scaling behaviors are also examined. These results provide a deeper understanding of the multiscaling behaviors of the SSTA series and may theoretically support climate prediction.