Predictability of Arctic Annual Minimum Sea Ice Patterns

Abstract

Abstract Prediction of the arctic annual sea ice minimum extent and melting patterns draws interest from numerous industries and government agencies but has been an ongoing challenge for forecasters and climate scientists using statistical and dynamical models. Using the dominant independent modes of interannual sea ice concentration (SIC) variability during September–October, a new approach combining statistical analysis with physically derived links to natural climate variability sources is used to predict each mode and the total anomaly pattern. Sea ice patterns associated with each mode are predominantly shaped by the wind-driven advective convergence, forced by circulation anomalies associated with local and remote forms of naturally occurring climate variability. The impacts of the Arctic Oscillation, beginning from the preceding winter, control the leading mode of SIC variability during the annual minimum. In the three final months of the melting period, the broad impacts of the Indian and East Asian summer monsoons produce unique SIC impacts along the arctic periphery, displayed as the second and third modes, respectively. El Niño–Southern Oscillation (ENSO) largely shapes the fourth SIC mode patterns through influencing variability early in the melting period. Using physically meaningful and statistically significant predictors, physical–empirical (P–E) models are developed for each SIC mode. Some predictors directly account for the circulation patterns driving anomalous sea ice, while the monsoon-related predictors convey early season sources of monsoonal variability, which subsequently influences the Arctic. The combined SIC predictions of the P–E models exhibit great skill in matching the observed magnitude and temporal variability along the arctic margins during the annual minimum.