The enhanced warming of the Arctic compared to global mean warming is termed Arctic Amplification (AA). According to observations, AA is as large as 4.2 during 1980-2022 (Arctic defined as poleward of 70^oN). State-of-the-art climate models robustly simulate AA but seldom replicate the observed magnitude. The discrepancy between simulated and observed AA has raised concerns that models may not correctly represent the Arctic’s response to increased greenhouse gas forcings. Sweeney et al. (2023) recently reconciled this discrepancy in the annual mean AA by applying an innovative machine learning approach to remove internal variability in observed Arctic and global mean trends, obtaining an observational externally forced AA of 3.0. Removing internal variability’s influence enables a direct comparison of forced AA between model simulations and observations, aiding in identifying model bias. AA has large seasonality, and so the simulated and observed AA discrepancy also strongly depends on season. For instance, the observed monthly AA in April is 5.1 for 1980-2022, while the multimodel mean AA for this month is only 2.5. This raises critical questions such as: 1) What is the role of internal variability in observed AA seasonality? 2) How is the model simulated AA compared with observations after removing the internal variability effects? To address these questions, we use climate model data to train a machine learning algorithm that uses the monthly multi-decadal surface air temperature (SAT) and pressure trend pattern maps as inputs to determine the influence of internal variability on monthly Arctic and global-mean SAT trends. We also use the machine learning approach to separate the internally generated and externally forced trends in sea ice extent (SIE). Forced trends in both SAT and SIE from this study would help better understand climate feedback processes, identify model biases, and constrain simulated forced trends in the Arctic. This will increase our confidence for predictions of Arctic climate change, including an ice-free Arctic, and advance our understanding of broader climate implications.

Sweeney, A., Fu, Q., Po-Chedley, S., Wang, H., Wang, M., 2023: Internal Variability Increased Arctic Amplification during 1980-2022. Geophysical Research Letters, 50, e2023GL106060 doi:10.1029/2023GL106060.