Machine learning feature analysis illuminates disparity between E3SM climate models and observed climate change

Distinct Role of a Spring Atmospheric Circulation Mode in the Arctic Sea Ice Decline in Summer

“…Zhang et al, 2008) (Table 2). In September 2020, the minimum SIE is the second lowest in the 42-year-old satellite record with a significant sea ice retreat over the eastern Arctic regions being observed (Nichol et al, 2021). It is demonstrated that the anomalous cyclonic circulation linked to the AO+ in spring 2020 tends to be the main reason related to the occurrence of this unprecedented case (Table 2).…”

“…In addition to the BBO, the springtime AO or AD index has also been previously reported as a trigger in leading to a dramatic SIE decline during summer over the PAS (Bi et al., 2019; Rigor et al., 2002), such as the earlier record low SIE minimum in September 2002 (Serreze, 2003) and another record low in September 2007 (J. Zhang et al., 2008) (Table 2). In September 2020, the minimum SIE is the second lowest in the 42‐year‐old satellite record with a significant sea ice retreat over the eastern Arctic regions being observed (Nichol et al., 2021). It is demonstrated that the anomalous cyclonic circulation linked to the AO+ in spring 2020 tends to be the main reason related to the occurrence of this unprecedented case (Table 2).…”

Statistical mechanics in climate emulation: Challenges and perspectives

“…Using nonparametric random forest regression and computing Gini feature importance measures, Nichol et al (2021) concluded that the energy exascale earth system model (E3SM) disproportionately relies on some of the climatological quantities when predicting September sea ice averages, which may explain why this model tends to underestimate Arctic sea ice loss. This approach outperformed their earlier contribution (Feng et al, 2015) based on wavelet analysis-support vector machines.…”

Comparison of Climate Model Large Ensembles With Observations in the Arctic Using Simple Neural Networks

“…Although the use of machine learning methods is still fairly new in climate science applications (Boukabara et al., 2021; Rasu et al., 2019), several studies have already demonstrated their utility over traditional multiple linear regression for identifying mechanistic processes and extracting patterns of climate change and variability (e.g., Barnes et al., 2020; Nichol, Peterson, Peterson, et al., 2021; Pasini et al., 2017). In this study, we use a form of deep learning called artificial neural networks (ANNs) for classifying Arctic maps of temperature data according to different GCMs.…”