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

Abstract In September of 2020, Arctic sea ice extent was the second-lowest on record. State of the art climate prediction uses Earth system models (ESMs), driven by systems of differential equations representing the laws of physics. Previously, these models have tended to underestimate Arctic sea ice loss. The issue is grave because accurate modeling is critical for economic, ecological, and geopolitical planning. We use machine learning techniques, including random forest regression and Gini importance, to show that the Energy Exascale Earth System Model (E3SM) relies too heavily on just one of the ten chosen climatological quantities to predict September sea ice averages. Furthermore, E3SM gives too much importance to six of those quantities when compared to observed data. Identifying the features that climate models incorrectly rely on should allow climatologists to improve prediction accuracy.

[1]  Laurence C. Smith,et al.  New Trans-Arctic shipping routes navigable by midcentury , 2013, Proceedings of the National Academy of Sciences.

[2]  G. G. Stokes.  "J." , 1890, The New Yale Book of Quotations.

[3]  M. Holland,et al.  Arctic sea ice decline: Faster than forecast , 2007 .

[4]  Gaël Varoquaux,et al.  Scikit-learn: Machine Learning in Python , 2011, J. Mach. Learn. Res..

[5]  Karl E. Taylor,et al.  An overview of CMIP5 and the experiment design , 2012 .

[6]  Tsuyoshi Murata,et al.  {m , 1934, ACML.

[7]  Veronika Eyring,et al.  Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6) experimental design and organization , 2015 .

[8]  J. Tukey,et al.  Variations of Box Plots , 1978 .

[9]  John F. B. Mitchell,et al.  THE WCRP CMIP3 Multimodel Dataset: A New Era in Climate Change Research , 2007 .

[10]  R. Treffeisen,et al.  September Arctic Sea Ice minimum prediction – a new skillful statistical approach , 2018 .

[11]  Wei Liu,et al.  Arctic sea-ice decline weakens the Atlantic Meridional Overturning Circulation , 2017 .

[12]  Karl Pfeiffer,et al.  Warm Arctic episodes linked with increased frequency of extreme winter weather in the United States , 2018, Nature Communications.

[13]  Lawrence O. Hall,et al.  A New Ensemble Diversity Measure Applied to Thinning Ensembles , 2003, Multiple Classifier Systems.

[14]  Ivana Cvijanovic,et al.  Future loss of Arctic sea-ice cover could drive a substantial decrease in California’s rainfall , 2017, Nature Communications.

[15]  Marvin N. Wright,et al.  Random forest as a generic framework for predictive modeling of spatial and spatio-temporal variables , 2018, PeerJ.

[16]  L. Breiman.  Arcing classifier (with discussion and a rejoinder by the author) , 1998 .

[17]  Wei-Yin Loh,et al.  Classification and regression trees , 2011, WIREs Data Mining Knowl. Discov..

[18]  Xiquan Dong,et al.  Evaluation and Intercomparison of Cloud Fraction and Radiative Fluxes in Recent Reanalyses over the Arctic Using BSRN Surface Observations , 2012 .

[19]  Julienne Stroeve,et al.  Changing state of Arctic sea ice across all seasons , 2018, Environmental Research Letters.

[20]  S. Geiger,et al.  Numerical simulations of seismicity-induced fluid flow in the Tjornes Fracture Zone, Iceland , 2011 .

[21]  Danna Zhou,et al.  d. , 1840, Microbial pathogenesis.

[22]  Tyler G. R. Reid,et al.  Arctic Sea Ice Extent Forecasting Using Support Vector Regression , 2014, 2014 13th International Conference on Machine Learning and Applications.

[23]  Gaël Varoquaux,et al.  The NumPy Array: A Structure for Efficient Numerical Computation , 2011, Computing in Science & Engineering.

[24]  M. Holland,et al.  Trends in Arctic sea ice extent from CMIP5, CMIP3 and observations , 2012 .

[25]  François Massonnet,et al.  Quantifying climate feedbacks in polar regions , 2018, Nature Communications.

[26]  Philip W. Jones,et al.  The DOE E3SM Coupled Model Version 1: Overview and Evaluation at Standard Resolution , 2019, Journal of Advances in Modeling Earth Systems.

[27]  Ron Kwok,et al.  Uncertainty in modeled Arctic sea ice volume , 2011 .

[28]  Lawrence O. Hall,et al.  A Comparison of Decision Tree Ensemble Creation Techniques , 2007, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[29]  L. Breiman.  Arcing Classifiers , 1998 .

[30]  C. Bitz,et al.  Processes Controlling Arctic and Antarctic Sea Ice Predictability in the Community Earth System Model , 2018, Journal of Climate.

[31]  Leo Breiman,et al.  Random Forests , 2001, Machine Learning.

[32]  E. DeWeaver,et al.  Persistence and Inherent Predictability of Arctic Sea Ice in a GCM Ensemble and Observations , 2011 .

[33]  K.,et al.  The Community Earth System Model (CESM) large ensemble project: a community resource for studying climate change in the presence of internal climate variability , 2015 .

[34]  Eric Rignot,et al.  Recent large increases in freshwater fluxes from Greenland into the North Atlantic , 2010 .

[35]  I. Eisenman,et al.  Sea Ice Trends in Climate Models Only Accurate in Runs with Biased Global Warming , 2016, 1606.08519.

[36]  M. H. Savoie,et al.  A long-term and reproducible passive microwave sea ice concentration data record for climate studies and monitoring , 2013 .

[37]  Leo Breiman,et al.  Bagging Predictors , 1996, Machine Learning.

[38]  Stefano Nembrini,et al.  The revival of the Gini importance? , 2018, Bioinform..