{"id":464,"date":"2025-12-29T19:08:11","date_gmt":"2025-12-29T19:08:11","guid":{"rendered":"https:\/\/atmos.ucla.edu\/dkondras\/?page_id=464"},"modified":"2026-02-06T02:02:23","modified_gmt":"2026-02-06T02:02:23","slug":"arctic-sea-ice-prediction","status":"publish","type":"page","link":"https:\/\/atmos.ucla.edu\/dkondras\/arctic-sea-ice-prediction\/","title":{"rendered":"Arctic sea ice prediction"},"content":{"rendered":"\n

Decline in the Arctic sea ice extent (SIE) has profound socio-economic implications and is a focus of active scientific research. Of particular interest is prediction of SIE on subseasonal time scales, i.e.~from early summer into fall, when sea ice coverage in Arctic reaches its minimum. However, subseasonal forecasting of SIE is very challenging due to the high variability of ocean and atmosphere over Arctic in summer, as well as shortness of observational data and inadequacies of the physics-based models to simulate sea-ice dynamics. The Sea Ice Outlook (SIO) <\/a>by Sea Ice Prediction Network (SIPN) <\/a>is a collaborative effort to facilitate subseasonal prediction of September SIE by physics-based and statistical models.<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

Data-adaptive Harmonic Decomposition (DAHD) and ML stochastic modeling techniques [Kondrashov et al. 2018] have been shown successful for retrospective and real-time summertime regional forecasting of Arctic Sea Ice extent. <\/p>\n\n\n\n

The real-time DAHD prediction of September SIE was fairly accurate and very competitive among statistical and physics-based models in 2016<\/a>, 2017<\/a>, 2018<\/a>, 2019,<\/a>  2020, <\/a>2021, <\/a> 2022,<\/a> 2023 <\/a>Sea Ice Outlook (SIO) submissions. The average of DAHD-based summertime Outlooks (June, July, August, September)  was within ~0.3 Mkm2 of the observed September pan-Arctic SIE for seven years in a row, given a total SIE area of ~5.0 Mkm2 and inter-quartile range spread of ~0.5 Mkm2:<\/p>\n\n\n\n

2016: 4.90 (predicted) vs 4.70 (observed)\u00a0 million km2<\/p>\n\n\n\n

                                          2017: 4.57 (predicted) vs 4.80 (observed)  million km2<\/p>\n\n\n\n

                                          2018: 4.53 (predicted) vs 4.71 (observed)  million km2<\/p>\n\n\n\n

                                          2019: 4.42 (predicted) vs 4.32 (observed)  million km2<\/p>\n\n\n\n

                                          2020: 4.40 (predicted) vs 3.92 (observed)  million km2<\/p>\n\n\n\n

                                          2021: 4.59 (predicted) vs 4.90 (observed)  million km2<\/p>\n\n\n\n

                                          2022: 4.80 (predicted) vs 4.87 (observed)  million km2<\/p>\n\n\n\n

                                          2023: 4.61 (predicted) vs 4.37 (observed)  million km2<\/p>\n\n\n\n

\u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 2024: 4.53\u00a0(predicted) vs. 4.39\u00a0(observed)\u00a0 million km2<\/p>\n\n\n\n

Also, DAHD-based predictions for the Alaska region solicited by SIO have been similarly accurate and are available in post-season reports. The key factors to this success are associated with DAHD\/ML capability to disentangle complex regional dynamics of Arctic Sea ice by data-adaptive harmonic spatio-temporal patterns. <\/p>\n\n\n\n

Bushuk et al. (2024)<\/a> performed assessment of the SIO multi-model predictive skill in retrospective fore- casts of regional summertime Arctic SIE. The results of DAHD-based reforecasts are consistent with the above-mentioned real-time SIO predictions over the past 7 years. The root-mean-square error (RMSE) skill of the DAHD model for the Pan-Arctic is close to the multi-model median of all models, including the dynamical ones, while its regional skill for the North Atlantic and the Siberian Seas is actually one of the best in RMSE among all models; see, for instance, Bushuk et al. (2024, Figs. 4 and 8)<\/a>.<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

2022 Outlook Contributions. <\/strong>The vertical black line is the observed value, and DAHD model forecast is marked by red box. Adapted from Fig. 9 at    https:\/\/www.arcus.org\/sipn\/sea-ice-outlook\/2022\/post-season<\/a><\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

2018 Outlook contributions<\/strong> by group for June (blue dot), July (green triangle), and August (orange diamond) are organized by general type of method; DAHD is marked by red box among statistical methods. The 2018 observed September SIE minimum is shown by dotted grey line, adapted from https:\/\/www.arcus.org\/sipn\/sea-ice-outlook\/2018\/post-season<\/a>.<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

Predictions of the Arctic SIE in the Sea Ice Outlook (SIO) for 2017<\/strong>; the red square marks DAHD prediction of September SIE. Contributions as box plots, broken down by type of method. Boxes show medians and interquartile ranges. Colors identify method types, and n denotes the number of contributions. Individual boxes for each method represent, from left to right, contributions to the June, July, and August SIO. The heavy gray line shows the 2017 observed September SIE from the NSIDC index, from https:\/\/www.arcus.org\/sipn\/sea-ice-outlook\/2017\/post-season<\/a>.<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

Predictions of the Arctic SIE in the Sea Ice Outlook (SIO) for 2016<\/strong>; the red square marks DAHD prediction of September SIE. Contributions as box plots, broken down by type of method. Boxes show medians and interquartile ranges. Colors identify method types, and n denotes the number of contributions. Individual boxes for each method represent, from left to right, contributions to the June, July, and August SIO. The heavy gray line shows the 2016 observed September SIE from the NSIDC index, from https:\/\/www.arcus.org\/sipn\/sea-ice-outlook\/2016\/post-season<\/a>.<\/p>\n\n\n\n

<\/p>\n\n\n\n

<\/p>\n\n\n\n

References<\/strong><\/p>\n\n\n\n

Dmitri Kondrashov,\u00a0Ivan Sudakow,\u00a0Valerie Livina,\u00a0Qingping Yang; Accurate and robust real-time prediction of September Arctic sea ice.\u00a0Cha<\/em><\/em>os, 2026; 36 (2): 023110.\u00a0https:\/\/doi.org\/10.1063\/5.0295634<\/a><\/p>\n\n\n\n

Kondrashov, D., M. D. Chekroun, and M. Ghil, 2018: 
Data-adaptive harmonic decomposition and prediction of Arctic sea ice extent, Dynamics and Statistics of the Climate System, 3(1), doi:10.1093\/climsys\/dzy001<\/a>.<\/p>\n\n\n\n

Kondrashov, D., M. D. Chekroun, X. Yuan, and M. Ghil, 2018: 
Data-adaptive Harmonic Decomposition and Stochastic Modeling of Arctic Sea Ice,in: Tsonis A. (eds) Advances in Nonlinear Geosciences. Springer, doi:10.1007\/978-3-319-58895-7_10<\/a>. Springer Nature 2019 Highlight in Earth Sciences (Book Chapters)<\/a><\/p>\n\n\n\n

Chekroun, M. D., and D. Kondrashov, 2017: Data-adaptive harmonic spectra and multilayer Stuart-Landau models,Chaos, 27, 093110: doi:10.1063\/1.4989400<\/a>, HAL preprint.<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"

Decline in the Arctic sea ice extent (SIE) has profound socio-economic implications and is a focus of active scientific research. Of particular interest is prediction of SIE on subseasonal time scales, i.e.~from early summer into fall, when sea ice coverage in Arctic reaches its minimum. However, subseasonal forecasting of SIE is very challenging due to the high…<\/p>\n","protected":false},"author":141,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"footnotes":""},"class_list":["post-464","page","type-page","status-publish","hentry"],"_links":{"self":[{"href":"https:\/\/atmos.ucla.edu\/dkondras\/wp-json\/wp\/v2\/pages\/464","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/atmos.ucla.edu\/dkondras\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/atmos.ucla.edu\/dkondras\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/atmos.ucla.edu\/dkondras\/wp-json\/wp\/v2\/users\/141"}],"replies":[{"embeddable":true,"href":"https:\/\/atmos.ucla.edu\/dkondras\/wp-json\/wp\/v2\/comments?post=464"}],"version-history":[{"count":6,"href":"https:\/\/atmos.ucla.edu\/dkondras\/wp-json\/wp\/v2\/pages\/464\/revisions"}],"predecessor-version":[{"id":482,"href":"https:\/\/atmos.ucla.edu\/dkondras\/wp-json\/wp\/v2\/pages\/464\/revisions\/482"}],"wp:attachment":[{"href":"https:\/\/atmos.ucla.edu\/dkondras\/wp-json\/wp\/v2\/media?parent=464"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}