Early warning of future rapid Arctic sea ice loss

Projektinformationen

ArcticWATCH

ID Finanzhilfevereinbarung: 101040858

DOI

10.3030/101040858

EK-Unterschriftsdatum 9 September 2022

Startdatum 1 Januar 2023

Enddatum 31 Dezember 2027

Finanziert unter

• European Research Council (ERC)

Gesamtkosten

€ 1 456 652,50

EU-Beitrag

€ 1 456 652,50

Koordiniert durch

UNIVERSITE CATHOLIQUE DE LOUVAIN
Belgium

Periodic Reporting for period 1 - ArcticWATCH (Early warning of future rapid Arctic sea ice loss)

Berichtszeitraum: 2023-01-01 bis 2025-06-30

Zusammenfassung vom Kontext und den Gesamtzielen des Projekts

The Arctic is currently transitioning toward a new climatic state that will be characterized by seasonally sea-ice-free conditions almost every year from the 2050s, with widespread ecological, climatic, and societal consequences. There is growing evidence that the future summer sea ice retreat will not occur at a constant rate. Indeed, climate model simulations are suggestive of pronounced sub-decadal fluctuations on top of the long-term trend, leading to periods of relative stability followed by abrupt sea ice decline in hardly 3-5 years. A lot remains to be understood regarding the precursors, mechanisms, predictability, and impacts of these rapid events. In particular, it is unclear how close we might be to the next one.

The overall objective of this project, ArcticWATCH, is to build an integrated early warning system that alerts on the possibility of rapid Arctic sea ice loss for the following summer up to five years. This system will provide annually updated assessments and will synthesize multiple lines of evidence harvested from various data sources (pre-existing and generated during the project), including climate model projections, initialized climate model and machine-learning-based predictions, satellite observations, and climate reconstructions. By introducing innovative targeted numerical experiments, ArcticWATCH will also identify the new pathways of sea ice predictability in a warmer world and will thereby provide evidence-based guidance regarding the design of the Arctic observing system for the next 30 years. Finally, ArcticWATCH will make a leap forward in depicting environmental impacts during and after rapid sea ice loss events, from short (Arctic heatwaves and precipitation extremes) to long (interactions with the Arctic and North Atlantic oceanic circulation) timescales.

The hypothesis that, after a decade of relatively stable conditions, Arctic sea ice is poised to an abrupt decline before 2030, will be paid utmost attention.

Arbeit, die ab Beginn des Projekts bis zum Ende des durch den Bericht erfassten Berichtszeitraums geleistet wurde, und die wichtigsten bis dahin erzielten Ergebnisse

The ArcticWATCH project has already contributed significant progress in understanding Arctic sea ice variability and predicting its future changes. Our research is structured around four key areas: rapid ice loss events, drivers of sea ice decline, data assimilation for improved reconstructions, and novel data-driven prediction methods.

1. Understanding Rapid Ice Loss Events (RILEs): Using climate models from CMIP6, including five large ensembles, we analyzed the frequency, magnitude, and seasonality of Rapid Ice Loss Events (RILEs). Our results show that RILEs can occur year-round, with winter/spring events being less frequent but longer-lasting, while summer/autumn events occur more often but for shorter durations. Notably, our projections indicate a 60% probability that at least one summer RILE will occur before 2030 in September. A preliminary sea ice mass balance analysis highlights the dominant role of thermodynamic processes in these events, particularly reduced ice growth and increased melt.

2. Drivers of Arctic Sea Ice Decline: We examined the causes behind the rapid decline of Arctic sea ice in the early 2000s and the relative stabilization since the 2010s. Our findings indicate that these fluctuations fall within the range of internal climate variability, with the rapid decline being statistically more unusual than the recent stabilization. Using initialized climate model simulations, we assessed the predictability of these extreme trends but found no significant improvement over uninitialized runs, suggesting that large model biases limit predictive skill.

3. Improved Arctic Sea Ice Reconstructions: We developed a 25-member ensemble simulation reconstructing Arctic sea ice conditions from 1979 to 2023 by assimilating satellite sea ice concentration observations into the ocean-sea ice model NEMO-SI3. This approach significantly improved the model's representation of seasonal sea ice variations, reducing biases in summer melt and annual ice volume estimates. This reconstruction will serve as a foundation for future ensemble predictions and the dataset will soon be publicly available.

4. Data-Driven Predictions and Methodological Advances: We pioneered two novel data-driven models—one based on transfer operators and another on neural networks—to provide probabilistic seasonal-to-interannual forecasts of Arctic sea ice extent. These models outperform simple persistence forecasts and demonstrate skillful predictions up to four years ahead. Additionally, a new statistical physics-inspired method, the rare event algorithm, was employed to simulate extreme sea ice states, revealing key atmospheric and oceanic conditions leading to extreme reductions.

Fortschritte, die über den aktuellen Stand der Technik hinausgehen und voraussichtliche potenzielle Auswirkungen (einschließlich der bis dato erzielten sozioökonomischen Auswirkungen und weiter gefassten gesellschaftlichen Auswirkungen des Projekts)

The project has delivered significant advances beyond the state-of-the-art in two aspects:

• The identification and exhaustive description of RILEs in CMIP6 models represents a significant advancement beyond the state-of-the-art, as no previous study has conducted a multi-model analysis combining large ensembles and a multi-model ensemble. Thanks to this analysis, we are now able to state that there is a 60% probability that at least one summer RILE will occur before 2030 in September.

• Using uninitialized fully-coupled climate simulations from the most recent generation of the Coupled Model Intercomparison Project (CMIP6), we found that the accelerated decline in sea ice during the early 2000s and the relatively stable conditions that followed were not unusual when accounting for accounting for the internal variability of the climate system. In fact, the accelerated decline in Arctic sea ice at the turn of the century is found to be more unusual than the sea ice pause we have observed since, such that we cannot really talk of a sea ice pause.

Probabilistic Prediction of Arctic Sea Ice Extent for 2024-2028. Credit L. Hoffman.