Seasonal Arctic sea ice forecasting with probabilistic deep learning

Tom R Andersson,Jeremy Wilkinson,Steffen Tietsche,Yevgeny Aksenov,James Byrne,Chris Russell,Stephen Law,Tony Phillips,J Scott Hosking,Rod Downie,Daniel C Jones,María Pérez-Ortiz,Andrew Elliott,Emily Shuckburgh,Eduardo Blanchard-Wrigglesworth,Brooks Paige,Beena Balan Sarojini

doi:10.1038/s41467-021-25257-4

Abstract

Anthropogenic warming has led to an unprecedented year-round reduction in Arctic sea ice extent. This has far-reaching consequences for indigenous and local communities, polar ecosystems, and global climate, motivating the need for accurate seasonal sea ice forecasts. While physics-based dynamical models can successfully forecast sea ice concentration several weeks ahead, they struggle to outperform simple statistical benchmarks at longer lead times. We present a probabilistic, deep learning sea ice forecasting system, IceNet. The system has been trained on climate simulations and observational data to forecast the next 6 months of monthly-averaged sea ice concentration maps. We show that IceNet advances the range of accurate sea ice forecasts, outperforming a state-of-the-art dynamical model in seasonal forecasts of summer sea ice, particularly for extreme sea ice events. This step-change in sea ice forecasting ability brings us closer to conservation tools that mitigate risks associated with rapid sea ice loss.

Highlights

Anthropogenic warming has led to an unprecedented year-round reduction in Arctic sea ice extent
Satellite observations of sea ice are presented as images: passive microwave measurements of microwave brightness temperature are converted to sea ice concentration (SIC) estimates of the fractional area covered by sea ice in a given grid cell, ranging between 0 and 100%
We have introduced an Arctic sea ice forecasting artificial intelligence (AI) system, IceNet, which outperforms the leading dynamical model, SEAS5, in seasonal predictions of Arctic sea ice

Summary

Introduction

Anthropogenic warming has led to an unprecedented year-round reduction in Arctic sea ice extent This has far-reaching consequences for indigenous and local communities, polar ecosystems, and global climate, motivating the need for accurate seasonal sea ice forecasts. We show that IceNet advances the range of accurate sea ice forecasts, outperforming a state-of-the-art dynamical model in seasonal forecasts of summer sea ice, for extreme sea ice events. Rising temperatures have played a key role in reducing Arctic sea ice, with September sea ice extent around half that of 1979 when satellite measurements of the Arctic began[4] This downward trend will continue, even in optimistic greenhouse gas emission reduction scenarios[5]. While there are inherent sea ice predictability limits, owing mostly to chaotic processes in the atmosphere[18,19,20], studies have demonstrated that potential predictability is higher, suggesting that forecasts could be improved[17,21,22]

Methods

Results

Conclusion