Subseasonal forecast of Arctic sea ice concentration via statistical approaches

Lei Wang,Xiaojun Yuan,Cuihua Li

doi:10.1007/s00382-018-4426-6

Lei Wang, Xiaojun Yuan + Show 1 more

Open Access

https://doi.org/10.1007/s00382-018-4426-6

Copy DOI

Abstract

Subseasonal forecast of Arctic sea ice has received less attention than the seasonal counterpart, as prediction skill of dynamical models generally exhibits a significant drop in the extended range (> 2 weeks). The predictability of pan-Arctic sea ice concentration is evaluated by statistical models using weekly time series for the first time. Two statistical models, the vector auto-regressive model and the vector Markov model, are evaluated for predicting the 1979–2014 weekly Arctic sea ice concentration (SIC) anomalies at the subseasonal time scale, using combined information from the sea ice, atmosphere and ocean. The vector auto-regressive model is slightly inferior to the vector Markov model for the subseasonal forecast of Arctic SIC, as the latter captures more effectively the subseasonal transition of the underlying dynamics. The cross-validated forecast skill of the vector Markov model is found to be superior to both the anomaly persistence and damped anomaly persistence at lead times > 3 weeks. Surface air and ocean temperatures can be included to further improve the forecast skill for lead times > 4 weeks. The long-term trends in SIC due to global warming and its polar amplification contribute significantly to the subseasonal sea ice predictability in summer and fall. The vector Markov model shows much higher skill than the NCEP CFSv2 model for lead times of 3–6 weeks, as evaluated for the period of 1999–2010.

Highlights

The trans-Arctic shipping routes are projected to be more and more navigable as the Arctic sea ice shrinks due to global warming and Arctic amplification
The vector autoregressive (VAR) model could avoid accumulating errors as the vector Markov model might do in the iterations, but might insufficiently represent the subseasonal transition of the underlying dynamics for long lead times
The pan-Arctic mean root-meansquare error (RMSE) of the vector Markov model is still quite skillful compared with the climatology and anomaly persistence in most seasons (Fig. 10e–h), with advantage reduces only slightly compared with the take-1-year-out cross-validation experiments

Summary

Introduction

The trans-Arctic shipping routes are projected to be more and more navigable as the Arctic sea ice shrinks due to global warming and Arctic amplification (i.e., the Arctic has shown much more warming than lower latitudes in the past few decades, e.g., Smith and Stephenson 2013). The longer opening season in the Arctic will stimulate more social activities such as transpolar shipping, tourism, and natural resources (e.g., O’Garra 2017; Petrick et al 2017) All these economic activities call for better understanding of the Arctic sea ice predictability, especially over subseasonal time scales. Subseasonal variability of Arctic sea ice concentration (SIC) is usually predicted using two types of models: dynamical (or numerical) and statistical models. Arctic SIC time series show large variability over marginal seas, in the Barents Sea (Fig. 1a; note that the total variability includes contributions from longterm trends and variability over all smaller time scales), and a visible reduction in variability can be seen after long-term linear trends are removed (Fig. 1b).

Data and method

Take‐1‐year‐out cross‐validated forecast skill

Retrospective forecast skill

Comparison with a dynamical model

Summary and discussion