Calibration of sea ice drift forecasts using random forest algorithms

Cyril Palerme,Malte Müller

doi:10.5194/tc-15-3989-2021

Abstract

Abstract. Developing accurate sea ice drift forecasts is essential to support the decision-making of maritime end-users operating in the Arctic. In this study, two calibration methods have been developed for improving 10 d sea ice drift forecasts from an operational sea ice prediction system (TOPAZ4). The methods are based on random forest models (supervised machine learning) which were trained using target variables either from drifting buoy or synthetic-aperture radar (SAR) observations. Depending on the calibration method, the mean absolute error is reduced, on average, between 3.3 % and 8.0 % for the direction and between 2.5 % and 7.1 % for the speed of sea ice drift. Overall, the algorithms trained with buoy observations have the best performances when the forecasts are evaluated using drifting buoys as reference. However, there is a large spatial variability in these results, and the models trained with buoy observations have particularly poor performances for predicting the speed of sea ice drift near the Greenland and Russian coastlines compared to the models trained with SAR observations.

Highlights

Passive microwave observations of sea ice concentration have been available for more than 40 years and have shown negative trends in Arctic sea ice extent since the beginning of the satellite era (e.g., Cavalieri and Parkinson, 2012; Comiso et al, 2017), with strong trends during the summer (e.g., Comiso et al, 2017)
The performances of sea ice drift forecasts have been evaluated using buoy observations, and the syntheticaperture radar (SAR) observations have been used to study the spatial variability in the forecast performances
For predicting the direction of sea ice drift, the models trained with buoy observations significantly outperform the TOPAZ4 prediction system and the models trained with SAR observations for all lead times, except 10 d

Summary

Introduction

Passive microwave observations of sea ice concentration have been available for more than 40 years and have shown negative trends in Arctic sea ice extent since the beginning of the satellite era (e.g., Cavalieri and Parkinson, 2012; Comiso et al, 2017), with strong trends during the summer (e.g., Comiso et al, 2017). Long-term negative trends in sea ice thickness have been assessed by comparing retrievals from satellite altimeters (ICESat and CryoSat-2). In order to ensure maritime safety, it is essential that accurate sea ice information is delivered to marine end-users. In addition to sea ice charts, short-term sea ice forecasts are necessary for planning activities and providing up-to-date information to end-users. The spatial resolution of the current sea ice models is often too coarse compared to user needs

Methods

Results

Conclusion