24 Oct 2023
 | 24 Oct 2023

MMSeaIce: Multi-task Mapping of Sea Ice Parameters from AI4Arctic Sea Ice Challenge Dataset

Xinwei Chen, Muhammed Patel, Fernando Pena Cantu, Jinman Park, Javier Noa Turnes, Linlin Xu, K. Andrea Scott, and David A. Clausi

Abstract. The AutoIce challenge, organized by multiple national and international agencies, seeks to advance the development of near-real-time sea ice products with improved spatial resolution, broader spatial and temporal coverage, and enhanced consistency. In this paper, we present a detailed description of our solutions and experimental results for the challenge. We have implemented an automated sea ice mapping pipeline based on a multi-task U-Net architecture, capable of predicting sea ice concentration (SIC), stage of development (SOD), and floe size (FLOE) using Sentinel-1 SAR data. For model training and evaluation, we utilize the AI4Arctic dataset, which includes SAR imagery, corresponding passive microwave and auxiliary data, and ice chart-derived label maps. Among the submissions from over 30 teams worldwide, our team achieved the highest combined score of 86.3 %, as well as the highest scores on SIC (92.0 %) and SOD (88.6 %). Additionally, our result analysis showcases the effectiveness of various techniques, such as input SAR variable downscaling, spatial-temporal encoding, input feature selection, and loss function selection, in significantly improving the accuracy, efficiency, and robustness of deep learning-based sea ice mapping.

Xinwei Chen et al.

Status: final response (author comments only)

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse
  • RC1: 'Review of egusphere-2023-1297', Anonymous Referee #1, 07 Nov 2023
  • RC2: 'Comment on egusphere-2023-1297', Karl Kortum, 20 Nov 2023
  • RC3: 'Comment on egusphere-2023-1297', Andreas Stokholm, 05 Dec 2023

Xinwei Chen et al.

Xinwei Chen et al.


Total article views: 221 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
154 54 13 221 8 5
  • HTML: 154
  • PDF: 54
  • XML: 13
  • Total: 221
  • BibTeX: 8
  • EndNote: 5
Views and downloads (calculated since 24 Oct 2023)
Cumulative views and downloads (calculated since 24 Oct 2023)

Viewed (geographical distribution)

Total article views: 208 (including HTML, PDF, and XML) Thereof 208 with geography defined and 0 with unknown origin.
Country # Views %
  • 1
Latest update: 07 Dec 2023
Short summary
This paper introduces an automated sea ice mapping pipeline utilizing a multi-task U-Net architecture. It attained the top score of 86.3 % in the AutoIce challenge. Ablation studies revealed that incorporating brightness temperature data and spatial-temporal information significantly enhanced model accuracy. Accurate sea ice mapping is vital for comprehending the Arctic environment and its global climate effects, underscoring the potential of deep learning.