Unsupervised machine learning detection of iceberg populations within sea ice from dual-polarisation SAR imagery

Accurate quantification of iceberg populations is essential to inform estimates of Southern Ocean freshwater and heat balances as well as shipping hazards. The automated operational monitoring of icebergs remains challenging, largely due to a lack of generality in existing approaches. Previous efforts to map icebergs have often exploited synthetic aperture radar (SAR) data but the majority are designed for open water situations, require significant operator input, and are susceptible to the substantial spatial and temporal variability in backscatter that characterises SAR time-series. We propose an adaptive unsupervised classification procedure based on Sentinel 1 SAR data and a recursive Dirichlet Process implementation of Bayesian Gaussian Mixture Model. The approach is robust to inter-scene variability and can identify icebergs even within complex environments containing mixtures of open water, sea ice and icebergs of various sizes. For the study area in the Amundsen Sea Embayment, close to the calving front of Thwaites Glacier, our classifier achieved a mean pixel-wise F1 score against manual iceberg delineations from the SAR scenes of 0.960 & PLUSMN; 0.018 with a corresponding object-level F1 score of 0.729 & PLUSMN; 0.086. The method provides an excellent basis for estimation of total near-shore iceberg populations and has inherent potential for scalability that other approaches lack.

Automated summary

Accurate quantification of iceberg populations is crucial for estimating Southern Ocean freshwater and heat balances, as well as shipping hazards. Existing automated monitoring methods lack generality. This study proposes an adaptive unsupervised classification approach based on SAR data, which can robustly identify icebergs in complex environments. The method achieved good results in the evaluation of the study area.