Our new paper, titled "Machine Learning-Based Classification of Lake Ice and Open Water from Sentinel-3 SAR Altimetry Waveforms," is now out in the prestigious Remote Sensing of Environment journal. The article is available open-access. Check out the article for some insight into what researchers at H2O Geomatics are working on!
Comparison of classification accuracies with a change in hyperparameter values for a) SVM, b) KNN, c) RF and d) GBT. The flat surface observed in the figures indicates the less sensitivity nature of the algorithms to their hyperparameters.
The study introduces a novel approach using machine learning algorithms and satellite radar altimetry data to predict lake surface conditions (open water, young ice, growing ice, or melting ice). It's worth noting that this is the pioneering study to employ Sentinel-3 SAR altimetry for lake ice classification, relying exclusively on altimetry data for the classification process.