CryoSat-2 waveform classification for melt event monitoring

Measuring the mass balance of ice sheets is important with respect to understanding among others sea level rise, glacier dynamics, global ocean circulation and marine ecosystems. One parameter surface melt, which can be estimated from different satellite data sources. In this study we investigate potential utilizing machine learning techniques for CryoSat-2 (CS2) radar altimeter waveform classification in order derive melt information. Training derived by spatio-temporally matching CS2 measurements MODIS land temperature measurements. We propose a time convolution network fully connected classifier tail classifcation. addition non-deep model implemented, providing baseline. main challenges high class imbalance, as temperatures on interior Greenland rarely reach freezing point. The performance measured several metrics: F1 score, average recall Matthews correlation coefficient. results proof concept indicate feasibility.