Mapping Greenland's subglacial topography is critical for projecting the future mass loss of the ice sheet and its contribution to global sea-level rise. However, the complex and sparse nature of observational data, particularly information about the bed topography under the ice sheet, significantly increases the uncertainty in model projections. Bed topography is traditionally measured by airborne ice-penetrating radars that measure the ice thickness directly underneath the aircraft, leaving data gaps of tens of kilometers in between flight lines. This study introduces a deep learning framework, DeepTopoNet, that integrates radar-derived ice thickness observations and BedMachine Greenland data through a novel dynamic loss-balancing mechanism. Among all efforts to reconstruct bed topography, BedMachine has emerged as one of the most widely used datasets, combining mass conservation principles and ice thickness measurements to generate high-resolution bed elevation estimates. The proposed loss function adaptively adjusts the weighting between radar and BedMachine's bed, ensuring robustness in areas with limited radar coverage while leveraging the high spatial resolution of BedMachine's bed estimates. Our approach incorporates gradient-based and trend surface features to enhance model performance and utilizes a convolutional neural network (CNN) architecture (i.e., BedTopoCNN) designed for subgrid-scale predictions. By systematically testing on the Upernavik Isstrøm) region in West Greenland, the model achieves high accuracy (MAE: 12.49 m, RMSE: 19.38 m, and R2: 0.99), outperforming baseline methods in reconstructing subglacial terrain. This work demonstrates the potential of deep learning in bridging observational gaps, providing a scalable and efficient solution to inferring subglacial topography. This framework paves the way for improved predictions of ice sheet flow and sea level rise.

There is a growing need for empirical benchmarks that support researchers and practitioners in selecting the best machine learning technique for given prediction tasks. In this article, we consider the next event prediction task in business process predictive monitoring, and we extend our previously published benchmark by studying the impact on the performance of different encoding windows and of using ensemble schemes. The choice of whether to use ensembles and which scheme to use often depends on the type of data and classification task. While there is a general understanding that ensembles perform well in predictive monitoring of business processes, next event prediction is a task for which no other benchmarks involving ensembles are available. The proposed benchmark helps researchers to select a high-performing individual classifier or ensemble scheme given the variability at the case level of the event log under consideration. Experimental results show that choosing an optimal number of events for feature encoding is challenging, resulting in the need to consider each event log individually when selecting an optimal value. Ensemble schemes improve the performance of low-performing classifiers in this task, such as SVM, whereas high-performing classifiers, such as tree-based classifiers, are not better off when ensemble schemes are considered.