The ubiquity of computing crosses all disciplines in the modern world, and with a domain as important as the cryosphere during the era of global climate change, integrating modern methods in computer science becomes paramount. As Earth observation data becomes increasingly available in unprecedented size and scope, it necessitates methodology which can efficiently process and extract value from Big Data. This thesis focuses on Light Detection and Ranging (LiDAR) data collected in the cryosphere, specifically over the Greenland Ice Sheet (GrIS). In order to leverage modern, high-resolution sensory technology my advisor Dr. Herzfeld has developed the Density Dimension Algorithm for Ice Surfaces (DDA-ice) to extract high-resolution ice-surface morphological characteristics from LiDAR data. This work improves the computational efficiency and scalability of such an algorithm, as to not compromise resolution for scale with the goal of analyzing the surface morphological characteristics of the GrIS over time in an efficient manner.

By integrating parallel computing techniques and distributed systems I can leverage the DDA- ice to process seasonal renderings of the ice-sheet morphology of the GrIS at a pace which keeps up with satellite data collection. In order to further highlight the morphological characteristics of the ice I focus on crevasses as manifestations of the underlying ice dynamics. A physically-informed multi-layer perceptron (MLP) is implemented to classify different crevasse types across the GrIS from LiDAR data by using the DDA-ice and relevant geostatistical methodology.