Publications

Lv, D; Cheng, Y; Xiao, H; Hai, G (2022). A Framework for Fracture Extraction Under Glaciological Property-Based Constraints: Scientific Application on the Filchner-Ronne Ice Shelf of Antarctica. EARTH AND SPACE SCIENCE, 9(8), e2022EA002293.

Abstract
Fractures are an important indicator of ice shelf instability and predecessors of calving events. It is of great importance to identify fractures and understand the formation and propagation processes of fractures. However, automated methods of fracture identification also erroneously extract non-fracture features, which are mixed with the automatically extracted fractures, negatively impacting the accuracy of the results. In this study, a framework for fracture extraction with glaciological property-based constraints was proposed to extract fractures and filter out non-fracture features. The extraction module consists of image preprocessing and linear feature extraction. The constraints were based on glaciological property analysis, including constraints based on physical properties of fractures and structural properties of ice shelves. The framework was applied to Mosaic of Antarctica and Landsat 8 Operational Land Imager images to extract fractures on the Filchner-Ronne Ice Shelf. The results suggest that most of the non-fracture features were successfully filtered out, accounting for about half of all linear features. Upon comparison, it was found that the proposed method detected similar to 97% of the rifts and similar to 42% of the crevasses detected by manual extraction. The spatial pattern and formation mechanisms of the fractures were analyzed based on strain maps. The formation of crevasses is closely related to shear margin and longitudinal tension, while the large rifts near the ice shelf front are likely to be related to the strikes of lateral ice flow tributaries. The constraint strategy can be further improved by utilizing more glaciological properties to obtain more-accurate fracture extraction. Plain Language Summary Ice shelf fracture is of great importance because it leads to ice shelf calving. Ice shelf calving will then accelerate ice loss in Antarctica and sea level rise. However, masses of non-fracture features are mixed into automatic fracture extraction results. To deal with this problem, an extraction framework with constraints was proposed to filter out non-fracture features. The constraints are based on the different patterns found in fractures and non-fractures. The framework was applied to two imagery sets of the Filchner-Ronne Ice Shelf. The results suggest that most of the non-fracture features were successfully filtered out, which account for about half of the extraction results. The automatic extraction identified similar to 97% of rifts and similar to 42% of the crevasses of the manual extraction results. Based on the analysis of the fracture map, crevasse formation is mostly caused by shear margin and longitudinal tension, while the large rifts near the ice shelf front may be caused by strikes of lateral ice flow tributaries. The constraint strategy can be further improved by utilizing more glaciological properties to more accurately extract features.

DOI:
10.1029/2022EA002293

ISSN:
2333-5084