Present-day erosion of the north polar scarps on Mars from automated detection and analysis of active block falls
In this thesis the evolution of the north polar scarps on Mars was studied with respect to the erosion rate caused by ice block falls. An automated method was developed to identify new blocks between individual High Resolution Imaging Science Experiment (HiRISE) images acquired at different times. T...
| Main Author: | |
|---|---|
| Other Authors: | , , |
| Format: | Doctoral or Postdoctoral Thesis |
| Language: | English |
| Published: |
2022
|
| Subjects: | |
| Online Access: | https://depositonce.tu-berlin.de/handle/11303/18228 https://doi.org/10.14279/depositonce-17021 |
| Summary: | In this thesis the evolution of the north polar scarps on Mars was studied with respect to the erosion rate caused by ice block falls. An automated method was developed to identify new blocks between individual High Resolution Imaging Science Experiment (HiRISE) images acquired at different times. This method was then applied to HiRISE images spanning three Martian years to derive the first observation-based erosion rate estimation for such scarps. Existing automated detection methods developed for Mars had focused either on large surface objects such as impact craters or on rocks at candidate landing sites, where the background environment is flat and constant over time. Developing an automated change detection method for block falls along the steep scarps of the north polar ice cap, however, presented a new challenge due to the small sizes of the ice blocks and the complex and dynamic background of the region. An automated method was developed based on machine learning and blob detection. Its performance was assessed through comparison with manual identification of new blocks. The method correctly identified 75.1% of the new blocks with a false detection rate of 8.5%, proving a robust and trust-worthy automated method. The developed method was used to identify new blocks and estimate their sizes and volumes. The results enabled estimating the retreat rate to be ~ 0.2 m/kyr, a rate which is similar to that of the ice accumulation on top of the scarp. This retreat rate is too low to counteract the viscous flow rate suggested on the basis of theoretical modelling studies, a process with a competing effect to erosion. The approach in this dissertation highlighted the benefits of inter-disciplinary research, provided novel insights into long-standing questions and can, thus, become a stepping stone in the application of machine learning in satellite image analysis with a focus on geo-morphological questions in planetary research. Im Rahmen dieser Doktorarbeit wird die gegenwärtige Entwicklung von Steilabbrüchen ... |
|---|