Enhancing Geological Mapping: Evaluating the Benefits and Drawbacks of Remote Sensing in Field Applications
Author(s)
Viengkham, Elysia N.
Advisor(s)
Editor(s)
Collections
Supplementary to:
Permanent Link
Abstract
Technological advancements are making geological mapping safer, more efficient, and increasingly autonomous. Drones are becoming increasingly available and are now capable of capturing ultra-high-resolution imagery at less than 3 centimeters per pixel resolution across multiple wavelengths. High-performance computing and GIS tools offer rapidly produced and highly detailed elevation models and derived data products. These tools aid in the mapping process and are only limited by the ‘human’ ability to analyze and classify these derived data products. Machine learning (ML) enhances this process by helping to cluster and classify complex multi-dimensional datasets, which enables a detailed analysis of geological features. While these tools complement rather than replace the expertise of geologists, they significantly enhance mapping capabilities, particularly in remote or hazardous environments, and may eventually support fully autonomous mapping on planetary bodies, such as Mars (e.g., NASA’s Ingenuity helicopter). Our objective was to develop a protocol for semi-autonomous mapping using ML techniques. This semi-autonomous ML mapping provides a powerful approach to synthesizing large datasets and enables predictions of surficial geological map units directly from imagery and elevation data. These ML-generated predictions offer geologists a valuable starting point in unmapped areas and alternative perspectives that can reveal biases or gaps in existing maps. This study explores the potential and limitations of using ML to produce surficial geological maps from high-resolution drone surveys, emphasizing its role in advancing modern geological mapping techniques.
Sponsor
Date
2025-04-30
Extent
Resource Type
Text
Resource Subtype
Thesis