Machine learning and computational image analysis leveraged to probe relationships between carbonate isotopic composition, diagenetic alteration, and sedimentary facies isotopic variability
Abstract
Carbon isotope chemostratigraphy regularly is utilized to construct temporal frameworks during intervals of Earth history with limited geochronologic and/or biostratigraphic constraints, such as the early Cambrian Period. This approach requires that carbon isotope values are modulated by a globally synchronous signal, yet every year new research finds additional reasons to doubt the notion that these records should be correlative (e.g., facies isotopic variability, diagenetic alteration, stratigraphic incompleteness, basin isolation, and more). Here, we leverage four high-resolution early Cambrian records from the Anti-Atlas Margin, Morocco to probe relationships between carbonate rock composition, diagenetic alteration, and sedimentary facies isotopic variability. Each of these records has bed-level, field-based facies assignments, carbon isotopic records, and estimated sediment accumulation rates, which we utilize to constrain the statistical likelihood that these carbon isotopic records are substantially modified by diagenetic alteration or sedimentary facies isotopic variability. However, any given rock classification (e.g., grainstone) has a high degree of variability: for example, size and type of skeletal and non-skeletal elements, and relative abundance of mud versus cement. This variability has proven difficult to rigorous assess and quantify, especially with hand samples, yet it may impact isotopic records. In an attempt to address this classification variability, we employ an ultra-high-resolution multispectral imaging technique. By extracting a set of quantitative image metrics from the image suite for each sample (which we term a facies vector), we probe and measure the variability within facies classifications, and investigate their relationships with geochemical data, including carbon isotopic values. Using these facies vectors, we re-visit questions regarding the statistical likelihood that diagenetic alteration and facies isotopic variability severely impacted these carbon isotopic records. We investigate whether detailed classification variability, as captured by the ultra-high-resolution multispectral imaging suites, can account for isotopic variability, indicate the likelihood of diagenesis, or further differentiate sedimentary facies.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2022
- Bibcode:
- 2022AGUFMPP46A..08H