Times are displayed in (UTC-05:00) Central Time (US & Canada) Change

Size, Shape, Abundance, and Age? Characterizing Detrital Zircon Grain Morphology Using Machine Learning to Identify Young Grains: A Case Study from the Blackhawk Formation, Utah, USA

Session: Advances and Applications in Geochronology for Interpreting Stratigraphic and Basin Records (Posters)

Presenting Author:

Olivia Wachob

Authors:

Wachob, Olivia M.¹, Stockli, Daniel F.², Malkowski, Matthew³, Sylvester, Zoltan⁴, Peak, Barra Augusta⁵

(1) Dept. of Earth and Planetary Sciences, University of Texas at Austin, Austin, TX, USA, (2) Dept. of Earth and Planetary Sciences, University of Texas at Austin, Austin, TX, USA, (3) Dept. of Earth and Planetary Sciences, University of Texas at Austin, Austin, TX, USA, (4) Dept. of Earth and Planetary Sciences, University of Texas Austin, Austin, TX, USA; Bureau of Economic Geology, University of Texas at Austin, Austin, TX, USA, (5) Dept. of Earth and Planetary Sciences, University of Texas at Austin, Austin, TX, USA,

Abstract:

Morphometric characterizations of detrital zircon (DZ) provide opportunities for enhanced provenance and maximum depositional age (MDA) interpretations beyond what is possible with geochronology alone. Grain measurements, such as size and shape, can aid in distinguishing sedimentary sources for provenance studies. These measurements can also help identify the youngest age component for MDA analyses, often ‘sharply faceted’, acicular grains. Yet, grain morphology remains underutilized in DZ studies due, in part, to limited quantitative understanding of a systematic relationship between grain size, shape, and age. Obtaining these morphometric measurements can be time consuming as well, particularly in analyses with hundreds or even one thousand grains. Our study aims to determine if and how size, shape, and DZ abundance trends correlate with U-Pb age distributions between different, age-equivalent lithofacies within the same basin. For example, we seek to evaluate how low-abundance populations of small or euhedral/acicular grains will affect the recovery of the youngest U-Pb date modes and MDA interpretations. We provide a proof-of-concept workflow to efficiently extract DZ morphometric characteristics using segmenteveryzircon, a machine-learning based semantic segmentation Python module. The module, an extension of segmenteverygrain, is fine-tuned to semi-automatically identify thousands of individual DZ grains for downstream morphometric and geochronologic analyses. This workflow utilizes legacy sandstone samples in addition to new mudstone samples from the Book Cliffs of Utah, USA, with emphasis on the Cretaceous Blackhawk Formation that is characterized by complex first-cycle and recycled Cordilleran zircon populations. The legacy sandstone samples were re-analyzed to obtain a more comprehensive zircon U-Pb dataset. U-Pb dates from sandstone samples were compared to dates acquired from new stratigraphically equivalent mudstone samples within the Blackhawk Formation. All dated sample mounts were imaged using reflected-light microscopy and morphometrically analyzed using our segmentation procedure for rapid identification of relationships between age and grain shape parameters. Preliminary findings suggest an offset in zircon grain size distribution between stratigraphically equivalent facies. Future work aims to continue model development for rapid identification of acicular, sharply faceted grains, and further understanding of volumetric changes between facies using x-ray computed tomography.