Skip to content

Volcanic stratigraphy reconstruction using crystal size distribution automated by machine learning

Martin Jutzeler 1, Rebecca Carey1, Yasin Dagasan2, Ray Cas1,3

  • Affiliations: 1Centre for Ore Deposit and Earth Sciences (CODES), University of Tasmania, Australia 2Datarock Pty Ltd, Melbourne, Australia 3Monash University, Melbourne, Australia

  • Presentation type: Poster

  • Presentation time: Monday 16:30 - 18:30, Room Poster Hall

  • Poster Board Number: 144

  • Programme No: 3.1.36

  • Theme 3 > Session 1

Abstract

Volcanic stratigraphy reconstruction is traditionally based on qualitative facies analysis complemented by geochemical analyses. Here we present a new technique based on crystal size distribution to quantitatively fingerprint lavas, shallow intrusions and coarse lava breccias even where moderately altered, and/or complexly tectonized. We built an automated image analysis workflow using machine-learning for crystal segmentation, followed by statistical algorithms to compare and match the size distribution of feldspar and ferromagnesian phenocrysts. The workflow comprises three instance segmentation models for pre-processing the images, automated scale measurement and grain segmentation using Mask RCNN that provide an unbiased and quantitative approach to determine crystal size distribution. This novel technique avoids the laborious and time-consuming task of manual picking by image analysis. We successfully tested this method in the mineralized Cambrian Mt Read Volcanics in Tasmania, Australia. Multiple dacitic bodies could be correlated across several drillholes, and independently validated by bulk-rock chemical analyses of key samples. This volcanic stratigraphy method can be applied to a large variety of igneous rocks and is complementary to geochemical analyses and qualitative crystal content assessment.