Skip to content

brendonhall/core_features

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

8 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Classifying Core Images from the Chixulub Impact Crater

Crater rendition, core section Source: DETLEV VAN RAVENSWAAY/SCIENCE SOURCE/IODP EXP 364

This repo contains notebooks that implement the hypercolumn 'Deep Feature' core image classification scheme presented in my SciPy 2020 talk.

IODP/ICDP Expedition 364 extracted ~830m of core from the peak ring of the Chixulub impact structure. This core provides dramatic evidence of how the crater was formed, and the processes that affected the resulting environmental changes. A section of core just above the K-Pg boundary consists of impactites with many different rock fragments and melt. This repo contains Python notebooks that use machine learning to classify these different rock fragments, or clasts.

The technique uses sparse labels supplied by a geoscientist. It requires a couple examples for each different class. Ilastik was used to annotate clasts in the image for training.

This example uses core line scan image dataset made available by:

IODP Expedition 364 Scientists (2017): Line scan images of IODP Hole 364-M0077A. PANGAEA, https://doi.org/10.1594/PANGAEA.881718

which is licensed CC-BY 3.0.

References

  • Morgan et al. (2016) The formation of peak rings in large impact craters, Science 354(6314).
  • Gulick et al. (2019) The first day of the Cenozoic, PNAS 116 (39).
  • Ormö et al. (2007) Sedimentological analysis of resurge deposits at the Lockne and Tvären craters: Clues to flow dynamics, Met. & Pl. Sc. 42(11).
  • Rae et al. (2019). Impact‐induced porosity and microfracturing at the Chicxulub impact structure. JGR: Planets.

Media:

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published