SEEM SEEM SEEM
  • Home
  • Research
  • People
  • Publications
  • Projects
  • Blog
  • Join / Contact

On this page

  • About
  • Why it matters
  • Key capabilities and audience
  • Public links
  • References

GeoSIRR

Conversational geological cross-section modeling

Geological modeling
Large language models
Research software
Research context and public links for GeoSIRR
Published

March 10, 2026

About

GeoSIRR—Geological Section Interpretation, Reconstruction & Refinement—is a Python application that translates geological narratives into a domain-specific language and coordinate-based cross-sections (Anikiev et al. 2026). Users can generate, validate, visualize, refine, and query a section through a command-line conversation.

Why it matters

Turning an informal geological description into consistent geometry is slow and requires both domain and drafting expertise. GeoSIRR makes the intermediate representation explicit: the generated geometry can be inspected, validated, revised, and reproduced rather than remaining hidden in an image-generation system.

Key capabilities and audience

GeoSIRR combines completeness checks, format and topology validation, Matplotlib visualization, iterative refinement, and section-aware question answering. It is designed for geologists, educators, subsurface modelers, and researchers evaluating large language models in scientific workflows.

Public links

  • GitHub repository
  • Discussion paper DOI
  • Version 1.0.0 on Zenodo
Note

This public SEEM-KFUPM repository is a fork of the maintained upstream project and is licensed under GPL-3.0.

References

Anikiev, Denis, Juan Esteban Mosquera, Korhan Ayranci, Judith Bott, and Umair Bin Waheed. 2026. “GeoSIRR 1.0: Conversational Geological Cross-Section Modeling Using Large Language Models.” EGUsphere, ahead of print. https://doi.org/10.5194/egusphere-2025-6545.

© 2026 Smart Earth Exploration and Monitoring Group

College of Petroleum Engineering & Geosciences

College of Petroleum Engineering & Geosciences · KFUPM

  • Content: CC BY 4.0

  • Powered by Quarto