Injection Risk Triage (IRT) Field Manual

A causal-inference pipeline for induced-seismicity management in the Permian Basin

Author
Affiliation

Lewis Matthews

Project Geminae

Published

June 2, 2026

Welcome

This is the field manual for Injection Risk Triage (IRT) — a causal-inference pipeline that answers, with regulatory-grade evidence standards, the question every saltwater-disposal permit decision in the Permian Basin implicitly turns on:

If this well injects less, does the earthquake risk actually go down — and by how much?

The manual is the translation layer between the technical work (a targeted-learning analysis of 7,581 TexNet earthquakes and 918,720 well-days of Railroad Commission injection records) and the three audiences who can act on it:

  • Regulators — who today curtail injection after felt events, using spatial-proximity rules, and who could instead condition permits on causal dose-response evidence before events occur.
  • Operators — who need to know which of their wells carry attributable risk, what a volume reduction buys, and how to demonstrate proactive management to regulators and insurers.
  • Researchers — who need the methods to be auditable, the failure modes documented, and the estimates reproducible.

How to read this book

If you have five minutes, read the Executive Summary. If you advise a regulator or legislator, the Regulator One-Pager is written to be lifted directly into a briefing packet. If you operate disposal wells, start with the Decision Tables. If you intend to attack the methods — please do — start with the Evidence Scoreboard and the Methods Defense appendix, which document not only the estimates but the artifacts we found in our own pipeline and how we fixed them.

The live system

This manual describes a running system, not a one-time study: