Fully Automatic Trace Gas Plume Detection

TL;DR

This paper introduces an automated framework combining machine learning and physics-based analysis for detecting trace gas plumes in spectrometer data, improving detection of large and previously overlooked plumes.

Contribution

The authors develop a fully automated detection system that operates without human input, extending detection capabilities to new trace gases and retrospective analysis.

Findings

Automated system detects large plumes with negligible false positives.

Retrospective analysis finds at least 25% of plumes were previously missed.

Extended detection to NH3, NO2, and CO in EMIT data.

Abstract

Future imaging spectrometers will increase data volumes by orders of magnitude, requiring automated detection of trace gas point sources. We present a fully automated framework that combines machine learning-based morphological analysis with physics-based spectroscopic fitting to detect plumes without human participation. Applied to EMIT imaging spectrometer data, the system operates in two modes: "daily digest" that runs automatically on all downlinked data, flagging the largest events for immediate response, and a retrospective analysis that identifies plumes missed by prior human review. The daily digest demonstrates that a significant fraction of the largest plumes can be detected automatically with negligible false positives, while retrospective analysis suggests at least 25% of plumes may have been overlooked. In addition to the previously observed methane point sources, we extend detection to three understudied trace gases: NH3, NO2 and the first observations of carbon monoxide (CO) plume in EMIT imagery.