Machine Learning Based Atmospheric Methane Concentration Estimation and Plume Complex Detection and Segmentation Using EMIT Hyperspectral Data

Date of Award

5-9-2026

Degree Name

M.S. in Electrical Engineering

Department

Department of Electrical and Computer Engineering

Advisor/Chair

Russell Hardie

Abstract

The role of hyperspectral imagers in greenhouse-gas detection is rapidly advancing, supported by the deployment of new sensors and the escalating demand for accurate climate monitoring. In 2022, NASA’s Jet Propulsion Laboratory (JPL) mounted the Earth Surface Mineral Dust Source Investigation (EMIT) hyperspectral imaging sensor on the International Space Station. In addition to the sensor’s first mission to study the effects on mineral dust suspended in the air, the instrument began detecting and characterizing greenhouse gases a month after deployment. JPL released Level 2 methane data products by applying physics-based matched-filter techniques on the Level 1 hyperspectral radiance data. These methane products include estimates of column methane concentrations and spatial delineations of identified plume complexes. These data products afford an opportunity to train and test machine learning methods to replace and/or enhance the physics-based methods and manual effort required to segment methane plumes. This paper focuses on the application of regression convolution neural networks and a semantic segmentation network for estimation of methane plume concentrations in addition to detection and segmentation of methane plumes using EMIT hyperspectral radiance data and available methane data products provided by JPL.

Keywords

Artificial Intelligence, Climate Change, Electrical Engineering, Engineering, Remote Sensing

Comments

OCLC No. 1591829874

Rights Statement

Copyright 2026, author.

Link to Full Text

Share

COinS
 
 
 

Links