Enabling fast greenhouse gas emissions inference  from satellites with GATES: a Graph-Neural-Network  Atmospheric Transport Emulation System

Fillola, Elena; Santos-Rodriguez, Raul; Tunnicliffe, Rachel; Clark, Jeffrey N.; Keshtmand, Nawid; Ganesan, Anita; Rigby, Matthew

doi:10.5194/gmd-19-1893-2026

Articles | Volume 19, issue 5

https://doi.org/10.5194/gmd-19-1893-2026

© Author(s) 2026. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/gmd-19-1893-2026

© Author(s) 2026. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 19, issue 5

Development and technical paper

|

05 Mar 2026

Development and technical paper |

| 05 Mar 2026

Enabling fast greenhouse gas emissions inference from satellites with GATES: a Graph-Neural-Network Atmospheric Transport Emulation System

Elena Fillola, Raul Santos-Rodriguez, Rachel Tunnicliffe, Jeffrey N. Clark, Nawid Keshtmand, Anita Ganesan, and Matthew Rigby

Data sets

LPDM footprint dataset - Brazil Elena Fillola and Rachel Tunnicliffe https://doi.org/10.5281/zenodo.16748754

Global NWP meteorological data for Met Office NAME dispersion model (Mk9: July 2015-2017) (Mk9), CEDA (Centre for Environmental Data Analysis) Met Office https://data.ceda.ac.uk/badc/name_nwp/data/global/UMG_Mk9

Global NWP meteorological data for Met Office NAME dispersion model (Mk10: June 2017-May 2022) (Mk10) Met Office https://data.ceda.ac.uk/badc/name_nwp/data/global/UMG_Mk10

University of Leicester GOSAT Proxy XCH4 v9.0 R. Parker and H. Boesch https://doi.org/10.5285/18ef8247f52a4cb6a14013f8235cc1eb

Model code and software

ACRG-Bristol/acrg: ACRG v0.2.0 (v0.2.0) Matthew Rigby et al. https://doi.org/10.5281/zenodo.6834888

GATES: A Graph-Neural-Network Atmospheric Transport Emulation System Elena Fillola et al. https://doi.org/10.5281/ZENODO.16679175

Short summary

Satellite-based greenhouse gas measurements can be used in “inverse models” to improve emissions reporting, but one of the key components, the simulations of atmospheric transport, struggle to scale to large datasets. We introduce the model GATES, an AI-driven emulator that outputs transport plumes 1000× faster than traditional models. Applied to Brazil’s methane emissions, GATES produces estimates consistent with physics-based methods, offering a scalable path for timely emissions monitoring.