Advancing Methane Biogeochemistry Modeling with Machine Learning Technique

Active Project
Advancing Methane Biogeochemistry Modeling with Machine Learning Technique

Focus Areas: 

  • Earth Systems

Methane (CH4) accounts for up to 25% of atmospheric warming to date, but large uncertainty exists in methane emissions estimates from wetlands (the largest natural CH4 source) using biogeochemistry models. This uncertainty arises largely because wetland CH4 dynamics depend on a diverse array of poorly-represented physical, biological and chemical processes, as well as a large number of poorly-constrained uncertain parameters to characterize these processes. This project will examine the sensitivity of CH4 emissions to a large set of parameters and optimize the most sensitive parameters, at observation sites covering a wide range of soil types, vegetation types and climatic conditions. The main goals are to provide insights into key parameters that drive uncertainty in wetland CH4 emissions at each site and parameter transferability between sites; enhance the process-level understanding of mechanisms and controls underlying CH4 biogeochemistry; and enable more reliable projections of the magnitude and variability of global and regional wetland CH4 emissions under a changing climate.

Funding Sources

Project Leaders

Research staff
CGCS; Joint Program