Multi-Decadal Dynamics of Wetland Methane Emissions Revealed by Knowledge-Guided Machine Learning

Measurement of methane fluxes (FCH₄) from natural systems, such as wetlands, has lagged far behind carbon dioxide fluxes. Short and fragmented wetland FCH₄ data limit our ability to assess its long-term dynamics and potential climate feedbacks. Extrapolating short-term FCH₄ records to recent decades remains challenging for both process-based models and data-driven machine learning (ML) approaches. Here, we develop a knowledge-guided ML framework that integrates eddy covariance (EC) FCH₄ observations, field warming experiments, and biogeochemical knowledge to reconstruct the long-term FCH₄ budgets and trends. Focusing on the 11 longest EC monitoring sites in the AmeriFlux network, we found considerable variability in multi-decadal trends of wetland FCH₄, with increases up to 14% per decade from 2000 to 2024. We also found that the strength of these increasing trends declines from high to low latitudes, highlighting the vulnerability of northern wetlands. This work presents novel and robust reconstructions of long-term wetland FCH₄, offering critical benchmark datasets for bottom-up ecosystem models and advancing fundamental understanding of wetland biogeochemistry.