Key Points
• Identified five key sensitive parameters for methane emissions using the Sobol sensitivity analysis method.
• Parameters linked to production and diffusion present the highest sensitivities despite apparent seasonal variation.
• Fourteen out of nineteen model parameters exert negligible influence on methane emissions.

Plain Language Summary
Methane is a critical greenhouse gas, and wetlands are the largest natural source of it. Accurately predicting methane emissions from wetlands is key to tackling climate change. But these predictions, made through computer models, are seldom spot-on. Why? Because there are many factors in the models that lead to uncertain predictions. A major source of this uncertainty arises from the empirical model parameters. Just as tuning a radio dial ensures clear reception, models need properly adjusted parameters for accurate predictions. 

A sensitivity analysis was performed to determine which parameters are most crucial for accurate predictions. Instead of running the complex numerical model every time, machine learning was employed to create a faster and simpler version.

Using this approach, five parameters were pinpointed as particularly sensitive, significantly impacting the predictions. The comparison of model-predicted methane emissions with real-world measurements showed that the model performed well in some cases but needed tweaking in others. Refining these sensitive parameters with more real-world observations could make better predictions in the future.

Abstract
Methane (CH4) is globally the second most critical greenhouse gas after carbon dioxide, contributing to 16-25% of the observed atmospheric warming. Wetlands are the primary natural source of methane emissions globally. However, wetland methane emission estimates from biogeochemistry models contain considerable uncertainty. One of the main sources of this uncertainty arises from the numerous uncertain model parameters within various physical, biological, and chemical processes that influence methane production, oxidation, and transport. Sensitivity Analysis (SA) can help identify critical parameters for methane emission and achieve reduced biases and uncertainties in future projections.

This study performs SA for 19 selected parameters responsible for critical biogeochemical processes in the methane module of the Energy Exascale Earth System Model (E3SM) land model (ELM). The impact of these parameters on various CH4 fluxes is examined at 14 FLUXNET- CH4 sites with diverse vegetation types. Given the extensive number of model simulations needed for global variance-based SA, we employ a machine learning (ML) algorithm to emulate the complex behavior of ELM methane biogeochemistry.

We found that parameters linked to CH4 production and diffusion generally present the highest sensitivities despite apparent seasonal variation. Comparing simulated emissions from perturbed parameter sets against FLUXNET-CH4 observations revealed that better performances can be achieved at each site compared to the default parameter values. This presents a scope for further improving simulated emissions using parameter calibration with advanced optimization techniques.