Challenges and Conundrums in Modeling Global Methane Emissions from Wetlands: An Empiricist's Viewpoint

Abstract:

Wetlands emit a third to half of the global CH₄ flux and have the largest uncertainty of any emission source. Moreover, wetlands have provided an important radiative feedback to climate in the geologic and recent past. A number of largescale wetland CH₄ models have been developed recently, but intermodel comparisons show wide discrepancies in their predictions. I present an empiricist's overview of the current limitations and challenges of more accurately modeling wetland CH₄ emissions. One of the largest limitations is simply the poor knowledge of wetland area, with estimated global values varying by a more than a factor of three. The areas of seasonal and tropical wetlands are particularly poorly constrained. There are also few wetlands with complete, multi-year datasets for all of the input variables for many models, and this lack of data is particularly alarming in tropical wetlands given that they are arguably the single largest natural or anthropogenic global CH₄ source. Almost all largescale CH₄ models have little biogeochemical mechanistic detail and treat anaerobic carbon cycling in a highly simplified manner. The CH₄:CO₂ ratio in anaerobic carbon mineralization is a central parameter in many models, but is at most set at a few values with no mechanistic underpinning. However, empirical data show that this ratio varies by five orders of magnitude in different wetlands, and tropical wetlands appear to be particularly methanogenic, all for reasons that are very poorly understood. The predominance of the acetoclastic pathway of methanogenesis appears to be related to total CH₄ production, but different methanogenesis pathways are generally not incorporated into models. Other important anaerobic processes such as humic substances acting as terminal electron acceptors, fermentation, homoacetogenesis, and anaerobic CH₄ oxidation are also not included in most models despite evidence of their importance in empirical studies. Moreover, there has been an explosion of microbial studies in wetlands using high-throughput molecular techniques, but microbial community and functional parameters are largely missing from models. However, recently developed trait-based models show promise for reducing the multivariate complexity of this data into manageable parameters for large-scale CH₄ models.