Abstract

Wetlands are responsible for 20%-31% of global methane (CH4) emissions and account for a large source of uncertainty in the global CH4 budget. Data-driven upscaling of CH4 fluxes from eddy covariance measurements can provide new and independent bottom-up estimates of wetland CH4 emissions. Here, we develop a six-predictor random forest upscaling model (UpCH4), trained on 119 site-years of eddy covariance CH4 flux data from 43 freshwater wetland sites in the FLUXNET-CH4 Community Product. Network patterns in site-level annual means and mean seasonal cycles of CH4 fluxes were reproduced accurately in tundra, boreal, and temperate regions (Nash-Sutcliffe Efficiency similar to 0.52-0.63 and 0.53). UpCH(4) estimated annual global wetland CH4 emissions of 146 +/- 43 TgCH4 y(-1) for 2001-2018 which agrees closely with current bottom-up land surface models (102-181 TgCH4 y(-1)) and overlaps with top-down atmospheric inversion models (155-200 TgCH4 y -1). However, UpCH4 diverged from both types of models in the spatial pattern and seasonal dynamics of tropical wetland emissions. We conclude that upscaling of eddy covariance CH4 fluxes has the potential to produce realistic extra-tropical wetland CH4 emissions estimates which will improve with more flux data. To reduce uncertainty in upscaled estimates, researchers could prioritize new wetland flux sites along humid-to-arid tropical climate gradients, from major rainforest basins (Congo, Amazon, and SE Asia), into monsoon (Bangladesh and India) and savannah regions (African Sahel) and be paired with improved knowledge of wetland extent seasonal dynamics in these regions. The monthly wetland methane products gridded at 0.25 degrees from UpCH4 are available via ORNL DAAC (https://doi.org/10.3334/ ORNLDAAC/2253).Plain Language Summary Wetlands account for a large share of global methane emissions to the atmosphere, but current estimates vary widely in magnitude (similar to 30% uncertainty on annual global emissions) and spatial distribution, with diverging predictions for tropical rice growing (e.g., Bengal basin), rainforest (e.g., Amazon basin), and floodplain savannah (e.g., Sudd) regions. Wetland methane model estimates could be improved by increased use of land surface methane flux data. Upscaling approaches use flux data collected across globally distributed measurement networks in a machine learning framework to extrapolate fluxes in space and time. Here, we train and evaluate a methane upscaling model (UpCH4) and use it to generate monthly, globally gridded wetland methane emissions estimates for 2001-2018. The UpCH4 model uses only six predictor variables among which temperature is dominant. Global annual methane emissions estimates and associated uncertainty ranges from upscaling fall within state-of-the-art model ensemble estimates from the Global Carbon Project (GCP) methane budget. In some tropical regions, the spatial pattern of UpCH4 emissions diverged from GCP predictions, however, inclusion of flux measurements from additional ground-based sites, together with refined maps of tropical wetlands extent, could reduce these prediction uncertainties.

Keywords

global; wetland; methane; eddy covariance; flux; random forest

Published in

AGU Advances
2023, Volume: 4, number: 5, article number: e2023AV000956
Publisher: AMER GEOPHYSICAL UNION

SLU Authors

• Nilsson, Mats

  • Department of Forest Ecology and Management, Swedish University of Agricultural Sciences
    

• Peichl, Matthias

  • Department of Forest Ecology and Management, Swedish University of Agricultural Sciences
    

Sustainable Development Goals

Ensure availability and sustainable management of water and sanitation for all


UKÄ Subject classification

Climate Research


Publication identifier

DOI: https://doi.org/10.1029/2023AV000956

Permanent link to this page (URI)

https://res.slu.se/id/publ/126488