You are here:
Methane and carbon dioxide emissions from reservoirs: Controls and Upscaling
Citation:
Beaulieu, J., S. Waldo, D. Balz, W. Barnett, A. Hall, M. Platz, AND K. White. Methane and carbon dioxide emissions from reservoirs: Controls and Upscaling. Journal of Geophysical Research: Biogeosciences. American Geophysical Union, Washington, DC, 125(12):e2019JG005474, (2020). https://doi.org/10.1029/2019JG005474
Impact/Purpose:
The US EPA reports an annual inventory of U.S. anthropogenic greenhouse gas emissions (GHG) to the United Nations in accordance with our obligations under the United Nations Framework Convention on Climate Change treaty. Data suggest that surface water reservoirs may be a nationally significant source of methane in the U.S., though additional feid data are required to generate a robust national-scale estimate. In this work we used a systematic approach to measure methane and carbon dioxide emission rates from 32 reservoirs in the Midwestern U.S. We then modeled the data using boosted regression trees and show that emission rates can be predicted from readily available data (i.e. reservoir size). Our predictions indicate that reservoirs are the fourth largest anthropogenic CH4 source in Indiana, Ohio, and Kentucky. These data and models will be of use to GHG inventory compilers in the US EPA's Office of Air and Radiation and across the globe.
Description:
Estimating carbon dioxide (CO2) and methane (CH4) emission rates from reservoirs is important for regional and national greenhouse gas inventories. A lack of methodologically consistent data sets for many parts of the world, including agriculturally intensive areas of the U.S., poses a major challenge to the development of models for predicting emission rates. In this study we used a systematic measurement approach to measure CO2 and CH4 diffusive and ebullitive emission rates from 32 reservoirs distributed across an agricultural and to forested land-use gradient in the U.S.. We found that all reservoirs were a source of CH4 to the atmosphere, with ebullition being the dominant emission pathway in 75% of the systems. Ebullition was a negligible emission pathway for CO2 emissions and 65% of sampled reservoirs were a net CO2 sink. Boosted regression trees (BRT), a type of machine learning algorithm, fit the data well and identified reservoir morphology and productivity as important predictors of emission rates. We used the BRT to predict CH4 emission rates for the 11,885 reservoirs in our study area and estimate that they are the fourth largest anthropogenic CH4 sources in the region, with emissions equivalent to 25% of CH4 emissions from municipal landfills in the region. Our work demonstrates that machine learning algorithms are a promising tool for predicting CH4 emission rates based on information readily available in national geodatabases.
URLs/Downloads:
DOI: Methane and carbon dioxide emissions from reservoirs: Controls and Upscaling
Free access through PubMed Central