1. Gap-filling eddy covariance methane fluxes : Comparison of machine learning model predictions and uncertainties at FLUXNET-CH4 wetlands
- Author
-
Sheel Bansal, Lisamarie Windham-Myers, Karina V. R. Schäfer, Christian Wille, Han Dolman, Hiroki Iwata, Mats Nilsson, Robert Shortt, Andrew D. Richardson, Pavel Alekseychik, Sarah Feron, Benjamin Poulter, David P. Billesbach, Kyle B. Delwiche, Walter C. Oechel, Anand Avati, Avni Malhotra, Jiquan Chen, Fred Lu, Ivan Mammarella, Lutz Merbold, Ankur R. Desai, Robert B. Jackson, Pia Gottschalk, Carole Helfter, Bhaskar Mitra, Kathrin Fuchs, Takashi Hirano, Manuel Helbig, Edward A. G. Schuur, M. Goeckede, Domenico Vitale, Zutao Ouyang, Andrew Y. Ng, Mangaliso J. Gondwe, Regine Maier, M. C. R. Alberto, Asko Noormets, Thomas Friborg, Patricia Y. Oikawa, Torsten Sachs, Franziska Koebsch, Eiko Nemitz, Andrej Varlagin, Dario Papale, Keisuke Ono, Jeremy Irvin, Matthias Peichl, Jordan P. Goodrich, Carlo Trotta, Gil Bohrer, Gerardo Celis, David I. Campbell, Camilo Rey-Sanchez, Vincent Liu, Sara H. Knox, Benjamin R. K. Runkle, Sébastien Gogo, Andrew Kondrich, Guan Xhuan Wong, Sharon Zhou, Housen Chu, Kuno Kasak, Lukas Hörtnagl, Timothy H. Morin, Oliver Sonnentag, George L. Vourlitis, Rodrigo Vargas, Derrick Y.F. Lai, Kyle S. Hemes, Ryan C. Sullivan, E. J. Ward, Masahito Ueyama, Annalea Lohila, Gerald Jurasinski, Daphne Szutu, Eeva-Stiina Tuittila, Gavin McNicol, Donatella Zona, Ayaka Sakabe, Cove Sturtevant, Aram Kalhori, Antje Lucas-Moffat, Mika Aurela, Dennis D. Baldocchi, Martin Heimann, Eugénie S. Euskirchen, Adrien Jacotot, Alex C. Valach, Ellen Stuart-Haëntjens, Joeseph G. Verfaillie, Higo J. Dalmagro, Etienne Fluet-Chouinard, Institute for Atmospheric and Earth System Research (INAR), Micrometeorology and biogeochemical cycles, Earth and Climate, Earth Sciences, Institut des Sciences de la Terre d'Orléans - UMR7327 (ISTO), Centre National de la Recherche Scientifique (CNRS)-Université d'Orléans (UO)-Institut national des sciences de l'Univers (INSU - CNRS)-Observatoire des Sciences de l'Univers en région Centre (OSUC), Institut national des sciences de l'Univers (INSU - CNRS)-Observatoire de Paris, Université Paris sciences et lettres (PSL)-Université Paris sciences et lettres (PSL)-Université d'Orléans (UO)-Centre National de la Recherche Scientifique (CNRS)-Institut national des sciences de l'Univers (INSU - CNRS)-Observatoire de Paris, Université Paris sciences et lettres (PSL)-Université Paris sciences et lettres (PSL)-Université d'Orléans (UO)-Centre National de la Recherche Scientifique (CNRS)-Bureau de Recherches Géologiques et Minières (BRGM) (BRGM), Biogéosystèmes Continentaux - UMR7327, and Université Paris sciences et lettres (PSL)-Université Paris sciences et lettres (PSL)-Université d'Orléans (UO)-Centre National de la Recherche Scientifique (CNRS)-Bureau de Recherches Géologiques et Minières (BRGM) (BRGM)-Centre National de la Recherche Scientifique (CNRS)-Université d'Orléans (UO)-Institut national des sciences de l'Univers (INSU - CNRS)-Observatoire des Sciences de l'Univers en région Centre (OSUC)
- Subjects
Atmospheric Science ,010504 meteorology & atmospheric sciences ,imputation ,computer.software_genre ,01 natural sciences ,CARBON-DIOXIDE ,FluxNet ,Imputation (statistics) ,ENVIRONMENTAL DRIVERS ,EMISSIONS ,Global and Planetary Change ,Artificial neural network ,methane ,Sampling (statistics) ,Forestry ,04 agricultural and veterinary sciences ,6. Clean water ,machine learning ,RESPIRATION ,CO2 ,Marginal distribution ,SDG 6 - Clean Water and Sanitation ,gap-filling ,CH4 FLUX ,time series ,methane flux ,wetlands ,ASSIMILATION ,Eddy covariance ,Decision tree ,Machine learning ,114 Physical sciences ,[SDU.ENVI]Sciences of the Universe [physics]/Continental interfaces, environment ,Baseline (configuration management) ,0105 earth and related environmental sciences ,NET ECOSYSTEM EXCHANGE ,business.industry ,15. Life on land ,flux ,PERSPECTIVES ,[SDU]Sciences of the Universe [physics] ,13. Climate action ,040103 agronomy & agriculture ,0401 agriculture, forestry, and fisheries ,Environmental science ,Artificial intelligence ,business ,METHODOLOGY ,Agronomy and Crop Science ,computer - Abstract
Time series of wetland methane fluxes measured by eddy covariance require gap-filling to estimate daily, seasonal, and annual emissions. Gap-filling methane fluxes is challenging because of high variability and complex responses to multiple drivers. To date, there is no widely established gap-filling standard for wetland methane fluxes, with regards both to the best model algorithms and predictors. This study synthesizes results of different gap-filling methods systematically applied at 17 wetland sites spanning boreal to tropical regions and including all major wetland classes and two rice paddies. Procedures are proposed for: 1) creating realistic artificial gap scenarios, 2) training and evaluating gap-filling models without overstating performance, and 3) predicting half-hourly methane fluxes and annual emissions with realistic uncertainty estimates. Performance is compared between a conventional method (marginal distribution sampling) and four machine learning algorithms. The conventional method achieved similar median performance as the machine learning models but was worse than the best machine learning models and relatively insensitive to predictor choices. Of the machine learning models, decision tree algorithms performed the best in cross-validation experiments, even with a baseline predictor set, and artificial neural networks showed comparable performance when using all predictors. Soil temperature was frequently the most important predictor whilst water table depth was important at sites with substantial water table fluctuations, highlighting the value of data on wetland soil conditions. Raw gap-filling uncertainties from the machine learning models were underestimated and we propose a method to calibrate uncertainties to observations. The python code for model development, evaluation, and uncertainty estimation is publicly available. This study outlines a modular and robust machine learning workflow and makes recommendations for, and evaluates an improved baseline of, methane gap-filling models that can be implemented in multi-site syntheses or standardized products from regional and global flux networks (e.g., FLUXNET). ISSN:0168-1923 ISSN:1873-2240
- Published
- 2021