Evaluating modeling approaches for estimating net ecosystem exchange in Canadian peatlands
10 affiliations
- 1Ecohydrology Research Group, Department of Earth and Environmental Sciences, University of Waterloo
- 2Water Institute, University of Waterloo
- 3Woodwell Climate Research Center
- 4Department of Geography and Planning, Queen’s University
- 5Geologic Survey of Canada, Natural Resources Canada
- 6School of Science and the Environment, Memorial University of Newfoundland
- 7Department of Geography, University of Winnipeg
- 8Département de Géographie, Université de Montréal
- 9Department of Geography, University of British Columbia
- 10Department of Geography, McGill University
Abstract
Peatlands are globally significant carbon reservoirs, yet models for peatland carbon cycling are often limited to the site level. To enable the prediction of carbon dioxide (CO2) fluxes in peatlands where no in situ measurements exist, machine learning algorithms must be trained on in situ CO2 flux measurements from multiple sites. In this study, year-round eddy covariance (EC)-derived net ecosystem exchange (NEE) measurements and 32 hydroclimatic predictor variables were compiled for 21 Canadian peatland sites spanning seven ecoregions. A comprehensive feature selection workflow carried out on the predictor variables (features) identified that model performance stabilized at four features, which were: evapotranspiration, burn area index, normalized difference water index, and modeled soil moisture. Four machine learning algorithms: ElasticNet Regression (EN), Light Gradient-Boosting Machine (LGBM), Random Forest Regression (RF), and Support Vector Regression (SVR) were trained and evaluated using held-out test data. The best performing model was the LGBM model, which was then assessed for generalizability via a leave-one-ecoregion-out sensitivity analysis, which highlighted the necessity of ensuring predictor variables fall within the range of the training data before applying this model framework to additional sites. These findings offer a framework for regional scaling to improve Canada’s spatially explicit CO2 emission estimates.
Related
- Machine learning and spatial regression approaches to estimating willingness to pay for ecosystem services in Jakarta and Taoyuan
- Active eye modeling: a tool for estimating retinal image quality across the visual field in everyday life
- Evaluating Clustering and Imputation Approaches for Multivariate Longitudinal Clinical Data Using a Real-World eCDF-Copula Resampling Framework
- Estimating latent neuronal nonlinear dynamics by sequential Monte Carlo method and sparse modeling
- How does the community structure of dung beetles affect ecosystem functions? Modeling of secondary seed dispersal and seedling establishment.
BibTeX
@article{hettinga2026evaluating,
title = {{Evaluating modeling approaches for estimating net ecosystem exchange in Canadian peatlands}},
author = {E. Kathryn Hettinga and Fereidoun Rezanezhad and Ali Reza Shahvaran and Mir Amir Mohammad Reshadi and Maria Strack and Kyle A. Arndt and Ian B. Strachan and June Skeeter and Jianghua Wu and Matthew Q. Morison and Haley Alcock and Oliver Sonnentag and Sara H. Knox and Rosemary Howard and Philippe Van Cappellen},
journal = {Scientific Reports},
year = {2026},
month = aug,
doi = {10.1038/s41598-026-67275-6},
url = {https://doi.org/10.1038/s41598-026-67275-6},
note = {CC BY},
}
RIS
TY - JOUR
TI - Evaluating modeling approaches for estimating net ecosystem exchange in Canadian peatlands
AU - E. Kathryn Hettinga
AU - Fereidoun Rezanezhad
AU - Ali Reza Shahvaran
AU - Mir Amir Mohammad Reshadi
AU - Maria Strack
AU - Kyle A. Arndt
AU - Ian B. Strachan
AU - June Skeeter
AU - Jianghua Wu
AU - Matthew Q. Morison
AU - Haley Alcock
AU - Oliver Sonnentag
AU - Sara H. Knox
AU - Rosemary Howard
AU - Philippe Van Cappellen
PY - 2026
DA - 2026/08/01
JO - Scientific Reports
DO - 10.1038/s41598-026-67275-6
UR - https://doi.org/10.1038/s41598-026-67275-6
AB - Peatlands are globally significant carbon reservoirs, yet models for peatland carbon cycling are often limited to the site level. To enable the prediction of carbon dioxide (CO2) fluxes in peatlands where no in situ measurements exist, machine learning algorithms must be trained on in situ CO2 flux measurements from multiple sites. In this study, year-round eddy covariance (EC)-derived net ecosystem exchange (NEE) measurements and 32 hydroclimatic predictor variables were compiled for 21 Canadian peatland sites spanning seven ecoregions. A comprehensive feature selection workflow carried out on the predictor variables (features) identified that model performance stabilized at four features, which were: evapotranspiration, burn area index, normalized difference water index, and modeled soil moisture. Four machine learning algorithms: ElasticNet Regression (EN), Light Gradient-Boosting Machine (LGBM), Random Forest Regression (RF), and Support Vector Regression (SVR) were trained and evaluated using held-out test data. The best performing model was the LGBM model, which was then assessed for generalizability via a leave-one-ecoregion-out sensitivity analysis, which highlighted the necessity of ensuring predictor variables fall within the range of the training data before applying this model framework to additional sites. These findings offer a framework for regional scaling to improve Canada’s spatially explicit CO2 emission estimates.
SN - 2045-2322
ER -