Groundwater–CO<sub>2</sub> emissions relationship in Dutch peatlands derived by machine learning using airborne and ground-based eddy covariance data

<p>Peatlands worldwide have been transformed from carbon sinks to carbon sources due to years of intensive agriculture requiring low water tables. In the Netherlands, carbon dioxide (CO<span class="inline-formula"><sub>2</sub></span>) emissions from drained pe...

Full description

Saved in:
Bibliographic Details
Main Authors: L. M. van der Poel, L. V. Bataille, B. Kruijt, W. Franssen, W. Jans, J. Biermann, A. Rietman, A. J. V. Buzacott, Y. van der Velde, R. Boelens, R. W. A. Hutjes
Format: Article
Language:English
Published: Copernicus Publications 2025-08-01
Series:Biogeosciences
Online Access:https://bg.copernicus.org/articles/22/3867/2025/bg-22-3867-2025.pdf
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849407078104301568
author L. M. van der Poel
L. V. Bataille
B. Kruijt
W. Franssen
W. Jans
J. Biermann
A. Rietman
A. J. V. Buzacott
Y. van der Velde
R. Boelens
R. W. A. Hutjes
author_facet L. M. van der Poel
L. V. Bataille
B. Kruijt
W. Franssen
W. Jans
J. Biermann
A. Rietman
A. J. V. Buzacott
Y. van der Velde
R. Boelens
R. W. A. Hutjes
author_sort L. M. van der Poel
collection DOAJ
description <p>Peatlands worldwide have been transformed from carbon sinks to carbon sources due to years of intensive agriculture requiring low water tables. In the Netherlands, carbon dioxide (CO<span class="inline-formula"><sub>2</sub></span>) emissions from drained peatlands mount up to 5.6 Mton annually and, according to the Dutch climate agreement, should be reduced by 1 Mton by 2030. It is generally accepted that mitigation measures should include raising the water level, and the exact influence of water table depth has been increasingly studied in recent years. Most studies do this by comparing annual eddy covariance (EC) site-specific CO<span class="inline-formula"><sub>2</sub></span> budgets to mean annual effective water table depths. However, here we apply a different approach: we integrate measurements from 16 EC towers with EC measurements from 141 flights by a low-flying research aircraft in an interpretable machine learning (ML) framework. We make use of the different strengths of tower and airborne data, temporal continuity, and spatial heterogeneity. We apply time frequency wavelet analysis and a footprint model to relate the measured fluxes to the underlying surface. Using spatiotemporal data, we train and optimize a boosted regression tree (BRT) machine learning algorithm to predict immediate CO<span class="inline-formula"><sub>2</sub></span> fluxes and use Shapley values and various simulations to interpret the model's outputs. We find that emissions increase by 4.6 t CO<span class="inline-formula"><sub>2</sub></span> ha<span class="inline-formula"><sup>−1</sup></span> yr<span class="inline-formula"><sup>−1</sup></span> (90 % confidence interval: 4.0–5.4) for every 10 cm lowering of the water table, down to a water table depth of 0.8 m below the surface. For more drained conditions, emissions decrease again. Furthermore, we find that the sensitivity of CO<span class="inline-formula"><sub>2</sub></span> emissions to drainage is stronger in winter than in summer and that it varies between sites. This study shows the added value of using ML with different types of instantaneous data and holds potential for future applications.</p>
format Article
id doaj-art-4e7276ada6a748a9af3c7c5f43a0287a
institution Kabale University
issn 1726-4170
1726-4189
language English
publishDate 2025-08-01
publisher Copernicus Publications
record_format Article
series Biogeosciences
spelling doaj-art-4e7276ada6a748a9af3c7c5f43a0287a2025-08-20T03:36:11ZengCopernicus PublicationsBiogeosciences1726-41701726-41892025-08-01223867389810.5194/bg-22-3867-2025Groundwater–CO<sub>2</sub> emissions relationship in Dutch peatlands derived by machine learning using airborne and ground-based eddy covariance dataL. M. van der Poel0L. V. Bataille1B. Kruijt2W. Franssen3W. Jans4J. Biermann5A. Rietman6A. J. V. Buzacott7Y. van der Velde8R. Boelens9R. W. A. Hutjes10Earth Systems and Global Change Group, Wageningen University, 6708 PB Wageningen, the NetherlandsEarth Systems and Global Change Group, Wageningen University, 6708 PB Wageningen, the NetherlandsEarth Systems and Global Change Group, Wageningen University, 6708 PB Wageningen, the NetherlandsEarth Systems and Global Change Group, Wageningen University, 6708 PB Wageningen, the NetherlandsEarth Systems and Global Change Group, Wageningen University, 6708 PB Wageningen, the NetherlandsEarth Systems and Global Change Group, Wageningen University, 6708 PB Wageningen, the NetherlandsEarth Systems and Global Change Group, Wageningen University, 6708 PB Wageningen, the NetherlandsInstitute for Atmospheric and Earth System Research, University of Helsinki, 00014 Helsinki, FinlandFaculty of Science, Earth and Climate, Vrije Universiteit Amsterdam, 1081 HV Amsterdam, the NetherlandsHydroLogic, 3811 HN Amersfoort, the NetherlandsEarth Systems and Global Change Group, Wageningen University, 6708 PB Wageningen, the Netherlands<p>Peatlands worldwide have been transformed from carbon sinks to carbon sources due to years of intensive agriculture requiring low water tables. In the Netherlands, carbon dioxide (CO<span class="inline-formula"><sub>2</sub></span>) emissions from drained peatlands mount up to 5.6 Mton annually and, according to the Dutch climate agreement, should be reduced by 1 Mton by 2030. It is generally accepted that mitigation measures should include raising the water level, and the exact influence of water table depth has been increasingly studied in recent years. Most studies do this by comparing annual eddy covariance (EC) site-specific CO<span class="inline-formula"><sub>2</sub></span> budgets to mean annual effective water table depths. However, here we apply a different approach: we integrate measurements from 16 EC towers with EC measurements from 141 flights by a low-flying research aircraft in an interpretable machine learning (ML) framework. We make use of the different strengths of tower and airborne data, temporal continuity, and spatial heterogeneity. We apply time frequency wavelet analysis and a footprint model to relate the measured fluxes to the underlying surface. Using spatiotemporal data, we train and optimize a boosted regression tree (BRT) machine learning algorithm to predict immediate CO<span class="inline-formula"><sub>2</sub></span> fluxes and use Shapley values and various simulations to interpret the model's outputs. We find that emissions increase by 4.6 t CO<span class="inline-formula"><sub>2</sub></span> ha<span class="inline-formula"><sup>−1</sup></span> yr<span class="inline-formula"><sup>−1</sup></span> (90 % confidence interval: 4.0–5.4) for every 10 cm lowering of the water table, down to a water table depth of 0.8 m below the surface. For more drained conditions, emissions decrease again. Furthermore, we find that the sensitivity of CO<span class="inline-formula"><sub>2</sub></span> emissions to drainage is stronger in winter than in summer and that it varies between sites. This study shows the added value of using ML with different types of instantaneous data and holds potential for future applications.</p>https://bg.copernicus.org/articles/22/3867/2025/bg-22-3867-2025.pdf
spellingShingle L. M. van der Poel
L. V. Bataille
B. Kruijt
W. Franssen
W. Jans
J. Biermann
A. Rietman
A. J. V. Buzacott
Y. van der Velde
R. Boelens
R. W. A. Hutjes
Groundwater–CO<sub>2</sub> emissions relationship in Dutch peatlands derived by machine learning using airborne and ground-based eddy covariance data
Biogeosciences
title Groundwater–CO<sub>2</sub> emissions relationship in Dutch peatlands derived by machine learning using airborne and ground-based eddy covariance data
title_full Groundwater–CO<sub>2</sub> emissions relationship in Dutch peatlands derived by machine learning using airborne and ground-based eddy covariance data
title_fullStr Groundwater–CO<sub>2</sub> emissions relationship in Dutch peatlands derived by machine learning using airborne and ground-based eddy covariance data
title_full_unstemmed Groundwater–CO<sub>2</sub> emissions relationship in Dutch peatlands derived by machine learning using airborne and ground-based eddy covariance data
title_short Groundwater–CO<sub>2</sub> emissions relationship in Dutch peatlands derived by machine learning using airborne and ground-based eddy covariance data
title_sort groundwater co sub 2 sub emissions relationship in dutch peatlands derived by machine learning using airborne and ground based eddy covariance data
url https://bg.copernicus.org/articles/22/3867/2025/bg-22-3867-2025.pdf
work_keys_str_mv AT lmvanderpoel groundwatercosub2subemissionsrelationshipindutchpeatlandsderivedbymachinelearningusingairborneandgroundbasededdycovariancedata
AT lvbataille groundwatercosub2subemissionsrelationshipindutchpeatlandsderivedbymachinelearningusingairborneandgroundbasededdycovariancedata
AT bkruijt groundwatercosub2subemissionsrelationshipindutchpeatlandsderivedbymachinelearningusingairborneandgroundbasededdycovariancedata
AT wfranssen groundwatercosub2subemissionsrelationshipindutchpeatlandsderivedbymachinelearningusingairborneandgroundbasededdycovariancedata
AT wjans groundwatercosub2subemissionsrelationshipindutchpeatlandsderivedbymachinelearningusingairborneandgroundbasededdycovariancedata
AT jbiermann groundwatercosub2subemissionsrelationshipindutchpeatlandsderivedbymachinelearningusingairborneandgroundbasededdycovariancedata
AT arietman groundwatercosub2subemissionsrelationshipindutchpeatlandsderivedbymachinelearningusingairborneandgroundbasededdycovariancedata
AT ajvbuzacott groundwatercosub2subemissionsrelationshipindutchpeatlandsderivedbymachinelearningusingairborneandgroundbasededdycovariancedata
AT yvandervelde groundwatercosub2subemissionsrelationshipindutchpeatlandsderivedbymachinelearningusingairborneandgroundbasededdycovariancedata
AT rboelens groundwatercosub2subemissionsrelationshipindutchpeatlandsderivedbymachinelearningusingairborneandgroundbasededdycovariancedata
AT rwahutjes groundwatercosub2subemissionsrelationshipindutchpeatlandsderivedbymachinelearningusingairborneandgroundbasededdycovariancedata