Knowledge-guided machine learning can improve carbon cycle quantification in agroecosystems

Licheng Liu; Wang Zhou; Kaiyu Guan; Bin Peng; Shaoming Xu; Jinyun Tang; Qing Zhu; Jessica Till; Xiaowei Jia; Chongya Jiang; Sheng Wang; Ziqi Qin; Hui Kong; Robert F Grant; Symon  Mezbahuddin; Vipin Kumar; Zhenong Jin

doi:10.1038/s41467-023-43860-5

Knowledge-guided machine learning can improve carbon cycle quantification in agroecosystems

Licheng Liu
, Wang Zhou
, Kaiyu Guan^*
, Bin Peng
, Shaoming Xu
, Jinyun Tang
, Qing Zhu
, Jessica Till
, Xiaowei Jia
, Chongya Jiang
, Sheng Wang
, Ziqi Qin
, Hui Kong
, Robert F Grant
, Symon Mezbahuddin
, Vipin Kumar
, Zhenong Jin

^*Corresponding author for this work

Department of Agroecology - Crop Health

Research output: Contribution to journal/Conference contribution in journal/Contribution to newspaper › Journal article › Research › peer-review

100 Citations (Scopus)

Abstract

Accurate and cost-effective quantification of the carbon cycle for agroecosystems at decision-relevant scales is critical to mitigating climate change and ensuring sustainable food production. However, conventional process-based or data-driven modeling approaches alone have large prediction uncertainties due to the complex biogeochemical processes to model and the lack of observations to constrain many key state and flux variables. Here we propose a Knowledge-Guided Machine Learning (KGML) framework that addresses the above challenges by integrating knowledge embedded in a process-based model, high-resolution remote sensing observations, and machine learning (ML) techniques. Using the U.S. Corn Belt as a testbed, we demonstrate that KGML can outperform conventional process-based and black-box ML models in quantifying carbon cycle dynamics. Our high-resolution approach quantitatively reveals 86% more spatial detail of soil organic carbon changes than conventional coarse-resolution approaches. Moreover, we outline a protocol for improving KGML via various paths, which can be generalized to develop hybrid models to better predict complex earth system dynamics.

Original language	English
Article number	357
Journal	Nature Communications
Volume	15
Issue	1
Number of pages	15
ISSN	2041-1723
DOIs	https://doi.org/10.1038/s41467-023-43860-5
Publication status	Published - Dec 2024

Access to Document

10.1038/s41467-023-43860-5Licence: CC BY

Library access?

Check availability with the Royal Danish Library

Cite this

@article{b45d1f19864e44978f00bef2258cbcc4,

title = "Knowledge-guided machine learning can improve carbon cycle quantification in agroecosystems",

abstract = "Accurate and cost-effective quantification of the carbon cycle for agroecosystems at decision-relevant scales is critical to mitigating climate change and ensuring sustainable food production. However, conventional process-based or data-driven modeling approaches alone have large prediction uncertainties due to the complex biogeochemical processes to model and the lack of observations to constrain many key state and flux variables. Here we propose a Knowledge-Guided Machine Learning (KGML) framework that addresses the above challenges by integrating knowledge embedded in a process-based model, high-resolution remote sensing observations, and machine learning (ML) techniques. Using the U.S. Corn Belt as a testbed, we demonstrate that KGML can outperform conventional process-based and black-box ML models in quantifying carbon cycle dynamics. Our high-resolution approach quantitatively reveals 86\% more spatial detail of soil organic carbon changes than conventional coarse-resolution approaches. Moreover, we outline a protocol for improving KGML via various paths, which can be generalized to develop hybrid models to better predict complex earth system dynamics.",

author = "Licheng Liu and Wang Zhou and Kaiyu Guan and Bin Peng and Shaoming Xu and Jinyun Tang and Qing Zhu and Jessica Till and Xiaowei Jia and Chongya Jiang and Sheng Wang and Ziqi Qin and Hui Kong and Grant, \{Robert F\} and Symon Mezbahuddin and Vipin Kumar and Zhenong Jin",

year = "2024",

month = dec,

doi = "10.1038/s41467-023-43860-5",

language = "English",

volume = "15",

journal = "Nature Communications",

issn = "2041-1723",

publisher = "Nature Research",

number = "1",

}

TY - JOUR

T1 - Knowledge-guided machine learning can improve carbon cycle quantification in agroecosystems

AU - Liu, Licheng

AU - Zhou, Wang

AU - Guan, Kaiyu

AU - Peng, Bin

AU - Xu, Shaoming

AU - Tang, Jinyun

AU - Zhu, Qing

AU - Till, Jessica

AU - Jia, Xiaowei

AU - Jiang, Chongya

AU - Wang, Sheng

AU - Qin, Ziqi

AU - Kong, Hui

AU - Grant, Robert F

AU - Mezbahuddin, Symon

AU - Kumar, Vipin

AU - Jin, Zhenong

PY - 2024/12

Y1 - 2024/12

N2 - Accurate and cost-effective quantification of the carbon cycle for agroecosystems at decision-relevant scales is critical to mitigating climate change and ensuring sustainable food production. However, conventional process-based or data-driven modeling approaches alone have large prediction uncertainties due to the complex biogeochemical processes to model and the lack of observations to constrain many key state and flux variables. Here we propose a Knowledge-Guided Machine Learning (KGML) framework that addresses the above challenges by integrating knowledge embedded in a process-based model, high-resolution remote sensing observations, and machine learning (ML) techniques. Using the U.S. Corn Belt as a testbed, we demonstrate that KGML can outperform conventional process-based and black-box ML models in quantifying carbon cycle dynamics. Our high-resolution approach quantitatively reveals 86% more spatial detail of soil organic carbon changes than conventional coarse-resolution approaches. Moreover, we outline a protocol for improving KGML via various paths, which can be generalized to develop hybrid models to better predict complex earth system dynamics.

AB - Accurate and cost-effective quantification of the carbon cycle for agroecosystems at decision-relevant scales is critical to mitigating climate change and ensuring sustainable food production. However, conventional process-based or data-driven modeling approaches alone have large prediction uncertainties due to the complex biogeochemical processes to model and the lack of observations to constrain many key state and flux variables. Here we propose a Knowledge-Guided Machine Learning (KGML) framework that addresses the above challenges by integrating knowledge embedded in a process-based model, high-resolution remote sensing observations, and machine learning (ML) techniques. Using the U.S. Corn Belt as a testbed, we demonstrate that KGML can outperform conventional process-based and black-box ML models in quantifying carbon cycle dynamics. Our high-resolution approach quantitatively reveals 86% more spatial detail of soil organic carbon changes than conventional coarse-resolution approaches. Moreover, we outline a protocol for improving KGML via various paths, which can be generalized to develop hybrid models to better predict complex earth system dynamics.

UR - https://www.scopus.com/pages/publications/85181710208

U2 - 10.1038/s41467-023-43860-5

DO - 10.1038/s41467-023-43860-5

M3 - Journal article

C2 - 38191521

SN - 2041-1723

VL - 15

JO - Nature Communications

JF - Nature Communications

IS - 1

M1 - 357

ER -

Knowledge-guided machine learning can improve carbon cycle quantification in agroecosystems

Abstract

Access to Document

Library access?

Other files and links

Fingerprint

Cite this