Preprints
https://doi.org/10.5194/acp-2021-631
https://doi.org/10.5194/acp-2021-631

  28 Jul 2021

28 Jul 2021

Review status: a revised version of this preprint is currently under review for the journal ACP.

Technical note: Uncertainties in eddy covariance CO2 fluxes in a semi-arid sagebrush ecosystem caused by gap-filling approaches

Jingyu Yao1, Zhongming Gao2, Jianping Huang1, Heping Liu2, and Guoyin Wang3 Jingyu Yao et al.
  • 1Key Laboratory for Semi-Arid Climate Change of the Ministry of Education, College of Atmospheric Sciences, Lanzhou University, Lanzhou, China
  • 2Laboratory for Atmospheric Research, Department of Civil and Environmental Engineering, Washington State University, Pullman, Washington, USA
  • 3Department of Atmospheric and Oceanic Sciences & Institute of Atmospheric Sciences, Fudan University, Shanghai, China

Abstract. Gap-filling eddy covariance CO2 fluxes is challenging at dryland sites due to small CO2 fluxes. Here, four machine learning (ML) algorithms including artificial neural network (ANN), k-nearest neighbours (KNN), random forest (RF), and support vector machine (SVM) are employed and evaluated for gap-filling CO2 fluxes over a semi-arid sagebrush ecosystem with different lengths of artificial gaps. The ANN and RF algorithms outperform the KNN and SVM in filling gaps ranging from hours to days, with the RF being more time efficient than the ANN. Performances of the ANN and RF are largely degraded for extremely long gaps of two months. In addition, our results suggest that there is no need to fill the daytime and nighttime NEE gaps separately when using the ANN and RF. With the ANN and RF, the gap-filling induced uncertainties in the annual NEE at this site are estimated to be within 16 g C m−2, whereas the uncertainties by the KNN and SVM can be as large as 27 g C m−2. To better fill extremely long gaps of a few months, we test a two-layer gap-filling framework based on the RF. With this framework, the model performance is improved significantly, especially for the nighttime data. Therefore, this approach provides an alternative in filling extremely long gaps to characterize annual carbon budgets and interannual variability in dryland ecosystems.

Jingyu Yao et al.

Status: final response (author comments only)

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse
  • RC1: 'Comment on acp-2021-631', Anonymous Referee #1, 17 Aug 2021
  • RC2: 'Comment on acp-2021-631', Anonymous Referee #2, 31 Aug 2021
    • AC1: 'Reply on RC1', Jingyu Yao, 23 Sep 2021

Jingyu Yao et al.

Jingyu Yao et al.

Viewed

Total article views: 280 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
203 66 11 280 3 3
  • HTML: 203
  • PDF: 66
  • XML: 11
  • Total: 280
  • BibTeX: 3
  • EndNote: 3
Views and downloads (calculated since 28 Jul 2021)
Cumulative views and downloads (calculated since 28 Jul 2021)

Viewed (geographical distribution)

Total article views: 280 (including HTML, PDF, and XML) Thereof 280 with geography defined and 0 with unknown origin.
Country # Views %
  • 1
1
 
 
 
 
Latest update: 23 Sep 2021
Download
Short summary
Gap-filling usually accounts for a large source of uncertainties in the annual CO2 fluxes, though gap-filling CO2 fluxes is challenging at dryland sites due to small fluxes. Using data collected from a semi-arid site, four machine learning methods are evaluated with different lengths of artificial gaps. The artificial neural network and random forest methods outperform the other methods. With these methods, uncertainties in the annual CO2 flux at this site are estimated to be within 16 g C m−2.
Altmetrics