Residual correlation and ensemble modelling to improve crop and grassland models

Renáta Sándor* (Corresponding Author), Fiona Ehrhardt, Peter Grace, Sylvie Recous, Pete Smith, Val Snow, Jean François Soussana, Bruno Basso, Arti Bhatia, Lorenzo Brilli, Jordi Doltra, Christopher D. Dorich, Luca Doro, Nuala Fitton, Brian Grant, Matthew Tom Harrison, Ute Skiba, Miko U.F. Kirschbaum, Katja Klumpp, Patricia LavilleJoel Léonard, Raphaël Martin, Raia Silvia Massad, Andrew D. Moore, Vasileios Myrgiotis, Elizabeth Pattey, Susanne Rolinski, Joanna Sharp, Ward Smith, Lianhai Wu, Qing Zhang, Gianni Bellocchi

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

3 Citations (Scopus)

Abstract

Multi-model ensembles are becoming increasingly accepted for the estimation of agricultural carbon-nitrogen fluxes, productivity and sustainability. There is mounting evidence that with some site-specific observations available for model calibration (with vegetation data as a minimum requirement), median outputs assimilated from biogeochemical models (multi-model medians) provide more accurate simulations than individual models. Here, we evaluate potential deficiencies in how model ensembles represent (in relation to climatic factors) the processes underlying biogeochemical outputs in complex agricultural systems such as grassland and crop rotations including fallow periods. We do that by exploring the correlation of model residuals. We restricted the distinction between partial and full calibration to the two most relevant calibration stages, i.e. with plant data only (partial) and with a combination of plant, soil physical and biogeochemical data (full). It introduces and evaluates the trade-off between (1) what is practical to apply for model users and beneficiaries, and (2) what constitutes best modelling practice. The lower correlations obtained overall with fully calibrated models highlight the centrality of the full calibration scenario for identifying areas of model structures that require further development.

Original languageEnglish
Article number105625
Number of pages12
JournalEnvironmental Modelling and Software
Volume161
Early online date24 Jan 2023
DOIs
Publication statusPublished - 1 Mar 2023

Bibliographical note

Funding Information:
This study was coordinated by the Integrative Research Group of the Global Research Alliance (GRA) on agricultural GHGs and was supported by five research projects (CN-MIP, Models4Pastures, MACSUR, COMET-Global and MAGGNET), which received funding by a multi-partner call on agricultural greenhouse gas research of the Joint Programming Initiative ‘FACCE’ through its national financing bodies. It falls within the thematic area of the French government IDEX-ISITE initiative (reference: 16-IDEX-0001; project CAP 20–25). We acknowledge funding for the data collection through the EU projects GREENGRASS (EC EVK2-CT2001-00105), CarboEurope (GOCE-CT-2003-505572) and NitroEurope (017841). US acknowledges SRUC's contribution (Stephanie K. Jones and Robert M. Rees) to compile the data of the C4 grassland site (Easter Bush, UK). The research in support of C1(Ottawa, ON, Canada) site data acquisition was conducted with the financial support of Agriculture and Agri-Food Canada A-base funding. Data for the C2 cropland site (Grignon, France) were obtained from the Fr-Gri ecosystem site ICOS (Integrated Carbon Observation System; https://www.icos-cp.eu), for which we thank Pauline Buysse and Benjamin Loubet (INRAE, Grignon) for access. Data for the G3 grassland site (Laqueuille, France) were obtained from the FR-Lq1 SOERE-ACBB (Système D'observation Et D'expérimentation Sur Le Long Terme Pour La Recherche En Environnement - Agro-Écosystème, Cycle Bio-Géochimique Et Biodiversité; https://www.soere-acbb.com) ecosystem site (ICOS) financed by French National Agency for Research (ANAEE-F, ANR-11-INBS-0001). SR (PIK) acknowledges financial support from the BMBF (Federal Ministry of Education and Research of Germany) for funding of the projects MACMIT (grant 01LN1317A) and Climasteppe (grant 01DJ18012). RS and GB received mobility funding from the French-Hungarian bilateral partnership through the BALATON (N° 44703 TF)/TéT (2019–2.1.11-TÉT-2019-00031) programme.

Data Availability Statement

Supplementary data to this article can be found online at https://doi.org/10.1016/j.envsoft.2023.105625

Keywords

  • Biogeochemical models
  • Correlation matrices
  • Ensemble modelling
  • Model calibration
  • Residual plot analysis

Fingerprint

Dive into the research topics of 'Residual correlation and ensemble modelling to improve crop and grassland models'. Together they form a unique fingerprint.

Cite this