Assessing uncertainties in crop and pasture ensemble model simulations of productivity and N2O emissions

F Ehrhardt, J-F Soussana, G Bellocchi, P Grace, R McAullife, S Recous, R Sandor, P Smith, V Snow, MDA Migliorati, B Basso, A Bhatia, L Brilli, J Doltra, CD Dorich, L Doro, N Fitton, SJ Giacomini, B Grant, MT HarrisonSK Jones, MUF Kirschbaum, K Klumpp, P Laville, J Leonard, M Liebig, M Lieffering, R Martin, RS Massad, E Meier, L Merbold, AD Moore, V Myrgiotis, P Newton, E Pattey, S Rolinski, J Sharp, WN Smith, L Wu, Q Zhang

Research output: Contribution to journalArticle

33 Citations (Scopus)
5 Downloads (Pure)

Abstract

Simulation models are extensively used to predict agricultural productivity and greenhouse gas (GHG) emissions. However, the uncertainties of (reduced) model ensemble simulations have not been assessed systematically for variables affecting food security and climate change mitigation, within multispecies agricultural contexts. We report an international model comparison and benchmarking exercise, showing the potential of multimodel ensembles to predict productivity and nitrous oxide (N2O) emissions for wheat, maize, rice and temperate grasslands. Using a multistage modelling protocol, from blind simulations (stage 1) to partial (stages 2-4) and full calibration (stage 5), 24 process-based biogeochemical models were assessed individually or as an ensemble against long-term experimental data from four temperate grassland and five arable crop rotation sites spanning four continents. Comparisons were performed by reference to the experimental uncertainties of observed yields and N2O emissions. Results showed that across sites and crop/grassland types, 23 to 40% of the uncalibrated individual models were within two standard deviations (s.d.) of observed yields, while 42 (rice) to 96% (grasslands) of the models were within one s.d. of observed N2O emissions. At stage 1, ensembles formed by the three lowest prediction model errors (RRMSE) predicted both yields and N2O emissions within experimental uncertainties for 44 and 33% of the crop and grassland growth cycles, respectively. Partial model calibration (stages 2 to 4) markedly reduced prediction errors of the full model ensemble E-median for crop grain yields (from 36% at stage 1 down to 4% on average) and grassland productivity (from 44 to 27%) and to a lesser and more variable extent for N2O emissions. Yield-scaled N2O emissions (N2O emissions divided by crop yields) were ranked accurately by 3-model ensembles across crop species and field sites. The potential of using process-based model ensembles to predict jointly productivity and N2O emissions at field scale is discussed.
Original languageEnglish
Pages (from-to)e603 - e616
JournalGlobal Change Biology
Volume24
Issue number2
Early online date28 Oct 2017
DOIs
Publication statusFirst published - 28 Oct 2017

Fingerprint

pasture
productivity
crop
simulation
grassland
rice
calibration
benchmarking
crop rotation
prediction
food security
nitrous oxide
crop yield
cereal
greenhouse gas
wheat
maize

Bibliographical note

1028529

Keywords

  • Agriculture
  • Benchmarking
  • Biogeochemical models
  • Climate change
  • Greenhouse gases
  • Nitrous oxide
  • Soil
  • Yield

Cite this

Ehrhardt, F., Soussana, J-F., Bellocchi, G., Grace, P., McAullife, R., Recous, S., ... Zhang, Q. (2017). Assessing uncertainties in crop and pasture ensemble model simulations of productivity and N2O emissions. Global Change Biology, 24(2), e603 - e616. https://doi.org/10.1111/gcb.13965
Ehrhardt, F ; Soussana, J-F ; Bellocchi, G ; Grace, P ; McAullife, R ; Recous, S ; Sandor, R ; Smith, P ; Snow, V ; Migliorati, MDA ; Basso, B ; Bhatia, A ; Brilli, L ; Doltra, J ; Dorich, CD ; Doro, L ; Fitton, N ; Giacomini, SJ ; Grant, B ; Harrison, MT ; Jones, SK ; Kirschbaum, MUF ; Klumpp, K ; Laville, P ; Leonard, J ; Liebig, M ; Lieffering, M ; Martin, R ; Massad, RS ; Meier, E ; Merbold, L ; Moore, AD ; Myrgiotis, V ; Newton, P ; Pattey, E ; Rolinski, S ; Sharp, J ; Smith, WN ; Wu, L ; Zhang, Q. / Assessing uncertainties in crop and pasture ensemble model simulations of productivity and N2O emissions. In: Global Change Biology. 2017 ; Vol. 24, No. 2. pp. e603 - e616.
@article{73fc43ba1f8e430f9dc338d293834a88,
title = "Assessing uncertainties in crop and pasture ensemble model simulations of productivity and N2O emissions",
abstract = "Simulation models are extensively used to predict agricultural productivity and greenhouse gas (GHG) emissions. However, the uncertainties of (reduced) model ensemble simulations have not been assessed systematically for variables affecting food security and climate change mitigation, within multispecies agricultural contexts. We report an international model comparison and benchmarking exercise, showing the potential of multimodel ensembles to predict productivity and nitrous oxide (N2O) emissions for wheat, maize, rice and temperate grasslands. Using a multistage modelling protocol, from blind simulations (stage 1) to partial (stages 2-4) and full calibration (stage 5), 24 process-based biogeochemical models were assessed individually or as an ensemble against long-term experimental data from four temperate grassland and five arable crop rotation sites spanning four continents. Comparisons were performed by reference to the experimental uncertainties of observed yields and N2O emissions. Results showed that across sites and crop/grassland types, 23 to 40{\%} of the uncalibrated individual models were within two standard deviations (s.d.) of observed yields, while 42 (rice) to 96{\%} (grasslands) of the models were within one s.d. of observed N2O emissions. At stage 1, ensembles formed by the three lowest prediction model errors (RRMSE) predicted both yields and N2O emissions within experimental uncertainties for 44 and 33{\%} of the crop and grassland growth cycles, respectively. Partial model calibration (stages 2 to 4) markedly reduced prediction errors of the full model ensemble E-median for crop grain yields (from 36{\%} at stage 1 down to 4{\%} on average) and grassland productivity (from 44 to 27{\%}) and to a lesser and more variable extent for N2O emissions. Yield-scaled N2O emissions (N2O emissions divided by crop yields) were ranked accurately by 3-model ensembles across crop species and field sites. The potential of using process-based model ensembles to predict jointly productivity and N2O emissions at field scale is discussed.",
keywords = "Agriculture, Benchmarking, Biogeochemical models, Climate change, Greenhouse gases, Nitrous oxide, Soil, Yield",
author = "F Ehrhardt and J-F Soussana and G Bellocchi and P Grace and R McAullife and S Recous and R Sandor and P Smith and V Snow and MDA Migliorati and B Basso and A Bhatia and L Brilli and J Doltra and CD Dorich and L Doro and N Fitton and SJ Giacomini and B Grant and MT Harrison and SK Jones and MUF Kirschbaum and K Klumpp and P Laville and J Leonard and M Liebig and M Lieffering and R Martin and RS Massad and E Meier and L Merbold and AD Moore and V Myrgiotis and P Newton and E Pattey and S Rolinski and J Sharp and WN Smith and L Wu and Q Zhang",
note = "1028529",
year = "2017",
month = "10",
day = "28",
doi = "10.1111/gcb.13965",
language = "English",
volume = "24",
pages = "e603 -- e616",
journal = "Global Change Biology",
issn = "1354-1013",
publisher = "Wiley",
number = "2",

}

Ehrhardt, F, Soussana, J-F, Bellocchi, G, Grace, P, McAullife, R, Recous, S, Sandor, R, Smith, P, Snow, V, Migliorati, MDA, Basso, B, Bhatia, A, Brilli, L, Doltra, J, Dorich, CD, Doro, L, Fitton, N, Giacomini, SJ, Grant, B, Harrison, MT, Jones, SK, Kirschbaum, MUF, Klumpp, K, Laville, P, Leonard, J, Liebig, M, Lieffering, M, Martin, R, Massad, RS, Meier, E, Merbold, L, Moore, AD, Myrgiotis, V, Newton, P, Pattey, E, Rolinski, S, Sharp, J, Smith, WN, Wu, L & Zhang, Q 2017, 'Assessing uncertainties in crop and pasture ensemble model simulations of productivity and N2O emissions', Global Change Biology, vol. 24, no. 2, pp. e603 - e616. https://doi.org/10.1111/gcb.13965

Assessing uncertainties in crop and pasture ensemble model simulations of productivity and N2O emissions. / Ehrhardt, F; Soussana, J-F; Bellocchi, G; Grace, P; McAullife, R; Recous, S; Sandor, R; Smith, P; Snow, V; Migliorati, MDA; Basso, B; Bhatia, A; Brilli, L; Doltra, J; Dorich, CD; Doro, L; Fitton, N; Giacomini, SJ; Grant, B; Harrison, MT; Jones, SK; Kirschbaum, MUF; Klumpp, K; Laville, P; Leonard, J; Liebig, M; Lieffering, M; Martin, R; Massad, RS; Meier, E; Merbold, L; Moore, AD; Myrgiotis, V; Newton, P; Pattey, E; Rolinski, S; Sharp, J; Smith, WN; Wu, L; Zhang, Q.

In: Global Change Biology, Vol. 24, No. 2, 28.10.2017, p. e603 - e616.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Assessing uncertainties in crop and pasture ensemble model simulations of productivity and N2O emissions

AU - Ehrhardt, F

AU - Soussana, J-F

AU - Bellocchi, G

AU - Grace, P

AU - McAullife, R

AU - Recous, S

AU - Sandor, R

AU - Smith, P

AU - Snow, V

AU - Migliorati, MDA

AU - Basso, B

AU - Bhatia, A

AU - Brilli, L

AU - Doltra, J

AU - Dorich, CD

AU - Doro, L

AU - Fitton, N

AU - Giacomini, SJ

AU - Grant, B

AU - Harrison, MT

AU - Jones, SK

AU - Kirschbaum, MUF

AU - Klumpp, K

AU - Laville, P

AU - Leonard, J

AU - Liebig, M

AU - Lieffering, M

AU - Martin, R

AU - Massad, RS

AU - Meier, E

AU - Merbold, L

AU - Moore, AD

AU - Myrgiotis, V

AU - Newton, P

AU - Pattey, E

AU - Rolinski, S

AU - Sharp, J

AU - Smith, WN

AU - Wu, L

AU - Zhang, Q

N1 - 1028529

PY - 2017/10/28

Y1 - 2017/10/28

N2 - Simulation models are extensively used to predict agricultural productivity and greenhouse gas (GHG) emissions. However, the uncertainties of (reduced) model ensemble simulations have not been assessed systematically for variables affecting food security and climate change mitigation, within multispecies agricultural contexts. We report an international model comparison and benchmarking exercise, showing the potential of multimodel ensembles to predict productivity and nitrous oxide (N2O) emissions for wheat, maize, rice and temperate grasslands. Using a multistage modelling protocol, from blind simulations (stage 1) to partial (stages 2-4) and full calibration (stage 5), 24 process-based biogeochemical models were assessed individually or as an ensemble against long-term experimental data from four temperate grassland and five arable crop rotation sites spanning four continents. Comparisons were performed by reference to the experimental uncertainties of observed yields and N2O emissions. Results showed that across sites and crop/grassland types, 23 to 40% of the uncalibrated individual models were within two standard deviations (s.d.) of observed yields, while 42 (rice) to 96% (grasslands) of the models were within one s.d. of observed N2O emissions. At stage 1, ensembles formed by the three lowest prediction model errors (RRMSE) predicted both yields and N2O emissions within experimental uncertainties for 44 and 33% of the crop and grassland growth cycles, respectively. Partial model calibration (stages 2 to 4) markedly reduced prediction errors of the full model ensemble E-median for crop grain yields (from 36% at stage 1 down to 4% on average) and grassland productivity (from 44 to 27%) and to a lesser and more variable extent for N2O emissions. Yield-scaled N2O emissions (N2O emissions divided by crop yields) were ranked accurately by 3-model ensembles across crop species and field sites. The potential of using process-based model ensembles to predict jointly productivity and N2O emissions at field scale is discussed.

AB - Simulation models are extensively used to predict agricultural productivity and greenhouse gas (GHG) emissions. However, the uncertainties of (reduced) model ensemble simulations have not been assessed systematically for variables affecting food security and climate change mitigation, within multispecies agricultural contexts. We report an international model comparison and benchmarking exercise, showing the potential of multimodel ensembles to predict productivity and nitrous oxide (N2O) emissions for wheat, maize, rice and temperate grasslands. Using a multistage modelling protocol, from blind simulations (stage 1) to partial (stages 2-4) and full calibration (stage 5), 24 process-based biogeochemical models were assessed individually or as an ensemble against long-term experimental data from four temperate grassland and five arable crop rotation sites spanning four continents. Comparisons were performed by reference to the experimental uncertainties of observed yields and N2O emissions. Results showed that across sites and crop/grassland types, 23 to 40% of the uncalibrated individual models were within two standard deviations (s.d.) of observed yields, while 42 (rice) to 96% (grasslands) of the models were within one s.d. of observed N2O emissions. At stage 1, ensembles formed by the three lowest prediction model errors (RRMSE) predicted both yields and N2O emissions within experimental uncertainties for 44 and 33% of the crop and grassland growth cycles, respectively. Partial model calibration (stages 2 to 4) markedly reduced prediction errors of the full model ensemble E-median for crop grain yields (from 36% at stage 1 down to 4% on average) and grassland productivity (from 44 to 27%) and to a lesser and more variable extent for N2O emissions. Yield-scaled N2O emissions (N2O emissions divided by crop yields) were ranked accurately by 3-model ensembles across crop species and field sites. The potential of using process-based model ensembles to predict jointly productivity and N2O emissions at field scale is discussed.

KW - Agriculture

KW - Benchmarking

KW - Biogeochemical models

KW - Climate change

KW - Greenhouse gases

KW - Nitrous oxide

KW - Soil

KW - Yield

U2 - 10.1111/gcb.13965

DO - 10.1111/gcb.13965

M3 - Article

VL - 24

SP - e603 - e616

JO - Global Change Biology

JF - Global Change Biology

SN - 1354-1013

IS - 2

ER -

Ehrhardt F, Soussana J-F, Bellocchi G, Grace P, McAullife R, Recous S et al. Assessing uncertainties in crop and pasture ensemble model simulations of productivity and N2O emissions. Global Change Biology. 2017 Oct 28;24(2):e603 - e616. https://doi.org/10.1111/gcb.13965