Multimodel ensembles of wheat growth: many models are better than one. Martre, P.; Wallach, D.; Asseng, S.; Ewert, F.; Jones, J.; Rotter, R.; Boote, K.; Ruane, A.; Thorburn, P.; Cammarano, D.; Hatfield, J.; Rosenzweig, C.; Aggarwal, P.; Angulo, C.; Basso, B.; Bertuzzi, P.; Biernath, C.; Brisson, N.; Challinor, A.; Doltra, J.; Gayler, S.; Goldberg, R.; Grant, R.; Heng, L.; Hooker, J.; Hunt, L.; Ingwersen, J.; Izaurralde, R.; Kersebaum, K.; Muller, C.; Kumar, S.; Nendel, C.; O’leary, G.; Olesen, J.; Osborne, T.; Palosuo, T.; Priesack, E.; Ripoche, D.; Semenov, M.; Shcherbak, I.; Steduto, P.; Stockle, C.; Stratonovitch, P.; Streck, T.; Supit, I.; Tao, F.; Travasso, M.; Waha, K.; White, J.; and Wolf, J. Global Change Biology, 21(2):911–925, 2015.
doi  abstract   bibtex   
Crop models of crop growth are increasingly used to quantify the impact of global changes due to climate or crop management. Therefore, accuracy of simulation results is a major concern. Studies with ensembles of crop models can give valuable information about model accuracy and uncertainty, but such studies are difficult to organize and have only recently begun. We report on the largest ensemble study to date, of 27 wheat models tested in four contrasting locations for their accuracy in simulating multiple crop growth and yield variables. The relative error averaged over models was 24-38% for the different end-of-season variables including grain yield (GY) and grain protein concentration (GPC). There was little relation between error of a model for GY or GPC and error for in-season variables. Thus, most models did not arrive at accurate simulations of GY and GPC by accurately simulating preceding growth dynamics. Ensemble simulations, taking either the mean (e-mean) or median (e-median) of simulated values, gave better estimates than any individual model when all variables were considered. Compared to individual models, e-median ranked first in simulating measured GY and third in GPC. The error of e-mean and e-median declined with an increasing number of ensemble members, with little decrease beyond 10 models. We conclude that multimodel ensembles can be used to create new estimators with improved accuracy and consistency in simulating growth dynamics. We argue that these results are applicable to other crop species, and hypothesize that they apply more generally to ecological system models.
@article{ Martre2015a,
  author = {Martre, P. and Wallach, D. and Asseng, S. and Ewert, F. and Jones, J.W. and Rotter, R.P. and Boote, K.J. and Ruane, A.C. and Thorburn, P.J. and Cammarano, D. and Hatfield, J.L. and Rosenzweig, C. and Aggarwal, P.K. and Angulo, C. and Basso, B. and Bertuzzi, P. and Biernath, C. and Brisson, N. and Challinor, A.J. and Doltra, J. and Gayler, S. and Goldberg, R. and Grant, R.F. and Heng, L. and Hooker, J. and Hunt, L.A. and Ingwersen, J. and Izaurralde, R.C. and Kersebaum, K.C. and Muller, C. and Kumar, S.N. and Nendel, C. and O’leary, G. and Olesen, J.E. and Osborne, T.M. and Palosuo, T. and Priesack, E. and Ripoche, D. and Semenov, M.A. and Shcherbak, I. and Steduto, P. and Stockle, C.O. and Stratonovitch, P. and Streck, T. and Supit, I. and Tao, F. and Travasso, M. and Waha, K. and White, J.W. and Wolf, J.},
  title = {Multimodel ensembles of wheat growth: many models are better than one.},
  journal = {Global Change Biology},
  volume = {21},
  number = {2},
  pages = {911–925},
  year = {2015},
  doi = {10.1111/gcb.12768},
  abstract = {Crop models of crop growth are increasingly used to quantify the impact of global changes due to climate or crop management. Therefore, accuracy of simulation results is a major concern. Studies with ensembles of crop models can give valuable information about model accuracy and uncertainty, but such studies are difficult to organize and have only recently begun. We report on the largest ensemble study to date, of 27 wheat models tested in four contrasting locations for their accuracy in simulating multiple crop growth and yield variables. The relative error averaged over models was 24-38% for the different end-of-season variables including grain yield (GY) and grain protein concentration (GPC). There was little relation between error of a model for GY or GPC and error for in-season variables. Thus, most models did not arrive at accurate simulations of GY and GPC by accurately simulating preceding growth dynamics. Ensemble simulations, taking either the mean (e-mean) or median (e-median) of simulated values, gave better estimates than any individual model when all variables were considered. Compared to individual models, e-median ranked first in simulating measured GY and third in GPC. The error of e-mean and e-median declined with an increasing number of ensemble members, with little decrease beyond 10 models. We conclude that multimodel ensembles can be used to create new estimators with improved accuracy and consistency in simulating growth dynamics. We argue that these results are applicable to other crop species, and hypothesize that they apply more generally to ecological system models.},
  `keywords = {`k`}
}
Downloads: 0