Multimodel ensembles of wheat growth: many models are better than one

Authors: MARTRE, PIERRE - French National Institute For Agricultural Research; WALLACH, DANIEL - French National Institute For Agricultural Research; ASSENG, SENTHOLD - University Of Florida; EWERT, FRANK; ROSENZWEIG, CYNTHIA - National Aeronautics And Space Administration (NASA); JONES, JAMES - University Of Florida; Hatfield, Jerry; RUANE, ALEX - National Aeronautics And Space Administration (NASA); BOOTE, KENNETH - University Of Florida; THORBURN, PETER - Commonwealth Scientific And Industrial Research Organisation (CSIRO); ROTTER, REIMUND - Mtt Agrifood Research Finland; CAMMARANO, DAVIDE - University Of Florida; AGGARWAL, PRAMOD - International Water Management Institute; ANGULO, CARLOS; BASSO, BRUNO - Michigan State University; BERTUZZI, PATRICK - French National Institute For Agricultural Research; BIERNATH, CHRISTIAN - German Research Center For Environmental Health; BRISSON, NADINE - French National Institute For Agricultural Research; CHALLINOR, ANDREW - University Of Leeds; DOLTRA, JORDI - Center For Agricultural Research And Training, Cantabria Government (CIFA); GAYLOR, SEBASTIAN - University Of Tubingen; GOLDBERG, RICHIE - National Aeronautics And Space Administration (NASA); GRANT, ROBERT - University Of Alberta; HENG, LEE - International Atomic Energy Agency (IAEA); HOOKER, JOHN - University Of Reading; HUNT, LESLIE - University Of Geulph; INGWERSEN, JOACHIM - University Of Hohenheim; IZAURRALDE, ROBERTO - Global Change Research Institute; CHRISTIAN, KURT - Leibniz Centre; MULLER, CHRISTOPH - Potsdam Institute; KUMAR, SOORA - Indian Agricultural Research Institute; NENDEL, CLAAS - Leibniz Centre; O'LEARU, GARRY - Department Of Primary Industries; OLESEN, JORGEN - University Of Aarhus; OSBORNE, TOM - University Of Reading; PALOSUO, TARU - Mtt Agrifood Research Finland; PRIESACK, ECKART - German Research Center For Environmental Health; RIPOCHE, DOMINIQUE - French National Institute For Agricultural Research; SEMENOV, MIKHAIL - Rothamsted Research; SHCHERBAK, IURII - Michigan State University; STEDUTO, PASQUALE - Food And Agriculture Organization Of The United Nations (FAO); STOCKLE, CLAUDIO - Washington State University; STRATONOVITCH, PIERRE - Rothamsted Research; STRECK, THILO - University Of Hohenheim; SUPIT, IWAN - Wageningen University; TAO, FULU - Chinese Academy Of Sciences; TRAVASSO, MARIA - National Institute Of Agronomic Research Of Morocco (INRA); WAHA, KATHARINA - Potsdam Institute; White, Jeffrey; WOLF, JOOST - Wageningen University

Submitted to: Global Change Biology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 9/25/2014
Publication Date: 12/3/2014
Citation: Martre, P., Wallach, D., Asseng, S., Ewert, F., Rosenzweig, C., Jones, J., Hatfield, J.L., Ruane, A.C., Boote, K.J., Thorburn, P.J., Rotter, R.P., Cammarano, D., Aggarwal, P.K., Angulo, C., Basso, B., Bertuzzi, P., Biernath, C., Brisson, N., Challinor, A.J., Doltra, J., Gaylor, S., Goldberg, R., Grant, R., Heng, L., Hooker, J.E., Hunt, L., Ingwersen, J., Izaurralde, R.C., Christian, K., Muller, C., Kumar, S.N., Nendel, C., O'Learu, G., Olesen, J.E., Osborne, T.M., Palosuo, T., Priesack, E., Ripoche, D., Semenov, M.A., Shcherbak, I., Steduto, P., Stockle, C.O., Stratonovitch, P., Streck, T., Supit, I., Tao, F., Travasso, M., Waha, K., White, J.W., Wolf, J. 2014. Multimodel ensembles of wheat growth: many models are better than one. Global Change Biology. 21:911-925.

Interpretive Summary: Computer-based simulation models of crop growth are increasingly used to forecast possible impacts of global changes, considering effects climate, crop management and other factors. This information is used to guide research and policy decisions with broad implications for future agriculture. Accuracy of such simulations thus is a major concern. One approach for overcoming the inaccuracies of individual models is to base decisions on multiple models, representing a “multimodel ensemble.” Studies of ensembles can give valuable information about model accuracy and uncertainty, but such studies are difficult to organize and have only recently begun. This paper describes the largest crop modeling ensemble study to date, which involved 27 wheat models tested in four contrasting locations. 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. Ensemble simulations, considering either the mean (e-mean) or median (e-median) of the simulated values from individual models, gave better estimates than any individual model. The error of the ensembles declined as the number of included models increased up to about 10 models. We conclude that multimodel ensembles can provide more accurate predictions of crop growth and yield. These results appear applicable to other crop species and more generally to other types of ecological models. The results open important new opportunities for improving our understanding of agriculture may be affected by a changing and uncertain climate.

Technical 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.