Abstract: To improve climate change impact estimates and to quantify their uncertainty, multi-model ensembles (MMEs) have been suggested. Model improvements can improve the accuracy of simulations and reduce the uncertainty of climate change impact assessments. Furthermore, they can reduce the number of models needed in a MME. Herein, 15 wheat growth models of a larger MME were improved through re-parameterization and/or incorporating or modifying heat stress effects on phenology, leaf growth and senescence, biomass growth, and grain number and size using detailed field experimental data from the USDA Hot Serial Cereal experiment (calibration data set). Simulation results from before and after model improvement were then evaluated with independent field experiments from a CIMMYT world-wide field trial network (evaluation data set). Model improvements decreased the variation (10th to 90th model ensemble percentile range) of grain yields simulated by the MME on average by 39% in the calibration data set and by 26% in the independent evaluation data set for crops grown in mean seasonal temperatures >24 °C. MME mean squared error in simulating grain yield decreased by 37%. A reduction in MME uncertainty range by 27% increased MME prediction skills by 47%. Results suggest that the mean level of variation observed in field experiments and used as a benchmark can be reached with half the number of models in the MME. Improving crop models is therefore important to increase the certainty of model-based impact assessments and allow more practical, i.e. smaller MMEs to be used effectively.

Abstract: Eleven widely used crop simulation models (APSIM, CERES, CROPSYST, COUP, DAISY, EPIC, FASSET, HERMES, MONICA, STICS and WOFOST) were tested using spring barley (Hordeum vulgare L.) data set under varying nitrogen (N) fertilizer rates from three experimental years in the boreal climate of Jokioinen, Finland. This is the largest standardized crop model inter-comparison under different levels of N supply to date. The models were calibrated using data from 2002 and 2008, of which 2008 included six N rates ranging from 0 to 150 kg N/ha. Calibration data consisted of weather, soil, phenology, leaf area index (LAI) and yield observations. The models were then tested against new data for 2009 and their performance was assessed and compared with both the two calibration years and the test year. For the calibration period, root mean square error between measurements and simulated grain dry matter yields ranged from 170 to 870 kg/ha. During the test year 2009, most models failed to accurately reproduce the observed low yield without N fertilizer as well as the steep yield response to N applications. The multi-model predictions were closer to observations than most single-model predictions, but multi-model mean could not correct systematic errors in model simulations. Variation in soil N mineralization and LAI development due to differences in weather not captured by the models most likely was the main reason for their unsatisfactory performance. This suggests the need for model improvement in soil N mineralization as a function of soil temperature and moisture. Furthermore, specific weather event impacts such as low temperatures after emergence in 2009, tending to enhance tillering, and a high precipitation event just before harvest in 2008, causing possible yield penalties, were not captured by any of the models compared in the current study.