No abstract

No abstract

1 15 ABSTRACT 16 Plant phenology, which describes the timing of plant development, is a major aspect of 17 plant response to environment and for crops, a major determinant of yield. Many studies have 18 focused on comparing model equations for describing how phenology responds to climate but 19 the effect of crop model calibration, also important for determining model performance, has 20 received much less attention. The objectives here were to obtain a rigorous evaluation of 21 prediction capability of wheat phenology models, to analyze the role of calibration and to 22 document the various calibration approaches. The 27 participants in this multi-model study 23were provided experimental data for calibration and asked to submit predictions for sites and 24 years not represented in those data. Participants were instructed to use and document their 25 "usual" calibration approach. Overall, the models provided quite good predictions of 26 phenology (median of mean absolute error of 6.1 days) and did much better than simply using 27 the average of observed values as predictor. The results suggest that calibration can 28 compensate to some extent for different model formulations, specifically for differences in 29 simulated time to emergence and differences in the choice of input variables. Conversely, 30 different calibration approaches were associated with major differences in prediction error 31 between the same models used by different groups. Given the large diversity of calibration 32 approaches and the importance of calibration, there is a clear need for guidelines and tools to 33 aid with calibration. Arguably the most important and difficult choice for calibration is the 34 choice of parameters to estimate. Several recommendations for calibration practices are 35 proposed. Model applications, including model studies of climate change impact, should 36 focus more on the data used for calibration and on the calibration methods employed. 37

Climate change, with increased temperatures and varied rainfall, poses a great challenge to food security around the world. Appropriately assessing the impacts of climate change on crop productivity and understanding the adaptation potential of agriculture to climate change are urgently needed to help develop effective strategies for future agriculture and to maintain food security. In this study, we studied future maize yield changes under 1.5°C (2018-2037) and 2°C (2044-2063) warming scenarios and investigated the adaptation potential across China's Maize Belt by optimizing the sowing date and cultivar using the APSIM-Maize model. In comparison to the baseline scenario, under the 1.5°C and 2°C warming scenarios, we found that without adaptation, maize yields would increase in the relatively cool regions with a single-cropping system but decrease in other regions. However, in comparison with the baseline scenario, under the 1.5°C and 2°C warming scenarios with adaptation, maize yields would increase by 11.1%-53.9% across the study area. Across the maize belt, compared with the baseline scenario, under warming of 1.5°C, the potential sowing window would increase by 2-17 d, and under warming of 2°C, this sowing window would increase by 4-26 d. The optimal sowing window would also be significantly extended in the regions with single-cropping systems by an average of 10 d under the 1.5°C warming scenario and 12 d under the 2°C warming scenario. Latematuring cultivar achieved higher yield than early-middle maturing cultivars in all regions except the north part of Northeast China. Adjusting the sowing date by increasing growth-period precipitation contributed more (44.5%-96.7%) to yield improvements than shifting cultivars (0%-50.8%) and climate change (−53.1% to 23.0%) across all maize planting regions except in the wet southwestern parts of the maize belt. The differences among the maize planting regions in terms of high adaptation potential provide invaluable information for policymakers and stakeholders of maize production to set out optimized agricultural strategies to safeguard the supply of maize.

Predicting wheat phenology is important for cultivar selection, for effective crop management and provides a baseline for evaluating the effects of global change. Evaluating how well crop phenology can be predicted is therefore of major interest. Twenty-eight wheat modeling groups participated in this evaluation. Our target population was wheat fields in the major wheat growing regions of Australia under current climatic conditions and with current local management practices. The environments used for calibration and for evaluation were both sampled from this same target population. The calibration and evaluation environments had neither sites nor years in common, so this is a rigorous evaluation of the ability of modeling groups to predict phenology for new sites and weather conditions. Mean absolute error (MAE) for the evaluation environments, averaged over predictions of three phenological stages and over modeling groups, was 9 days, with a range from 6 to 20 days. Predictions using the multimodeling group mean and median had prediction errors nearly as small as the best modeling group. About two thirds of the modeling groups performed better than a simple but relevant benchmark, which predicts phenology by assuming a constant temperature sum for each development stage. The added complexity of crop models beyond just the effect of temperature was thus justified in most cases. There was substantial variability between modeling groups using the same model structure, which implies that model improvement could be achieved not only by improving model structure, but also by improving parameter values, and in particular by improving calibration techniques.

57Predicting wheat phenology is important for cultivar selection, for effective crop 58 management and provides a baseline for evaluating the effects of global change. Evaluating 59 how well crop phenology can be predicted is therefore of major interest. Twenty-eight wheat 60 modeling groups participated in this evaluation. Model predictions depend not only on model 61 structure but also on the parameter values. This study is thus an evaluation of modeling groups, 62 which choose the structure and fix or estimate the parameters, rather than an evaluation just of 63 model structures. Our target population was wheat fields in the major wheat growing regions 64 of Australia under current climatic conditions and with current local management practices. 65The environments used for calibration and for evaluation were both sampled from this same 66 target population. The calibration and evaluation environments had neither sites nor years in 67 common, so this is a rigorous evaluation of the ability of modeling groups to predict phenology 68 for new sites and weather conditions. Mean absolute error (MAE) for the evaluation 69 environments, averaged over predictions of three phenological stages and over modeling 70 groups, was 9 days, with a range from 6 to 20 days. Predictions using the multi-modeling group 71 mean and median had prediction errors nearly as small as the best modeling group. For a given 72 modeling group, MAE for the evaluation environments was significantly correlated with MAE 73 for the calibration environments, which suggests that it would be of interest to test ensemble 74 predictors that weight individual modeling groups based on performance for the calibration 75 data. About two thirds of the modeling groups performed better than a simple but relevant 76 benchmark, which predicts phenology by assuming a constant temperature sum for each 77 development stage. The added complexity of crop models beyond just the effect of temperature 78 was thus justified in most cases. Finally, there was substantial variability between modeling 79 groups using the same model structure, which implies that model improvement could be 80 4 achieved not only by improving model structure, but also by improving parameter values, and 81 in particular by improving calibration techniques. 82 uncertainty 84 85 5 A second aspect of evaluation that must be specified is the modeling group or groups 131 that are being evaluated. We reserve the term "model" specifically for model structure, i.e. the 132 model equations, while modeling group determines both the model structure and the parameter 133 values, which are chosen or estimated by the group running the model. It is clear that predictions 134

No abstract

No abstract