ABSTRACT

A comparison was made between the recently used crop models (DSSAT-CERES and AquaCrop), based on high accuracy, to simulate wheat growth under different water and nitrogen levels. Hence, four irrigation treatments (rainfed, irrigate at 50% and 75% of field capacity (FC), and irrigate at 100% of FC) were considered as the main factor with three nitrogen fertilizer levels as the sub-main factor (viz. non-fertilizer, 100 kg N ha⁻¹, and 200 kg N ha⁻¹) during 2014–2016. Model efficiency (E), Wilmott’s index of agreement (d), Root Mean Square Error (RMSE), and Normalized Root Mean Square Error (NRMSE) were used to test the model performance. AquaCrop model was calibrated to simulate the grain and biomass yield with the prediction error statistics of 0.87 < E < 0.90, 0.24 < RMSE < 0.36 t ha⁻¹, 6 < NRMSE < 7.2%, and 0.90 < d < 0.93, respectively. Subsequently, the CERES model was calibrated and the performance was in line (0.88 < E < 0.93, 0.92 < d < 0.96, 0.19 < RMSE < 0.34 t ha⁻¹, and 5.7 < NRMSE < 5.8%), respectively. Upon validation, E was 0.89 and 0.91; d was 0.93 and 0.96, RMSE was 0.25 and 0.21 t ha⁻¹, and NRMSE was 6.4 and 5.9 for grain yield by AquCrop and CERES models for all treatments, respectively. Overall, the simulation results by the DSSAT model were relatively more accurate than the AquaCrop model. But, considering the low required input data, the AquaCrop model is recommended for wheat yield simulation in the conditions of unavailability of equipment and crop data.

摘要

基于高精度，对最近使用的作物模型（DSSAT-CERES 和 AquaCrop）进行了比较，以模拟不同水和氮水平下的小麦生长。因此，四种灌溉处理（雨养、田间持水量 (FC) 的 50% 和 75% 的灌溉以及 FC 的 100% 的灌溉）被认为是主要因素，三个氮肥水平作为次主要因素（即。2014-2016 年期间不施肥，100 kg N ha ^-1和 200 kg N ha ^-1 )。模型效率 (E)、威尔莫特一致性指数 (d)、均方根误差 (RMSE) 和归一化均方根误差 (NRMSE) 用于测试模型性能。AquaCrop 模型经过校准以模拟谷物和生物量产量，预测误差统计为 0.87 < E < 0.90、0.24 < RMSE < 0。^-1，6 <NRMSE <7.2％，和0.90 <d <0.93，分别。随后，CERES 模型被校准并且性能分别符合（0.88 < E < 0.93、0.92 < d < 0.96、0.19 < RMSE < 0.34 t ha ^-1和5.7 < NRMSE < 5.8%）。经验证，E 分别为 0.89 和 0.91；对于所有处理的 AquCrop 和 CERES 模型，d 分别为 0.93 和 0.96，RMSE 为 0.25 和 0.21 t ha ^-1，NRMSE 分别为 6.4 和 5.9。总体而言，DSSAT 模型的模拟结果相对比 AquaCrop 模型更准确。但是，考虑到所需的输入数据较少，在设备和作物数据不可用的情况下，建议使用 AquaCrop 模型进行小麦产量模拟。