基于APSIM-Maize模型的石羊河流域春玉米适宜种植区产量模拟及参数优化

Optimization of Yield Parameters for APSIM-Maize Model in Suitable Planting Areas for Spring Maize in the Shiyang River Basin

为提高APSIM-Maize模型在西北干旱条件下对春玉米产量模拟的精准性,以甘肃省石羊河流域春玉米适宜种植区(永昌县、民勤县、凉州区、古浪县)为研究区域,通过查阅与研究地区春玉米相关的49篇大田试验文献获得2009-2022年共115对观测数据,基于APSIM-Maize模型,采用敏感性指数SI法和修正的Morris法对水分利用率、光周期斜率等21个模型参数进行敏感性分析,针对4个适宜种植区分别对筛选的敏感参数进行优化验证。结果表明:(1)对春玉米产量较敏感的模型参数有5个,敏感大小分别为蒸腾效率系数>出苗到拔节结束积温>光合作用与辐射利用效率>开花至开始灌浆积温>茎向籽粒转移的生物量;(2)针对不同适宜种植区,参数优化效果有所差异,其中永昌县、凉州区及古浪县效果较好,民勤县的优化效果次之;(3)春玉米适宜种植区参数优化后的产量实测值与模拟值之间的相关系数分别由0.566提高到0.978、0.341提高到0.809、0.455提高到0.953、0.537提高到0.936,均方根误差分别由1837.10~3088.72 kg/hm^(2)减小到341.64~996.64 kg/hm^(2),优化后的参数有效的提高了模型模拟产量的精度。

In order to improve the accuracy of the APSIM-Maize model in simulating spring maize yield under drought conditions in northwest China,this study selected the suitable planting areas for spring maize in the Shiyang River Basin of Gansu Province(Yongchang County,Minqin County,Liangzhou District,Gulang County)as the research area.A total of 115 pairs of observation data from 2009 to 2022 were obtained by reviewing and researching 49 field experimental literature related to spring maize in the research area.Based on the APSIMMaize model,sensitivity analysis was conducted on 21 model parameters such as water use efficiency and photoperiod slope using sensitivity index SI method and modified Morris method.The selected sensitive parameters were optimized and verified for each of the four suitable planting areas.The results showed,①that there were 5 model parameters that were sensitive to spring maize yield,with sensitivity values as follows:transpiration efficiency coefficient>accumulated temperature from emergence to elongation>photosynthesis and radiation utilization efficiency>accumulated temperature from flowering to beginning of grain filling>biomass transferred from stem to grain;②The optimization effect of parameters varies for different suitable planting areas,with Yongchang County,Liangzhou District,and Gulang County showing better results,followed by Minqin County;③The correlation coefficients between the measured and simulated yield values in the optimized planting area parameters for spring corn increased from 0.566 to 0.978,0.341 to 0.809,0.455 to 0.953,0.537 to 0.936,and the root mean square error decreased from 1837.10~3088.72 kg/hm^(2)to 341.64~996.64 kg/hm^(2),respectively.The optimized parameters effectively improved the accuracy of the model's simulated yield.