| 刘家辉,刘强,李广.不同施氮梯度下影响春小麦产量的参数敏感性分析及优化[J].干旱地区农业研究,2025,(2):253~261 |
| 不同施氮梯度下影响春小麦产量的参数敏感性分析及优化 |
| Analysis and optimization of the sensitivity of parameters affecting spring wheat yield under different nitrogen application gradients |
| |
| DOI：10.7606/j.issn.1000-7601.2025.02.25 |
| 中文关键词: 春小麦 施氮梯度 产量相关参数 敏感性分析 APSIM模型 甘肃省 |
| 英文关键词:spring wheat nitrogen application gradient yield\|related parameters sensitivity analysis APSIM model Gansu Province |
| 基金项目:国家自然科学基金（32360438）；甘肃省拔尖领军人才项目（GSBJLJ-2023-09）；甘肃省重点研究发展计划项目（22YF7FA116） |
|
| 摘要点击次数: 75 |
| 全文下载次数: 36 |
| 中文摘要: |
| 于甘肃省定西市安定区凤翔镇安家坡村，根据2005—2022年当地气象数据、2015—2018年旱地春小麦实测数据，结合该地区2005—2022年的统计年鉴数据，通过APSIM模型模拟小麦播种期不同的施氮量梯度(0、21、42、63、84、105、126 kg·hm-2)下25个模型参数，利用EFAST方法对参数进行敏感性分析。结果表明，不同施氮梯度下，对春小麦产量影响较大的敏感参数有11个，分别为灌浆期积温、灌浆到成熟期积温、拔节到开花期积温、单株最大籽粒质量、作物水分需求、出苗到拔节期积温、每克茎籽粒数量、遮阴导致老化的最大叶面积指数、单株质量、开花到灌浆期籽粒日潜在灌浆速率、消光系数，其中以消光系数和每克茎籽粒数量对产量的影响最为显著，其他参数在不同施氮梯度下对春小麦产量的敏感性顺序存在差异。使用马尔可夫链蒙特卡罗算法对这11个参数进行优化，其产量实测值与模拟值的均方根误差从194.60 kg·hm-2减少到了25.75 kg·hm-2，相对均方根误差从34.10%减少到了4.53%，模型决定系数从0.778提高到了0.943。利用EFAST敏感性分析方法结合马尔可夫链蒙特卡罗算法对影响春小麦产量的敏感性参数进行优化，提高了APSIM模型对春小麦产量模拟的精准度和可靠性。 |
| 英文摘要: |
| The study area is located in Anjiapo Village, Fengxiang Town, Anding District, Dingxi City, Gansu Province. Utilizing local meteorological data from 2005 to 2022, measured data on dryland spring wheat from 2015 to 2018, and regional statistical yearbook data from 2005 to 2022, the APSIM model was employed to simulate 25 model parameters under different nitrogen application gradients during the wheat sowing period (0, 21, 42, 63, 84, 105 kg·hm-2, and 126 kg·hm-2). The EFAST method was applied to analyze the sensitivity of these parameters. The results indicated that under different nitrogen application gradients, 11 parameters had a significant impact on spring wheat yield: accumulated temperature during the grain filling period, accumulated temperature from grain filling to maturity, accumulated temperature from stem elongation to flowering, maximum grain mass per plant, crop water demand, thermal time from emergence to jointing, number of grains per gram of stem, maximum leaf area index reduction due to shading, plant weight per plant, daily potential grain filling rate from flowering to grain filling, and the extinction coefficient. Among these, the extinction coefficient and the number of grains per gram of stem had the most significant impact on yield, while the sensitivity ranking of other parameters to spring wheat yield varied under different nitrogen application gradients. Subsequently, the Markov Chain Monte Carlo algorithm was used to optimize these 11 parameters. The optimized parameters significantly improved the accuracy of APSIM yield simulations. The root mean square error (RMSE) between the measured and simulated yield decreased from 194.6 kg·hm-2 to 25.75 kg·hm-2, while the relative RMSE dropped from 34.10% to 4.53%. Additionally, the model’s coefficient of determination (R2) improved from 0.778 to 0.943. This study employed the EFAST sensitivity analysis method in conjunction with the Markov Chain Monte Carlo algorithm to identify and optimize key parameters influencing spring wheat yield under varying nitrogen application gradients. This approach enhances the accuracy and reliability of the APSIM model in simulating spring wheat yield. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| | |