基于DNDC模型的红枣生长模拟参数敏感性和产量不确定性分析  

Analyses of Parameter Sensitivity and Yield Uncertainty of Jujube Growth Simulation Based on DNDC Model

作　　者：王德娟 汪健平 冯建中[2] 井双泉[4] 许士东 隋立春[6] 黄光辉 WANG Dejuan;WANG Jianping;FENG Jianzhong;JING Shuangquan;XU Shidong;SUI Lichun;HUANG Guanghui(Changqing Engineering Design Co.,Ltd.,Xi'an 710018,China;Institute of agricultural information,Chinese Academy of Agricultural Sciences,Beijing 100081,China;Shandong Marine Resource and Environment Research Institute,Yantai Shandong 264000,China;Agricultural Science Research Institute of the 14th division of Xinjiang production and Construction Corps,Kunyu Xinjiang 848100,China;Institute of Agricultural Economics and Science and Technology Information,Xinjiang Academy of Agricultural Sciences,Urumqi 830091,China;School of geological engineering and surveying and mapping,Chang'an University,Xi'an 710054,China)

机构地区：[1]长庆工程设计有限公司,西安710018 [2]中国农业科学院农业信息研究所,北京100081 [3]山东省海洋资源与环境研究院,山东烟台264000 [4]新疆生产建设兵团第十四师农业科学研究所,新疆昆玉848100 [5]新疆农业科学院农业经济与科技信息研究所,乌鲁木齐830091 [6]长安大学地质工程与测绘学院,西安710054

出　　处：《新疆农业科学》2023年第3期651-663,共13页Xinjiang Agricultural Sciences

基　　金：新疆生产建设兵团重点领域科技攻关计划(2019AB002);中国农业科学院科技创新(CAAS-ASTIP-2016-AⅡ)。

摘　　要：【目的】研究红枣生长模型模拟输入参数的敏感性和产量预测不确定性,为红枣生长模拟模型的本地化和区域化参数调整优化提供依据,以提高模型模拟预测精度和效率。【方法】以新疆昆玉市现代农业示范区为研究区,应用可扩展傅里叶振幅敏感分析法(EFAST)和蒙特卡罗法分析基于DNDC模型系统新构建的红枣生长模型的输入参数敏感特性和产量预测不确定性。【结果】作物参数中全株生物量中果实比例(Gfra)、最大作物产量(MaxY)、生长积温(TDD)和需水量(WaterR)等指标敏感度最高,土壤参数中田间持水率(FC)和孔隙度(Por)等指标敏感度最高,田间管理参数中灌溉量(IrrAm)、施肥量(FerAm)和有机肥施肥量(ManAm)等指标敏感度最高;随着参数的波动范围由±5%增大到±10%,红枣预测产量正态分布的相关一致性系数增大,模型的平稳性增加。【结论】调整参数优化模型,并对2015~2019年各年份进行产量模拟测试验证,预测产量结果相对误差控制在±8%以内(最小误差为-1.99%),调整红枣产量预测模型参数,提高了模型预测产量的精度,优化趋于合理。【Objective】Analyses of input-parameters sensitivity and yield uncertainty of jujube growth simulation are very significant steps in the hope of providing suggestions for localization and regionalization of jujube growth model to serve as improving the accuracy and efficiency of model simulation prediction.【Methods】In this paper,the Extended Fourier Amplitude Sensitivity Test(EFAST)and Monte Carlo(MC)method were used to analyze the input-parameter sensitivity and output-parameter uncertainty of a new jujube-yield prediction Denitrification-Decomposition(DNDC)model,which was generated by the crop-type generator of the DNDC model system,in a modern-agriculture demonstration area located in Kunyu city,Xinjiang Uygur Autonomous Region.【Results】The results showed that the jujube-crop parameters including grain fraction of total biomass(Gfra),maximum grain yield(MaxY),thermal degree days(TDD)and water requirement(WaterR),the soil parameters including field capacity(FC)and porosity(POR),and the field management parameters including Irrigation(IrrAm),Fertilizing amount(FerAm)and Manure amount(ManAm)were the most sensitive to modeling output(i.e.,jujube yields),respectively.According to the simulation of jujube yields in a typical year of 2018,when the fluctuation ranges of input-parameters were extended from±5%to±10%,the correlation consistency coefficients of jujube-yield prediction with the corresponding normal distribution increased,respectively,which showed the stability of the jujube-yield model increasing.【Conclusion】Based on the sensitivity and uncertainty analysis of the jujube-yield model parameters,the model parameters were adjusted and optimized,and then,the model was tested and verified to jujube yields from 2015 to 2019.The relative errors of the prediction yields,compared with the in situ data,were controlled within±8%(and the minimum error reached-1.99%),which presented a great improvement of the accuracy of the prediction yields and meant that the optimization and adjustment of the model

关 键 词：DNDC模型 枣树 模型本地化 敏感性与不确定性分析 

分 类 号：S665.1[农业科学—果树学]