【作者】 邹薇；

【导师】 曹卫星；

【作者基本信息】 南京农业大学 ， 作物栽培学与耕作学， 2009， 博士

【摘要】 作物生长模型的研究和应用对于农业生产的信息化和数学化具有重要的理论意义和应用价值。本研究借鉴国际上先进的建模理论和本实验室形成的作物模拟方法体系和基本框架,运用系统分析原理和动态建模技术,通过实施不同生态点、不同类型品种的播期和氮素处理试验,对大麦生长发育的基本规律及其与环境因子之间的关系予以定量解析和综合,构建了基于生理生态过程的大麦生长发育模拟模型(BarleyGrow),为数学化麦作系统的构建奠定了基础。以扬州地区5个大麦品种春播条件下的顶端发育和物候发育观测资料和历史资料为基础,构建了基于生理发育时间的顶端发育和物候期机理模型。模型量化了热效应、光周期、春化效应对大麦发育的影响,引入了播种到出苗所需的有效积温、灌浆期发育基点温度、生理春化时间、临界日长、光周期反应起始点、最短苗穗期、最短灌浆期7个品种生物学参数。模型将每日生理发育时间的增量除以水肥丰缺因子来表现水肥对大麦发育的影响,客观体现了大麦在水肥丰缺条件下的发育延迟或提早现象；用非线性函数来表达春化效应和相对热效应,确立了不同品种相对春化效应和相对热效应的曲线族；用正弦函数来表达不同品种光周期效应。经测算,各大麦品种到达单棱期、二棱期、雌雄蕊分化期、药隔形成期、雌蕊柱头二裂分叉期、雌蕊柱头毛状突起期等顶端发育阶段的生理发育时间分别为2.6、5.6、11.3、13.1、15.3、18.2、28.7 d,到达出苗期、拔节期、抽穗期、灌浆期和成熟期等主要物候期的生理发育时间为0、13.1、28.7、32.8、51.5 d,从而形成了不同大麦品种在不同气候和栽培条件下统一的衡量发育的定量尺度。BarleyGrow模型中利用遗传-模拟退火算法来确定各品种的生物学参数,提高了应用程序求算参数的精度。采用4个生态区(南京、扬州、武汉、昆明)、10个大麦品种在不同播期下的顶端发育和物候期资料,对BarleyGrow进行了检验和评价,并将其模拟效果与YDmodel和SUCROS模型进行了对比分析。结果表明,BarleyGrow对不同地区、不同播期、不同品种各顶端发育和物候期预测准确而稳定,其均方根差RMSE在1.06～7.94 d之间,而YDmodel为6.26～13.35 d, SUCROS为11.22～20.28 d。各参试品种对BarleyGrow模型中灌浆期基点温度、生理春化时间、临界日长、最短.苗穗期4个参数反应敏感。BarleyGrow模型对中国广大地区不同温光条件下的大麦顶端发育和物候发育均具有较好的预测效果,尤其对药隔期、二裂期、毛状期、抽穗期、灌浆期、成熟期的模拟精度高而稳定,表现出较强的机理性以及较好的预测性。准确模拟叶面积指数是作物生长模拟模型可靠预测作物生长和产量的关键。通过系统分析扬州和武汉地区不同大麦品种高产群体叶面积指数变化动态,建立了高产群体叶面积指数与生理发育时间、每日光合有效辐射累积量和孕穗期最适叶面积指数之间的关系模型,并通过水肥丰缺因子的修订得到实际条件下叶面积指数动态。模型较为全面地考虑了影响大麦品种叶面积变化的内外因素,内因主要体现为品种叶面积指数扩展的遗传特性,外因主要包括温度日较差、日照时数、辐射量和水肥丰缺因子。利用扬州、南京和昆明地区不同品种的播期试验及氮肥试验资料对模型进行了检验,结果表明模型能较好地模拟不同地区、不同气候、不同栽培管理条件下大麦叶面积指数的变化动态。在综合已有作物模拟模型优点的基础上,构建了基于生态生理过程的大麦光合生产与干物质积累模拟模型。模型将一天均分为96个时间点,分别计算各时间点的光合有效辐射,以复化辛普森积分计算每日冠层光合同化量,并采用高斯积分法与辛普森积分法对昆明和武汉试验区地上部干重进行了对比检验。结果表明辛普森积分法模型具有更优的预测效果,模拟值与观测值吻合度高。因此,辛普森积分法进行光合作用模拟可以作为大田环境中精确模拟作物生长的新方法。通过系统分析南京和扬州地区不同处理下大麦品种干物质分配的变化动态,建立了高产大麦群体物质分配指数与生理发育时间、每日光合有效辐射累积量及生物学参数(收获指数和绿叶分配指数的遗传差异)之间的关系模型,并通过水肥丰缺因子的修订得到实际条件下大麦物质分配指数动态。各器官重量为该器官物质分配指数与生物量的乘积。利用昆明和武汉地区不同品种不同播期的试验资料对本模型进行了检验与评价,并将其模拟效果与Ydmodel进行了对比分析,结果表明,本模型对大麦各器官间的物质分配及各器官干重的模拟效果较好,模拟值与观测值吻合度高,能较好地模拟不同地区、不同气候、不同栽培管理下大麦物质分配和器官生长动态。通过产量构成法构建了大麦产量预测模型。以南京、昆明、武汉3个地区各试验处理中不同大麦品种最适条件下的产量因素为基础,建立了最适条件下每株穗数相对值、每穗粒数相对值、千粒重相对值与累积光合有效辐射之间的的回归方程,构建了实际条件下不同大麦品种每株穗数、每穗粒数、千粒重与最适条件下三指标潜在值及水肥丰缺因子之间的函数关系。模型较为全面地考虑了大麦产量形成的内外因素,内因主要体现为品种遗传特性(潜在的每株穗数、潜在的每穗粒数、潜在的千粒重和灌浆期因子),外因包括光合有效辐射和水肥丰缺因子。运用武汉、昆明、扬州3个地区不同品种、不同播期的田间试验资料对模型进行了测试和检验。结果表明,模型对大麦产量构成因素及理论产量的模拟效果较好,模拟值与观测值吻合度高,显示模型具有较高的预测性和适用性。更多还原

【Abstract】 The research and application of crop growth model would be important for facilitating development of informational and digital agriculture. In the present study, the relationships of growth and development to environment factors were analyzed and integrated by using the field experiments data with different genotypes, sowing dates and nitrogen application levels in different ecosites. By adopting advanced modeling technology abroad and the methodology of crop growth model developed by our lab, a physiological process-based barley simulation model (BarleyGrow) was developed through the system analysis and mathematical modeling. The present study should be useful for prediction of growth performance under different conditions and development of barley digital management system in barley crop.Based on the data from Yangzhou field experiment with five cultivars at spring sowing date, three processes of thermal effectiveness, photoperiod and vernalization in barley were quantified, and a physiological development time (PDT) based barley model for phasic and phenological development was developed. Seven cuitlvar parameters were used in the model, including the accumulated temperature from sowing to germination (GDDo), basic temperature in filling period (Tbmax), physiological vernalization time (PVT), critical day length (DLc), start time of photoperiod response (PPs), minimum time from emergence to heading (EHmin) and from heading to maturity (FDmin). The ratio of daily increment of PDT and the deficit of both water and nitrogen was used to estimate the effect of environment on phasic and phenological development. A nonlinear function was adopted to describe vernalization and thermal effectiveness, and the sinusoidal function was used to describe photoperiod curve cluster for different cultivars. Physiologically, our model estimated 2.6 days to reach the single ridge stage,5.6 days to the double ridge stage,11.3 days to the stamen and pistil initiation stage,13.1 days to the anther separation stage,15.3 days to the pollen mother cells stage and 18.2 days to the tetrad stage. Phenologically, estimations were 13.1 days to reach the jointing stage,28.7 days to the heading stage,32.8 days to the grain filling stage and 51.5 days to the maturity stage. PDT was consequently used as a unified scale for measuring developmental progress of different cultivars under different climate and cultural practices.In BarleyGrow, the optimum values of the model parameters were obtained through genetic-simulated annealing algorithms. Based on the field experiments with 14 barley cultivars on different sowing dates at 4 ecosites (Nanjing, Yangzhou, Wuhan and Kunming), the submodel for apical and phenological development was validated by comparing with YDmodel and SUCROS model. As a whole, the BarleyGrow model had an accurate and stable estimation on apical and phenological development. The root mean square error (RMSE) with the BarleyGrow model was ranged between 1.06 and 8.13 days for various cultivars, compared to 6.26 and 13.35 days with YDmodel, and 8.84 and 20.28 days with SUCROS. Compared with YDmodel and SUCROS model, the BarleyGrow model was quite sensitive to basic temperature in grain filling time, physiological vernalization time, critical daylength and minimum time from emergence to heading. The BarleyGrow model gave good predictions of apical and phenological development for a diverse range of temperature and photoperiod conditions across China. Especially, effects in anther separation, pollen mother cells, tetrad, heading, grain filling and maturity stages were better predicted.Accurate simulation of leaf area index (LAI) is critical for reliable prediction of crop growth and yield using a crop growth model. Based on the systematic analysis of barley experimental data from different cultivars and sowing dates at Wuhan and Yangzhou, the submodel for LAI estimation was developed. Along with the expansion coefficient of LAI for cultivar genetic properties, the relationships of leaf area index dynamic in barley under high yield to physiological development time (PDT), the accumulation of photosynthetic available radiation after sowing (∑PAR) and the optimum LAI at booting were simulated. The actual dynamic of LAI was modified with water and nitrogen factors based on the dynamic of LAI under high yield. The internal and external factors of leaf growth and development in barley were integrated into the model. The internal factor was genetic property of LAI expansion. Environmental conditions included daily temperature difference, sunlight,∑PAR, water and nitrogen factor. The submodel was tested with the different cultivars under different sowing dates and different nitrogen rates at Yangzhou, Nanjing and Kunming. The results showed that this submodel gave good predictions of LAI in barley under different ecosites, climates and cultivation practices. Based on the merits of existing crop simulation models, a submodel for photosynthesis and dry matter accumulation was developed. In the submodel, a day was separated into 96 time segments, and the corresponding photosynthetically active radiation during each time segment was calculated. The daily canopy assimilation was simulated using complex Simpson integration method, and the simulation result was evaluated by comparing the method of Gauss integration. Testing results showed that the Simpson integration method was better than Gauss integration method, and featured with higher predictability and broader applicability. Thus the Simpson integration method could be used as a new method to accurately simulate crop dry matter accumulation.Based on the systematic analysis of barley experimental data from various cultivars at different treatments in Nanjing and Yangzhou, a process-based submodel was developed for predicting dry matter partitioning and organ growth in barley. Along with harvest index and partitioning coefficient of leaf for cultivar genetic properties, the relationships of dry matter partitioning dynamic under high yield to physiological development time (PDT), the accumulation of photosynthetic available radiation after sowing (∑PAR), and cultivar genetic properties. The actual dry matter partitioning indices of organs in barley were modified by water and nitrogen limitation factors based on dry matter partitioning dynamic under high yield. Organ weights were the products of corresponding organ partitioning index and biomass. The submodel for organ partitioning indices and organ weights was tested with dataset from different cultivars at different sowing dates in Wuhan and Kunming, and showed good predictions of partitioning indices and organ weights under various conditions.Submodel for yield prediction was established through the method of yield components. Based on the experiment dataset from different cultivars under the optimal condition at Wuhan, Yangzhou and Kunming, the regression equations were built between the relative values of ear per plant, kernel per ear, and thousand-grain weight, and accumulated photosynthetic effective radiation (∑PAR). Ears per plant, kernels per ear, and thousand-grain weight under the actual condition were the equations of their potential values under the optimal condition, and water and nitrogen factor at the actual condition. The internal and external factors of yield components formation on barley were integrated into the model. The internal factors were genetic properties of cultivar, including potential ears per plant, potential kernels per ear, potential thousand grain weight and grain filling duration. Environmental conditions included∑PAR, water and nitrogen factor. The submodel was calibrated and validated with field experimental data from different cultivars at different sowing dates in Wuhan, Kunming and Yangzhou. The results showed that the model could well simulate the yield components and theoretic yield with high applicable levels.更多还原