【作者】 刘小军；

【导师】 曹卫星；

【作者基本信息】 南京农业大学 ， 农业信息学， 2011， 博士

【摘要】 水分精确管理作为作物生产管理的重要环节,对于保障粮食健康生产和节约水资源具有重要意义。本研究以水稻为研究对象,基于系列池栽试验,综合运用信息获取、生长分析、数理统计及建模方法,探明了不同土壤水分处理下水稻的生长发育规律及水分状况变化特征；明确了水稻植株水分状况与冠-气温差及叶-气温差的定量关系；确定了水稻植株水分状况与反射光谱及植被指数的定量关系,建立了基于反射光谱的水稻水分状况无损监测模型；通过定量描述水稻需水规律与环境因子之间的动态关系,根据土壤水分平衡原理,构建了水稻水分实时调控模型；进一步采用面向对象的程序设计思想和软构件技术,建立了综合性的基于模型的水稻水分管理决策支持系统。研究结果为水稻水分状况的实时监测和灌溉管理提供了理论依据和关键技术。以武香粳14和两优培九为试验材料,基于池栽试验,研究了不同土壤水分处理、不同生育时期和不同品种条件下水稻的生长及水分状况的动态规律。结果表明,不同水分处理下水稻株高、节间长度、穗长、一次和二次枝梗数、植株和叶层含水率差异显著,表现为：CK>W3>W2>W1;各生育期叶位叶片含水率值均表现为WL1<WL2<WL3<WL4。轻度土壤干旱胁迫处理(W3)延缓了叶片衰老,具有较高的叶面积指数、叶绿素含量和叶片净光合速率；抽穗期后,水稻上部叶片叶绿素含量反而高于下部叶片,且各叶位间的单叶净光合速率也表现为：L1>L2>L3>L4。各处理间产量最终表现为W3>CK>W2>W1,轻度水分胁迫(W3)处理产量最高。综合运用红外热成像仪和测温仪,比较了水稻冠-气温差和叶-气温差的变化特征,分析了与植株水分状况的定量关系。结果表明,不同土壤水分状况下,冠-气温差和叶-气温差值在不同生育期均表现为：W1>W2>W3>W4。顶部3叶的叶-气温差差异不明显,与顶4叶有一定的差异,各生育期表现一致；冠-气温差、L1/L3叶片含水率比值与植株和叶层含水率均呈显著线性相关关系；叶-气温差值与叶片含水率、气孔导度及蒸腾速率呈较好的相关性。表明冠(叶)-气温差及叶片含水率可用于水稻植株水分状况的监测。系统分析了不同土壤水分状况下叶片高光谱反射率和含水率的变化特征,及350-2500 nm范围内任意两波段组合而成的比值(RSI)、归一化差值(NDSI)及差值(DSI)光谱指数与水稻叶片含水率的量化关系。结果表明,不同土壤水分处理和顶部4张叶片间,叶片反射光谱具有明显的时空变化特征,叶片含水率的敏感光谱波段位于近红外和短波红外波段区域；比值RSI(R1402, R2272)及归一化差值NDSI(R1402, R2272)光谱指数与叶片含水率呈现良好的线性相关。独立试验资料对所建模型进行测试检验也显示,预测值和观察值的拟合精度较高,表明RSI(R1402, R2272)、NDSI(R1402, R2272)均可用于水稻叶片含水率的定量监测。基于冠层高光谱特征,综合分析了水稻植株水分状况与光谱反射率的定量关系。结果显示,水稻冠层高光谱反射率在不同土壤水分处理间差异显著；植株及叶层含水率的敏感光谱波段位于可见光区域和近红外波段区域；比值光谱指数RSI(R773, R1287)与植株含水率、叶层含水率呈良好的线性相关。基于独立试验资料对所建模型进行检验的结果显示,预测值和实测值的拟合精度较高,表明RSI(R773, R1287)可有效地监测拔节期后水稻植株及叶层含水率。在分析和提炼水稻水分管理理论与技术资料的基础上,通过定量描述水稻需水规律与栽培技术、品种类型、生态环境之间的动态关系,采用植株含水率、土壤水势作为中期调控指标,结合水稻主要水分指标无损监测模型,构建了水稻水分实时调控模型。进一步结合现有水稻水分管理知识模型、生长模拟模型、气象数据生成模型,基于系统工程思想和软构件技术,在Visual Studio.NET2005平台上开发了基于模型的水稻水分管理决策支持系统。系统具有基础信息管理、水分动态模拟、灌溉量设计、因苗实时调控及帮助等功能。对系统进行了实例分析结果表明,按照模型设计方案进行水分管理的田块,其产量和水分生产率均高于常规方案田块。表明系统设计的水分灌溉管理处方具有较强的适用性和指导意义。更多还原

【Abstract】 As the core content of crop production and management, the precision water management is of great significance to ensure food security and save water resource. Remote sensing, crop model and decision support system are important tools for monitoring crop growth characters, water indices and irrigation management. In this study, a series of field experiments with rice were carried out, the change characteristics of growth, development and water content condition under different soil water levels were defined with the methods of information query, growth analysis, mathematical statistics and mathematical modeling. The quantitative relationships of canopy-air temperature difference(CATD), leaf-air temperature difference(LATD) with plant water status were confirmed. By studying quantitative relationship between plant water status and hyperspectra, spectral index, nondestructive models for monitoring plant and leaf water contents were developed. Based on the theory of soil water balance, a sub-model for water real-time regulation was established by quantifying the relationships between water demand of rice and environment factors. By further adopting the methodology of software component design and the technology of object oriented program, a model-based decision support system for rice water management was developed. This work has provided a basic theory and key technologies for monitoring water status and irrigation management.The dynamic characteristics of growth, development and water condition were studied under different soil water conditions and rice cultivars of Wuxiangjingl4 and Liangyoupeijiu. Results showed the plant height, panicle length, internodes length, number of primary and second branch, plant and total leaf water contents were remarkably affected by soil water in different water treatments, the treads were CK>W3>W2>W1. The distribution of leaf water contents on main stem in all treatments and development stages were WL1<WL2<WL3<WL4. The leaf area index, SPAD value of leaves and leaf photosynthesis were the highest under low-grade soil dry stress (W3) for it delayed aging of leaf. After heading stage, the SPAD values of leaves at upper position were higher than basal position, and the distribution of leaf photosynthesis at different positions were L1>L2>L3>L4. The yield tread in all treatments were W3>CK>W2>W1, W3 treatment had the highest yield.By applying the measuring instruments of THREMOCAM P25 and Raytec ST60, the change characteristics of canopy-air temperature difference(CATD) and leaf-air temperature difference(LATD), and their quantitative relationships of CATD, LATD with plant water content(PWC) and leaf water content(LWC) were studied. Results showed the values of CATD and LATD in all treatments and stages were W1>W2>W3>W4. The leaf-air temperature differences among upper three positions were very small, but varied distinctly with the upper forth leaf at all development stages. CATD or radio of L1/L3 LWC with PWC and canopy leaf water contents were significantly related. LATD with LWC, stomatic conductance and transpiration rate showed certain relativity. It is concluded that CATD, LATD and LWC can be used to monitor plant water conditions.The change characteristics of leaf hyperspectral reflectance and water content were studied. Ratio spectral indices (RSI), normalized difference spectral indices (NDSI) and difference spectral indices (DSI) with all combinations of two wavebands between 350 and 2500 nm were calculated, and the relationship between spectral indices and LWC was analyzed. The results indicated that the leaf spectral reflectance varied distinctly with soil water treatments and different top leaves, the sensitivity bands mostly occured within near-infrared and short-infrared spectral regions. The spectral indices as RSI (R1402, R2272) and NDSI (R1402, R2272) were linear with LWC, giving the determination coefficient of linear regression (S-R2) of 0.80, and the predicted R2 (P-R2) based on the testing performance with independent datasets as 0.86. It is concluded that the RSI (R1402, R2272) and NDSI (R1402, R2272) can be used to monitor leaf water content in rice.Based on the characteristic of canopy hyperspectra, the relationship between plant water condition and hyperspectral reflectance were analyzed. The results indicated that canopy spectral reflectance varied distinctly with soil water conditions. The sensitivity bands mostly occured within visible light region and near infrared region. The RSI (R773, R1287) were the most effective approach for predicting plant and canopy leaf water contents. The predicted R2 (P-R2) based on the testing performance with independent datasets were 0.802 and 0.774 respectively. It is concluded that the RSI (R773, R1287) can be used to monitor plant and canopy leaf water contents.Based on analysis and extraction of the research results in rice water management theories and technologies, further with plant water contents or soil water potential as the regulation indices, a sub-model for water real-time regulation in rice was developed by combining rice water indices nondestructive monitoring model, quantifying the relationships of water demand and cultivation technologies, variety traits, and eco-environments. Furthermore, guided by the methodology of software component design and the technology of object oriented program, based on rice knowledge model for water management, rice growth simulation model and weather data generating model, a model-based decision support system for rice water management was developed based on platform of Visual Studio. NET 2005. The system realized the functions as basal information management, dynamic water simulation, design of irrigation amount, real-time dynamic regulation and help, etc. The field comparison experiments showed that the average grain yields and irrigation water use efficiencies were improved in two eco-sites under the model-designed irrigation strategies, as compared to conventional irrigation regimes. It appears that the present model has good applicability and proper guidance for rice water irrigation management.更多还原