基于最小平方误差的单倍体和二倍体分类方法研究

基于玉米油分的花粉直感效应,采用核磁共振鉴别单倍体,快速、准确、低成本地确定单倍体与二倍体之间的含油率阈值,是实现大批量自动鉴别的前提和关键。提出一种基于最小平方误差的单倍体和二倍体分类方法,通过少量籽粒确定含油率阈值的研究,对单倍体和二倍体进行识别。以10个玉米组合为研究对象,利用核磁共振分析仪测定各组合一定数量单倍体和二倍体籽粒的含油率,依据最小平方误差方法确定各组合油分阈值,并通过逐级递增训练集样本数量的方法,确定油分阈值所需训练集规模,获得含油率阈值,能够满足单倍体正确识别率及对二倍体正确拒识率的均值达到90%以上的要求,为核磁共振全自动单倍体鉴别系统的后期大规模分类提供参数指导,提高分选效率。同时因单倍体生产效率低,运用最小平方误差的方法,通过少量的单倍体及二倍体籽粒来构建油分鉴别单倍体模型,以较小的投入获得更优的分类效果在玉米单倍体自动化分选中具有非常重要的现实意义,将该分类方法应用到全自动核磁分选单倍体系统中,可以达到较好的效果,为加快单倍体工程化育种提供保障。     

最小二乘支持向量机模型在径流预报中的应用

为解决径流中长期预报模型精度较低的问题,提出了一种基于混沌时间序列的最小二乘支持向量机预测方法,该方法依据相空间重构技术以及最小二乘支持向量机模型(以下简称LS_SVM)。利用该模型对黄河三门峡1919-1989年的71年实测年径流量进行预测。经过反复计算分析表明LS_SVM模型在径流中的应用具有较快的计算速度、较好的泛化能力以及较高的预测精度。     

Deficit irrigation based on drought tolerance and root signalling in potatoes and tomatoes

Agriculture is a big consumer of fresh water in competition with other sectors of the society. Within the EU-project SAFIR new water-saving irrigation strategies were developed based on pot, semi-field and field experiments with potatoes (Solanum tuberosum L.), fresh tomatoes (Lycopersicon esculentum Mill.) and processing tomatoes as model plants. From the pot and semi-field experiments an ABA production model was developed for potatoes to optimize the ABA signalling; this was obtained by modelling the optimal level of soil drying for ABA production before re-irrigation in a crop growth model. The field irrigation guidelines were developed under temperate (Denmark), Mediterranean (Greece, Italy) and continental (Serbia, China) climatic conditions during summer. The field investigations on processing tomatoes were undertaken only in the Po valley (North Italy) on fine, textured soil. The investigations from several studies showed that gradual soil drying imposed by deficit irrigation (DI) or partial root zone drying irrigation (PRD) induced hydraulic and chemical signals from the root system resulting in partial stomatal closure, an increase in photosynthetic water use efficiency, and a slight reduction in top vegetative growth. Further PRD increased N-mineralization significantly beyond that from DI, causing a stay-green effect late in the growing season. In field potato and tomato experiments the water-saving irrigation strategies DI and PRD were able to save about 20-30% of the water used in fully irrigated plants. PRD increased marketable yield in potatoes significantly by 15% due to improved tuber size distribution. PRD increased antioxidant content significantly by approximately 10% in both potatoes and fresh tomatoes. Under a high temperature regime, full irrigation (FI) should be undertaken, as was clear from field observations in tomatoes. For tomatoes full irrigation should be undertaken for cooling effects when the night/day average temperature >26.5 °C or when air temperature >40 °C to avoid flower-dropping. The temperature threshold for potatoes is not clear. From three-year field drip irrigation experiments we found that under the establishment phase, both potatoes and tomatoes should be fully irrigated; however, during the later phases deficit irrigation might be applied as outlined below without causing significant yield reduction:

•

Potatoes

°

After the end of tuber initiation, DI or PRD is applied at 70% of FI. During the last 14 days of the growth period, DI or PRD is applied at 50% of FI.

•

Fresh tomatoes

°

From the moment the 1st truce is developed, DI is applied at 85-80% of FI for two weeks. In the middle period, DI or PRD is applied at 70% of FI. During the last 14 days of the growth period, DI or PRD is applied at 50% of FI.

•

Processing tomatoes

°

From transplanting to fruit setting at 4th-5th cluster, the PRD and DI threshold for re-irrigation is when the plant-available soil water content (ASWC) equals 0.7 (soil water potential, Ψ_soil = −90 kPa). During the late fruit development/ripening stage, 10% of red fruits, the threshold for re-irrigation for DI is when ASWC = 0.5 (Ψ_soil = −185 kPa) and for PRD when ASWC (dry side) = 0.4 (Ψ_{soil, dry side} = −270 kPa).

The findings during the SAFIR project might be used as a framework for implementing water-saving deficit irrigation under different local soil and climatic conditions.     

基于自然光照反射光谱的温室黄瓜叶片含氮量预测

利用便携式光谱辐射仪测量了自然光照条件下温室黄瓜叶片的光谱反射率,并计算了反射率光谱的一次微分光谱。反射率光谱以及一次微分光谱与叶片含氮量的相关分析表明,温室内光谱特性与叶片含氮量相关性最大的敏感波段分别是505～664 nm和685～722 nm。当利用原始光谱进行分析时,通过变量筛选得到了4个敏感波长,分别是568、596、640和664 nm。偏最小二乘回归分析(PLSR)以及归一化颜色指数(NDCI)分析都表明,建模时的相关系数R_C>0.800,模型验证时的相关系数R_V>0.700。对微分光谱进行的相关分析结果表明,利用单一敏感波长520 nm就可获得理想模型,建模时的相关系数为0.880,模型验证时的相关系数为0.787。对比原始光谱的PLSR模型与一阶微分光谱的一元线性回归模型可以得知,原始光谱以及一阶微分光谱都可用于温室内叶片含氮量的预测,而且一阶微分光谱在一些特殊的波长处具有更高的预测能力,这些模型将成为开发便携式作物长势诊断仪器的技术基础。     

Estimation of genetic variance components and heritabilities for cut-flower yield in gerbera using least squares and maximum likelihood methods

Summary Genetic variance components and heritability were estimated for cut-flower yield of gerbera in the Davis population using ordinary least squares and maximum likelihood methods. The overall estimate of narrow-sense heritability is 0.33 based on least squares (LS) and 0.31 based on maximum likelihood (ML). The results of the study indicate that (1) ML and LS provide very similar results if sample size is large enough, suggesting both are useful for plant breeding programs; (2) about one third of the variation in gerbera cut-flower yield is additive, implying selection in cut-flower will be successful; and (3) although additive variation gradually decreased, heritability remained near 0.27 suggesting there is still potential variation in the population for further selection.     

随机取样技术中的数据处理

取样技术是一种广泛应用的信息处理技术。通过对模拟信号的取样量化处理，可以得到便于计算机处理的数字信号。若使被取样信号不失真，取样频率必须高于被取样信号最高频率分量2倍，这对于高频信号不容易做到。为此提出了随机取样技术，即用低频脉冲序列对高频信号随机取样，通过大量样值来统计出原信号的某些特征。     

监测采样间隔对作物模型性能的影响研究

为了获得不同监测采样间隔对作物生长模型的影响规律,避免现有研究和应用中存在的主观性或随意性,以番茄叶片蒸腾速率和CO2交换率为对象,采用3 种采样间隔数据及2 种常用方法进行建模、预测和比较研究。研究结果显示,在15、30、60 min的监测采样间隔下,番茄叶片蒸腾速率模型和CO2交换率模型的预测误差是不同的,GA-BP神经网络模型的预测能力普遍优于纯二次回归模型,但两者的结论是一致的,即对蒸腾速率和CO2交换率而言,最合适的监测采样间隔分别为30 min和15 min。本研究结果为监测采样间隔的设定、试验数据的选取和建模方法的运用提供依据,对作物模型研究和应用具有重要意义。     

耕作土壤表面不平度分析

利用激光式不平度测试仪测定犁耕耕地地面、圆盘耙耙地地面和驱动耙耙地地面的不平度，测定地面不平度沿着3个方向进行，即平行、倾斜和垂直于耕作方向。对每一种地面不平度的RMS高度、自相关长度和自相关函数进行了分析，结果表明：对于短的地块，可用单一尺度来表征地表不平度的特征，地表不平度特征属于指数相关函数，RMS高度和自相关长度具有明显的相关性；对于长的地块，平整的农业地表不平度具有分形性或多尺度性，其自相关函数是一种分形过程。     

精准农业中不同取样间距下Kriging插值精度对比研究

为了研究土壤不同养分的合理取样间距,研究选择了平顶山地区典型烟田为研究对象,在面积为4 hm2的地块上以20 m×20 m网格采样,共采集耕层(0~20 cm)土壤样品111个。在GIS软件平台上对采样点进行行列删除,人为改变取样间距大小,并运用地质统计学和普通克里格插值方法进行插值,生成了土壤各养分分布图。同时,随机选取7个样点,并将其在不同取样密度下的Kriging插值结果与实测值进行比较。由插值误差分析、独立检验、面积分析和养分分布图可以看出:在该研究区域内,碱解氮、速效磷、速效钾、速效铁和速效锌以20 m取样间距较为合适,有机质和速效铜的合理取样间距为60 m,速效锰则以40 m取样间距较适宜。研究目的是为该地区确定合理的土壤取样间距提供理论依据。     

桁杆张网渔具菱形和方形网目网囊的选择性研究

使用套网法作为选择性试验方法,Logistic曲线作为选择性曲线模型,并运用极大似然估计法进行曲线参数估算,对桁杆张网渔具不同网目尺寸(25mm、30mm、35mm、40mm以及45mm网目大小)、不同网目形状(菱形网目和方形网目)的网囊选择性进行分析。结果显示:各菱形网目网囊对黄鲫和银鲳的50%选择体长(或叉长)比相应方形网目的大(分别为3.79%和31.25%),而对小黄鱼的50%选择体长比相应方形网目小(7.92%);各菱形网目对黄鲫、小黄鱼和银鲳的选择范围都比相应方形网目大(分别为20.76%、5.36%以及45.99%)。这与传统认为方形网目选择性优于菱形网目的观点存在差异。通过讨论,认为黄鲫和银鲳体形较扁,游泳能力较弱是造成这些差异的主要原因。因此,在进行网囊结构改革时,不能盲目遵循方形网目优于菱形网目的传统观点。