葡萄酒发酵过程主要参数近红外光谱分析

用近红外光谱结合化学计量学方法对葡萄酒酒精发酵中葡萄糖、果糖、乙醇和甘油4个指标进行了定量分析,化学指标的测定采用高效液相色谱法.通过对近红外数据进行筛选、变量标准化等预处理,比较了主成分回归和偏最小二乘回归定量分析的模型质量,以决定系数、校正均方根误差、预测均方根误差为模型质量的评价指标.通过比较发现,对于葡萄糖和果糖,主成分回归与偏最小二乘回归的预测精度相当;对于乙醇,主成分回归预测结果较优;对于甘油,偏最小二乘回归的预测结果要优于主成分回归.主成分回归所采用的成分数要多于偏最小二乘回归,但二者都可以用于上述4种成分的定量分析,其预测精度也相近.     

基于支持向量机的有机肥总养分含量NIRS分析

以我国22个省市的120份畜禽粪便为原料的有机肥产品样品为研究对象,利用近红外漫反射光谱法与支持向量机相结合建立了有机肥产品中总养分含量的快速测定模型,并与偏最小二乘法所建模型作了比较.利用偏最小二乘回归法所建立的基于原始样品和干燥粉碎样品的总养分含量近红外模型验证决定系数R_v~2、预测标准差SEP和验证相对分析误差RPD分别为0.96、7.95 g/kg、2.47和0.93、8.02 g/kg、3.58.利用支持向量机回归法所建立的干燥粉碎样品中总养分含量的近红外模型验证决定系数R_v~2、预测标准差SEP和验证相对分析误差RPD分别为0.93、7.38 g/kg和3.88.结果表明,近红外漫反射光谱法可以快速测定畜禽粪便为原料的有机肥产品中总养分含量,与偏最小二乘法相比,支持向量机所建模型具有更高的预测精度.     

西瓜可溶性固形物含量近红外透射检测技术

设计的近红外透射式西瓜可溶性固形物含量（SSC）检测系统主要包括光纤光谱仪、光纤透射附件、数据采集卡以及自制光源。对50个麒麟瓜SSC进行了预测试验,采用主成分回归和偏最小二乘回归法分别建立了样品的原始光谱、一阶微分光谱、二阶微分光谱和SSC的预测模型,结果表明偏最小二乘回归法建立的模型具有较高的相关性,相关系数为0．951,均方根校正误差0．347,均方根预测误差0．302,样本真实值与预测值的相关系数为0．910。     

绿茶杀青叶料含水率可见-近红外光谱检测

为实现绿茶杀青叶料含水率的快速无损检测,基于可见-近红外光谱分析建立含水率的预测模型。使用FieldSpec 3型便携式地物光谱仪,采集192个杀青叶料样品的漫反射光谱信息,基于X-Y共生距离的样本划分算法SPXY,确定144个样本的校正集和48个样本的预测集。进行一阶微分和移动平滑滤波预处理后,采用相关系数法优选出11个特征波段,建立了含水率检测的偏最小二乘回归、主成分回归、人工神经网络及其组合的模型。结果表明,选用5个主成分的偏最小二乘回归模型最佳,其校正和预测模型的相关系数分别为0.990和0.819,均方根误差分别为0.011和0.037,预测含水率的平均相对误差为3.30%。     

基于高光谱成像技术的面条中马铃薯全粉含量检测

为了快速检测面条中马铃薯全粉含量,研究近红外高光谱成像技术定量检测面条中马铃薯全粉含量的可能性,自制了马铃薯全粉质量分数在0~35%内随机均匀分布的120个面条样品,在900~2500nm范围采集高光谱图像,随机选取80个样品作为校正集,分别采用原始光谱和经过6种预处理方法预处理后的光谱建立了偏最小二乘回归、主成分回归、支持向量机回归模型。结果表明经标准化预处理后用偏最小二乘回归建模效果最好,校正集决定系数（R2C）为0.8653,交叉验证集决定系数（R2CV）为0.6914。用回归系数法在经过标准化预处理后的光谱数据中提取了与全粉含量相关的特征波长,建立了马铃薯全粉含量偏最小二乘回归简化模型, 校正集决定系数（R2C）为0.8685,交叉验证集决定系数（R2CV）为0.8021,基于特征波长建立的模型效果优于全波段模型,模型效果得到了一定的提高。以剩余40个未参与校正模型建立的样品作为预测集,基于特征波长建立了标准化-偏最小二乘回归简化预测模型,预测集决定系数（R2P）为0.8546,模型具有较好的预测能力。结果表明利用近红外高光谱成像技术可检测面条中马铃薯全粉含量,可为马铃薯全粉面条的快速无损检测建立新的方法。     

近红外漫反射光谱检测土壤有机质和速效N的研究

利用近红外漫反射光谱检测技术对土壤有机质和速效N含量进行了相关研究。通过自行设计的NIR光谱系统测定了150个土壤样品有机质和速效N。126个土壤样品用来建立校正集模型,其余24个用来验证模型的性能。采集完整土壤样品的近红外漫反射光谱,原始光谱经移动窗口平滑处理、SNV和一阶微分预处理后,分别采用最小二乘支持向量机(LS-SVM)和偏最小二乘法(PLS),建立土壤有机质和速效N含量的定量预测数学模型。结果表明采用一阶微分结合最小二乘支持向量机(LS-SVM)所建模型的预测效果较好,土壤有机质和速效N含量定量预测数学模型的决定系数分别为0.8255和0.8015,均方根误差分别为2.84和16.80。近红外漫反射光谱作为一种检测方法,可用于评价土壤有机质和速效N含量。     

PLS在提取需水量预测影响因素中的应用

需水预测模型的输入变量直接影响了需水预测模型精度的高低。为了找出影响需水量的主要因素,选择偏最小二乘回归模型,建立了需水量的影响因素与需水量之间的相关关系模型。偏最小二乘法是多元线性回归、典型相关分析和主成分分析的集成与发展。偏最小二乘法提取的成分既能够很好的概括自变量系统中的信息,又能最好的解释因变量。通过计算影响因素与需水量之间的回归系数等信息,得出了影响需水量的主要成分,确定了需水预测模型的输入项。     

基于近红外光谱的活性米品质检测研究

选取105份具有代表性的活性米样品,研究光谱范围在918~1 045nm内12个波长点的近红外反射光谱与其品质的相关性。利用近红外光谱技术和多元线性回归、主成分回归和偏最小二乘法3种校正方法对活性米中γ-氨基丁酸、水分、蛋白质、淀粉、明度值、红度值及黄度值进行定量分析;选出最优校正模型,模型的预测决定系数分别为0.903、0.964、0.953、0.951、0.949、0.961、0.916,预测标准偏差分别为0.598、1.367、1.367、2.688、0.996、50.144、0.952。研究结果表明:应用近红外光谱技术检测活性米品质是可行性的。     

苹果内部品质的CT成像结合傅里叶变换方法检测

通过对苹果的CT扫描灰度图像进行傅里叶变换,在变换后的频域图像中提取16个参数,并结合苹果的可溶性固形物质量分数、可滴定酸度、pH值和含水率,采用主成分回归和偏最小二乘回归的方法建立模型,预测苹果的内部品质。结果表明,采用主成分回归分析建立模型,主成分的累计贡献率选取99%的情况下,F检验的P均小于0.05,预测效果良好。采用偏最小二乘回归时潜变量的个数为12时,各模型的误差平方和最小,预测效果良好。对两个模型进行误差率验证表明主成分回归的模型略优于偏最小二乘回归模型。     

黄酒总糖含量的中红外光谱定量检

利用中红外光谱技术对黄酒中的总糖含量进行了快速检测.实验以3个厂家的共30个样品为研究对象,在1000～1500cm-1和2500～3000cm-1波段范围内建立了多元线性回归、主成分回归以及偏最小二乘回归模型,并比较了微分的光谱预处理方法对模型精度的影响.二阶微分光谱建立的偏最小二乘模型的精度最佳,交互验证的相关系数为0.984,RMSECV为0.765g/L, RPD为5.75,表明中红外光谱技术可以用于黄酒中总糖含量的快速检测.     

Induction of soil water repellency following irrigation with treated wastewater: effects of irrigation water quality and soil texture

We hypothesized that organic matter (OM) content originating from treated wastewater (TWW) irrigation and soil texture dominate the intensity of soil water repellency. The relationship between soil texture, wastewater treatment level, and water repellency was examined in a 3-year lysimeter experiment (2008–2010). Soil type–water quality combinations, consisting of three soils with different specific surface area (SSA) and four levels of water quality differing in OM content, were tested. In each year, water repellency developed in all TWW quality treatments, but not in freshwater-irrigated controls. At the end of each year (except 2009), the highest degree of repellency was exhibited by sandy soil treated with the lowest quality TWW (highest OM content). The lowest degree of water repellency was consistently exhibited by the soil with the highest SSA irrigated with the highest quality TWW (lowest OM content). Water quality, rather than SSA, was the dominant factor in determining degree of repellency induced by TWW irrigation.     

Logistic model application for prediction of maize yield under water and nitrogen management

Simulation of crop yield allows better planning and efficient management under different environmental inputs such as water and nitrogen application. However, most of the models are complicated and difficult to understand. Furthermore, input data are not readily available. The objectives of this investigation were to use logistic equation to quantify the influence of seasonal water and nitrogen application on maize biomass accumulation and grain yield and to develop empirical models for prediction of maize biomass and grain yield. Logistic equations were fitted to dray matter (DM) yield at different times in the growing season at different irrigation water and nitrogen levels. The parameters of the logistic equations were then fitted to irrigation water and nitrogen as empirical functions. Further, the harvest index (HI) was related to the applied water and nitrogen as another empirical model. The empirical logistic models were used to estimate the DM and grain yield based on data from another experiment in the same area. Results indicated that the empirical models predicted the DM yield during the growing season with an acceptable accuracy, but dry matter (DM) prediction at harvest was very good. The grain yield also was predicted with a very good accuracy. It is concluded that logistic equation along with the presented empirical models for prediction of constants in logistic equation and HI are appropriate for accurate prediction of DM and grain yield of maize at the study region.     

The Effect of Formic Acid,Formic Acid Salt and Formic Acid with Formalin on Silage Fermentation,Digestibility and Intake,and on Liveweight Change of Young Cattle

Data from four experiments conducted at ADAS Liscombe Research Centre during 1977–1982 were used to compare unwilted silage treated with either formic acid (FA), a complex formic salt (FS) and formic acid with formalin (FF) with non-additive-treated silage (C), made from herbage with dry matter (DM) 166 g/kg and water-soluble carbohydrate (WSC) 21 g/kg. Following ensilage, the DM content of C was significantly less than that of the other treated silages. Compared with C, the additive treatments significantly decreased pH, acetic acid, butyric acid, total short-chain fatty acids and total acid content and increased the WSC content. Compared with FA and FS, FF significantly increased the WSC content. DM losses from FA and FF treatment were significantly less than C but not FS. Additive treatment significantly increased silage DM intake and liveweight gain compared with C. It is suggested that additive treatment irrespective of the nature of the formic acid applied produced well-fermented silage, which when fed to growing cattle significantly increased both silage DM intake and liveweight gain compared with non-additive-treated silage.     

The Effect of Formic Acid with Formalin Addition and Wilting on Silage Fermentation and Intake,and on Liveweight Change of Young Cattle

Data from seven experiments conducted at ADAS Liscombe Centre during 1980–1983 were used to compare formic acid with formalin-treated unwilted (FF) and wilted silage (WFF) with wilted non-additive-treated (W) silage made from herbage [dry matter (DM) of 172 g/kg and water-soluble carbohydrate (WSC) 114 g/kg DM]. Following ensilage, the DM content of FF was significantly less than W and WFF. Compared with W, WFF treatment significantly decreased silage pH, ammonia–N and butyric-acid content and significantly increased the residual WSC, whereas FF significantly reduced pH, ammonia–N, butyric acid, protein and ash content and significantly increased lactic acid, acetic acid and lactic acid expressed as a proportion of total acid content. DM losses were significantly higher for FF than from W or WFF treatments. W and WFF significantly increased subsequent silage DM intake compared to FF. Daily liveweight gains were, however, significantly higher on FF compared with W but not WFF treatment. It is suggested that FF treatment of unwilted silage produced well-fermented silage, which when fed to growing cattle had a lower DM intake, but higher liveweight gain than wilted silage, and that when compared with wilted silage (W) formic acid with formalin treatment (WFF) improved fermentation, but provided no significant benefit in terms of either silage DM intake or liveweight gain.     

马铃薯干物质含量高光谱检测中变量选择方法比较

为提高利用高光谱成像技术快速检测马铃薯干物质含量的精度,比较了主成分分析法(PCA)、组合间隔偏最小二乘法(siPLS)、遗传偏最小二乘法(GA-PLS)、无信息变量消除法(UVE)以及竞争性自适应重加权算法(CARS)等变量选择方法。在此基础上提出一种竞争性自适应重加权算法与连续投影算法(SPA)相结合的波长选择方法,最终将原始光谱变量从678个减少到了27个。用27个变量建立多元线性回归模型,模型预测集相关系数Rp为0.86,预测均方根误差为1.06%。实验结果表明:高光谱成像技术能够对马铃薯干物质含量进行检测,同时CARS-SPA是一种有效的变量选择方法。     

Ameliorating water repellency under turfgrass of contrasting soil organic matter content: Effect of wetting agent formulation and application frequency

Water conservation strategies are being developed in regions of the world expected to experience decreases in water resources due to changing climates. Strategies advocated for improving water-use efficiency may increase the incidence of soil water repellency in sandy-textured soils. We evaluated the effect of soil wetting agent formulation, and application frequency, on water repellency in sandy soil with two contrasting organic matter (OM) contents under kikuyugrass [Pennisetum clandestinum (Holst. Ex Chiov)], and irrigated at 60% replacement of net evaporation in a climate subject to hot, dry summers. The randomized plot design included two turfgrass ages [established from 20 week (7.7% OM) or 20 year old (30% OM) turfgrass in 2005, the latter included a 50 mm ‘mat’ layer], two soil wetting agent formulations (granular or liquid); two application frequencies (one or two applications per irrigation season); and plots of both turfgrass ages that did not receive any wetting agent (nil control). Both wetting agent formulations contained the same active ingredient (propylene oxide-ethylene oxide block polymer), and all wetting agent treatments received the same rate (69 L active ingredient ha⁻¹). Water repellency in the surface soil (0-5 mm), measured using the molarity of ethanol droplet test (MED), ranged from 1.09 M to 4.32 M during the irrigation season, and was more severe in the soil with high OM (average MED, 3.3 M) than low OM content (average MED, 2.7 M). Applying one application of either granular or liquid soil wetting agent at the commencement of the irrigation season decreased the severity of soil water repellency by up to 30% in the high OM soil and by up to 60% in the low OM soil during the summer, and without the need for a second application. The decline in soil water repellency in response to soil wetting agent application was not matched by an increase in soil VWC in summer, and turfgrass quality was considered acceptable throughout the study. The soil wetting agents were less effective at treating water repellent sand containing a significant amount of OM than sand with low OM content.     

The Effect of Formic Acid with Formalin on Grass Silage Fermentation and the Performance of Dairy Cows

In the first experiment, perennial ryegrass with dry matter (DM) 260 g/kg and water-soluble carbohydrate (WSC) 140 g/kg DM was harvested as first-cut material on 25 May 1982. It was pre-wilted for 24 h prior to ensilage and treated with either a formic acid with formalin additive (WFF) at 4l/t or no additive (W) and ensiled in two adjoining covered 150 t bunker silos. Subsequently, both silages were satisfactorily fermented. Nevertheless, the pH, ammonia–N g/kg N and lactic-acid content were significantly lower and the residual WSC content significantly higher for WFF than W. Each silage was subsequently individually fed to ten Friesian and ten Holstein dairy cows, together with 10 kg/d of a compound feed containing 160 g/kg crude protein in a 70 d feeding trial. Additive treatment had no significant effect upon silage DM intake, milk output or composition. However, the daily liveweight losses of cows given WFF-treated silage were significantly higher than those given W.In the second experiment, perennial ryegrass (DM 180 g/kg and WSC 170–200 g/kg DM) harvested as first-cut material on 2 June 1983, was either cut and picked up within 20 min of cutting and treated with either the formic acid with formalin additive at 4l/t (NWFF) or no additive (NW); or wilted for a maximum of 24 h and treated with either the formic acid with formalin additive at 4l/t (WFF) or no additive (W). Wilting significantly increased subsequent silage DM. The pH of NW was significantly less than W, whilst the ammonia–N g/kg N of NWFF was significantly less than NW and W and that of W significantly higher than the other treatments. Additive treatment significantly increased the residual WSC content. Each silage was subsequently individually feed to 12 Friesian and 12 Holstein dairy cows together with 10 kg/d of a compound feed to 12 Friesian and 12 Holstein dairy cows together with 10 kg/d of a compound feed containing 160 g/kg protein in a 180 d feeding trial. The silage DM intake of WFF was significantly higher than NW. Silage treatment had no significant effect upon milk output or composition. The daily liveweight loss of cows given WFF-treated silage was significantly less than the other treatments, and the cows on NW had a significantly longer calving index than the other treatments.It is concluded that compared with non-additive-treated silage, formic acid with formalin applied at ensilage produced better fermented silage which, when subsequently given to lactating dairy cows, produced no significant increase in daily silage DM intake, milk output or composition, but had small significant effects upon liveweight change.     

河套灌区覆盖对盐渍土壤养分迁移与分布的影响

在河套灌区对盐渍土壤覆盖后进行秋浇田间试验,研究不同覆盖下秋浇、冻融及玉米生育期3个阶段的土壤养分含量变化规律。试验设7个处理,分别为秸秆覆盖量1.2(F1.2)、0.9(F0.9)、0.6(F0.6)、0.3 kg/m2(F0.3)、玉米整秆覆盖(YZ)、地膜覆盖(DM)、未覆盖(CK)。结果表明,秋浇后地下水位较高,秋浇至冻融期间土壤水分与地下水具有补排关系,土壤养分的转化、迁移同时发生,秋浇后至冻融期内土壤养分的变化规律复杂;地表覆盖改变了土壤的水土环境,不同处理间养分含量存在差异,在试验期内全效养分的变化幅度较小,速效养分变化幅度较大;分析玉米播前耕层土壤供肥能力,秸秆覆盖处理的全氮、全磷含量小但供应强度大,春播时需补充氮肥、磷肥,以保证作物的生长。秸秆覆盖处理的全钾含量与钾素供应强度均高于CK,与CK相比较,秸秆覆盖使耕层土壤较好地满足了玉米播种及后续生长过程中的钾肥需求;相同施肥条件下,经过一年试验后,对于土壤耕层,处理DM的全氮含量较试验前降低了0.10 g/kg,其余处理变化幅度较小。各处理的全磷含量较试验前增加。处理CK、DM的全钾含量较试验前增加,处理F1.2、F0.9降低了全钾含量。处理YZ、F1.2和DM的碱解氮含量升高,其余处理则降低。各处理的速效钾、速效磷含量均较试验前升高。处理F1.2、F0.9和CK的有机质含量较试验前增加,其余处理则降低。     

渭北台塬区耕地土壤有机质与全氮空间特征

以渭北台塬农业县——蒲城县为研究区,采用2011年耕地地力调查与质量评价项目数据,运用经典统计学、地统计学结合GIS技术,探讨以有机质为辅助变量的协同克里格法在区域土壤全氮及碳氮比估值中的适用性,研究耕地土壤有机质和全氮的空间变异特征、影响因素及其与碳氮比的分布特征。结果表明,当前蒲城县耕地土壤有机质质量比平均值为13.58 g/kg,全氮质量比为0.81 g/kg,均处于偏低水平,空间上整体随地貌自北向南递增;土壤碳氮比平均值为9.84,整体呈南北高中间低格局。土壤有机质与全氮是相关性密切的本征协同区域化变量,均具有极显著性的空间自相关性,表现出中等强度的空间变异性;其空间变异性是地形地貌、土壤性质等自然因素和种植模式、施肥、灌溉等人为因素交互作用的结果,其中人为耕作管理的地理区位导向作用愈发明显。相同采样数量下,以有机质为辅助变量的全氮协同克里格法插值精度优于普通克里格法,适于当下蒲城县土壤全氮的估值研究,能够提供更多局部细节信息,但其不适于进行主辅变量的栅格比值运算来构建土壤碳氮比值图。     

Soil surface water repellency induced by treated wastewater irrigation: physico-chemical characterization and quantification

Irrigation with treated wastewater (TWW) is commonly practiced in Israel to relieve freshwater (FW) shortages. We hypothesized that the organic matter (OM) originating from TWW irrigation alters the physico-chemical properties of the soil, induces water repellency in the soil’s top layer, and consequently alters water distribution in the soil profile. In measurements taken in an avocado orchard on a clayey soil, water repellency was found in TWW-irrigated plots. In addition, smaller wetted surface areas were recorded around the drippers in comparison with FW drippers. Drier zone below soil surface was observed in TWW-emitting drippers. OM extraction from the different plots exhibited differences in quantity and quality of organic substances between TWW- and FW-irrigated soils, with a higher quantity of hydrophobic substances in the TWW-irrigated soil extract.