基于近红外技术的落叶松木材密度预测模型

运用近红外光谱对落叶松(Larix gemelinii Rupr)样品密度进行了研究,分别运用偏最小二乘法及主成分回归建立预测模型,并用建立的模型分别对每一个样品进行了预测。基于偏最小二乘法的校正模型及验证模型相关系数分别为0.964和0.918,校正标准误差及预测标准误差分别为0.016和0.021,模型预测值与实测值决定系数为0.93;主成分回归模型中,校正模型及验证模型相关系数分别为0.954和0.911,校正标准误差及预测标准误差分别为0.017和0.023,模型预测值与实测值决定系数为0.91。研究表明:基于主成分回归法与偏最小二乘法的近红外光谱分析建模,都可以实现对落叶松木材密度的有效预测,但相比较而言,偏最小二乘法略优于主成分回归法,所建立的模型对落叶松木材密度预测更加准确可靠。     

Rapid,non-destructive carbon analysis of forest soils using neutron-induced gamma-ray spectroscopy

Forest soils are pivotal to understanding global carbon (C) cycling and evaluating policies for mitigating global change. However, they are very difficult to monitor because of the heterogeneity of soil characteristics, the difficulty of representative sampling, and the slow time scale of response to environmental change. Here we demonstrate that use of gamma-ray spectroscopy facilitates in situ non-destructive analysis of C and other elements in forest soils. In this approach the element-specific gamma-rays are induced by fast and thermal neutrons interacting with the nuclei of the elements present in the soil. Background gamma-rays emanating from naturally occurring radionuclides in the forest are recorded as well. We applied this approach in a mature northern hardwood forest on glacial till soils at the Bartlett Experimental Forest in New Hampshire, USA. The inelastic neutron scattering (INS) system yielded strong signals in gamma-ray counts/h, from C and other elements present in the soil matrix that included silicon, oxygen, hydrogen, iron, aluminum, manganese and potassium. The INS sensitivity for carbon was 20.656 counts h⁻¹ kg⁻¹ C m⁻² based on current net C gamma-ray counts and the data for the O horizon and mineral soil to a depth of 30 cm obtained from a nearby quantitative soil pit (7.35 kg C m⁻²). We estimate the minimum detectable change to be ∼0.34 kg C m⁻², which is ∼5% of the current soil C content, and the minimum detectable limit to be ∼0.23 kg C m⁻². Eight % reproducibility from 11 measurements was limited, in part, by the large variability in the system counting geometry due to the uneven forest microtopography. The INS approach has the potential to revolutionize belowground monitoring of C and other elements, because the possibility of detecting a 5% change in forest soils has not been possible with destructive sampling methods.     

基于田间原位土壤含水量估测的可见/近红外光谱建模方法

为实时、准确地获取原位土壤含水量信息,利用可见/近红外光谱技术,分别使用全局偏最小二乘(PLS)建模、局部PLS建模方法,对田间原位土壤含水量进行快速估测。结果表明:全局PLS模型中,其建模集的决定系数(R~2)、交叉验证均方根误差(RMSECV)分别为0.943和1.750%,检验集的决定系数(R~2)、预测均方根误差(RMSEP)分别为0.956和1.260%。局部PLS模型中,分别比较了选取定标子集的2种方法(欧氏距离法和马氏距离法),采用欧氏距离法和马氏距离法选取定标子集进行建模的R~2值分别为0.974和0.979,RMSEP值分别为0.976%和0.943%。因此,将可见/近红外光谱技术应用到田间原位含水量测量是可行的,其中,使用局部建模方法的效果优于全局建模。     

FT-NIR快速检测制丝过程膨丝与叶丝掺配比例及均匀性

将膨胀烟丝与某品牌正常生产切丝后烟丝,分别按计算好的不同比例混合均匀后,应用傅立叶变换近红外光谱技术采集混合烟丝的近红外光谱,采用偏最小二乘法(PLS)建立预测膨胀烟丝与片烟烟丝掺配比例值的近红外模型,其相关系数是0.993 0,并进行了准确性和精密度试验。试验表明,FT-NIR光谱法可以很好地应用于膨胀烟丝与片烟烟丝掺配比例的快速检测,以监控制丝过程中膨胀烟丝与片烟烟丝的掺配均匀性。     

NIR Spectroscopy: An Alternative for Soil Analysis

Advances in laboratory instrumentation and chemometrics provide alternatives to traditional methods of conducting soil chemical analysis. One of these is infrared diffuse reflectance spectroscopy in the near-infrared spectral range (NIRS). Herein we report the results of a multinational study to develop useful calibrations associating NIRS spectra with laboratory-measured results for total soil carbon (C), total soil nitrogen (N), δ¹³C, and δ¹⁵N from a single soil site in Mexico subjected to zero- and conventional-tillage regimens with and without crop residues and crop rotations of maize and wheat across 16 years. Modified partial least squares regression (MPLS) was used to obtain useful NIR predictions for total soil C and N, with ratio performance deviation (RPD) values of 6.8 and 2.6, respectively. Corresponding multiple correlation coefficients (RSQs) for C and N were 0.98 and 0.85, with standard errors of prediction (SEPs) of ±0.45 g C kg^–1 and ±0.09g Nkg^–1, respectively. The generation of δ¹⁵N and δ¹³C models produced different NIR recordings in soils with and without crop residues. Application of discriminant partial least squares (DPLS) statistics to the NIR spectral data allowed us to discriminate soils with and without residues. The prediction confidence for stable isotopes was 90% (internal validation) and 94% (external validation). Modified partial least squares regression was used to estimate δ¹⁵N and δ¹³C. Ratio performance deviation, RSQ, and SEP values obtained for δ¹³C and δ¹⁵N were 2.44 and 3.57, 0.83 and 0.81, ±0.5‰ (parts per thousand) and ±0.45‰ in soils with residues and 2.5 and 3.8, 0.93 and 0.92, and ±0.2‰ and ±0.23‰ in soils without residues, respectively. Overall, results obtained with NIRS were comparable to those obtained using conventional analytical methods, a finding that has wide relevance to agricultural soils and environmental studies in tropical locations. However, further testing is necessary to confirm that the calibration models are neither site nor instrument specific.     

Microwave l-band (1730 MHz) accurately estimates the relative water content in poplar leaves. A comparison with a near infrared water index (R1300/R1450)

In this study the estimation of reflectivity at 1730 MHz (l-band), measured with a microwave digital cordless telephony (DCT) patch antenna, is presented as an easy-to-handle and non-destructive new method to assess the relative water content (RWC) of poplar leaves and filter discs at different levels of dehydration. The accuracy of this new method has been contrasted with the R₁₃₀₀/R₁₄₅₀ index, determined by a portable near infrared (NIR) spectrometer. The close correlations found between RWC and the reflectivity at a frequency of 1730 MHz, both for filters and leaves, indicate that microwave determinations are rather independent of the physical properties of the material analysed. On the contrary, the differences found between poplar leaves and leaf filters in the relationships established between RWC and the R₁₃₀₀/R₁₄₅₀ index demonstrate a strong influence of the properties of the material in NIR reflectance measurements, specifically as they relate to changes in leaf thickness during dehydration. It should be noted that the amount of energy received by the leaf for the microwave technique (0.1 mW) was much lower than that received for the measuring of the R₁₃₀₀/R₁₄₅₀ index (2.5 W). Moreover, R-square coefficients were higher for microwaves than for the R₁₃₀₀/R₁₄₅₀ index. The use of a technologically simple, low cost and portable device, based on a microwave DCT patch antenna, could yield a solid support for the development of a commercial apparatus enabling the determination of plant water status under field conditions.     

Measurement of genetic and environmental variation in barley (Hordeum vulgare) grain hardness

In this study, we assessed a broad range of barley breeding lines and commercial varieties by three hardness methods (two particle size methods and one crush resistance method (SKCS—Single-Kernel Characterization System), grown at multiple sites to see if there was variation in barley hardness and if that variation was genetic or environmentally controlled. We also developed near-infrared reflectance (NIR) calibrations for these three hardness methods to ascertain if NIR technology was suitable for rapid screening of breeding lines or specific populations. In addition, we used this data to identify genetic regions that may be associated with hardness. There were significant (p<0.05) genetic effects for the three hardness methods. There were also environmental effects, possibly linked to the effect of protein on hardness, i.e. increasing protein resulted in harder grain. Heritability values were calculated at >85% for all methods. The NIR calibrations, with R² values of >90%, had Standard Error of Prediction values of 0.90, 72 and 4.0, respectively, for the three hardness methods. These equations were used to predict hardness values of a mapping population which resulted in genetic markers being identified on all chromosomes but chromosomes 2H, 3H, 5H, 6H and 7H had markers with significant LOD scores. The two regions on 5H were on the distal end of both the long and short arms. The region that showed significant LOD score was on the long arm. However, the region on the short arm associated with the hardness (hordoindoline) genes did not have significant LOD scores. The results indicate that barley hardness is influenced by both genotype and environment and that the trait is heritable, which would allow breeders to develop very hard or soft varieties if required. In addition, NIR was shown to be a reliable tool for screening for hardness. While the data set used in this study has a relatively low variation in hardness, the tools developed could be applied to breeding populations that have large variation in barley grain hardness.     

奶粉中三聚氰胺含量的NIR检测技术

以奶粉为原料,采用偏最小二乘法(PLS)建立奶粉中添加的三聚氰胺含量的预测模型,旨在利用近红外分析方法无损、快速、准确的特点,探究其在奶粉掺假识别中应用的可行性。结果表明:大梯度模型覆盖范围广,模型参数Rc为0.9997,Rp为0.9996,RMSEP为0.245,RMSEC为0.275,RMSEC/RMSEP为1.122,模型线性关系良好,预测值与实际值拟合良好,在较大浓度情况下可以应用于未知样品中三聚氰胺含量的检测;小梯度预测模型受仪器检测限的影响,Rc为0.9991,Rp为1.0000,虽然线性关系良好,但是模型参数RMSEP为0.176,RMSEC为0.0275,RMSEC/RMSEP为0.156,预测值与实际值的拟合情况不理想,无法实现三聚氰胺含量的精确预测,实际应用中具有一定的局限性,但可以对其是否掺假正确定性。     

Comparison of predictions of baking volume using large deformation rheological properties

Three large deformation rheological tests, the Kieffer dough extensibility system, the D/R dough inflation system and the 2 g mixograph test, were carried out on doughs made from a large number of winter wheat lines and cultivars grown in Poland. These lines and cultivars represented a broad spread in baking performance in order to assess their suitability as predictors of baking volume. The parameters most closely associated with baking volume were strain hardening index, bubble failure strain, and mixograph bandwidth at 10 min. Simple correlations with baking volume indicate that bubble failure strain and strain hardening index give the highest correlations, whilst the use of best subsets regression, which selects the best combination of parameters, gave increased correlations with R²=0.865 for dough inflation parameters, R²=0.842 for Kieffer parameters and R²=0.760 for mixograph parameters.     

近红外光谱在转基因农产品检测中的研究进展

简述了近年来近红外光谱分析技术在农产品和食品定量分析中的应用,重点介绍了近红外光谱分析在转基因植株筛选以及转基因和非转基因农产品分类和识别等方面的研究现状,比较和分析了近红外光谱与传统的转基因农产品检测方法的特点,并展望了近红外光谱分析技术在转基因农产品检测研究中的发展前景。