Visible Near-infrared Spectra Research Articles

Application of Visible/Near-Infrared Spectra in Modeling of Soil Total Phosphorus

Overabundance of phosphorus (P) in soils and water is of great concern and has received much attention in Florida, USA. Therefore, it is essential to analyze and predict the distribution of P in soils across large areas. This study was undertaken to model the variation of soil total phosphorus (TP) in Florida. A total of 448 soil samples were collected from different soil types. Soil samples were analyzed by chemical reference method and scanned in the visible/near-infrared (VNIR) region of 350–2500 nm. Partial least squares regression (PLSR) calibration model was developed between chemical reference values and VNIR values. The coefficient of determination (R) and the root mean squares error (RMSE) of calibration and validation sets, and the residual prediction deviation (RPD) were used to evaluate the models. The R in calibration and validation for log-transformed TP (log TP) were 0.69 and 0.65, respectively, indicating that VNIR calibration obtained in this study accounted for at least 65% of the variance in log TP using only VNIR spectra, and the high RPD of 2.82 obtained suggested that the spectral model derived in this study was suitable and robust to predict TP in a wide range of soil types, being representative of Florida soil conditions.

Coordinated spectral and XRD analyses of magnesite‐nontronite‐forsterite mixtures and implications for carbonates on Mars

Mineral detection on Mars largely relies on laboratory data of minerals and mineral mixtures. The objective of this study is to provide reflectance spectra in the visible/near‐infrared (VNIR) and mid‐IR regions, X‐ray diffraction (XRD) data and Mössbauer spectra of a suite of carbonate, phyllosilicate and olivine mixtures in order to facilitate identification and characterization of these minerals on Mars. Remote sensing observations indicate that combinations of these minerals are present in ancient rocks on Mars around the Isidis Basin and in Gusev crater. Magnesite, nontronite, and forsterite size fractions <125 µm were selected for this study. Results of the VNIR reflectance analyses illustrate the complexity of VNIR spectra of mixtures. Analyses of the NIR band depths near 2.3, 2.5, 3.4, and 4 µm showed clear trends with carbonate abundance, although the data are not linear. Mixtures of magnesite and nontronite exhibited a band near 2.3 µm much closer to that observed for nontronite than that for magnesite. VNIR analyses of the mixtures indicated that a small amount of forsterite in any of the mixtures contributed a large increase in the broad ~1 µm band and, hence, the red slope characteristic of Fe2+‐bearing minerals. Mid‐IR mixture spectra were dominated by magnesite and forsterite, and nontronite was much more difficult to detect by mid‐IR spectra in the mixtures. This could be related to why phyllosilicates are detected in many locations on Mars using data collected by the Compact Reconnaissance Imaging Spectrometer for Mars, but not detected using data collected by the Thermal Emission Spectrometer. Mössbauer spectroscopy is well suited for analyses of Fe2+‐ and Fe3+‐bearing minerals, and modeling of the peak areas gave well‐correlated trends for nontronite and forsterite abundances where abundant Fe was present. XRD full‐pattern fitting analyses were performed on the magnesite‐forsterite series, giving results within 6 wt % of the actual values, with a mean difference between actual and calculated values of 2.4 wt %. This study provides important laboratory data for characterizing the spectral and XRD properties of mineral mixtures that will facilitate mineral identification on Mars. Carbonates, in particular, have been primarily observed at low abundances and in small outcrops, and they are frequently found mixed with other minerals. Through analyses of mineral mixtures using multiple data sets, this study seeks to provide ground truthing that will enable better coordination of carbonate detections in the dust and rocks of Mars.

On-line prediction of beef quality traits using near infrared spectroscopy

The aim of this study was to test the ability of visible-near infrared (Vis-NIR) reflectance spectroscopy to predict beef quality traits in the slaughterhouse by directly applying a fiber-optic probe on the carcass surface. Carcasses from 230 young bulls and heifers slaughtered in two commercial abattoirs were included in the experiment. Vis-NIR spectra were recorded on an exposed surface of M. gracilis in the abattoirs 4 to 6 and 14 to 16h post mortem. Traits evaluated were pH, color indexes (L*, a*, b*, H, and SI), cooking loss, and Warner–Bratzler shear force. Prediction models were satisfactory for pH and color indexes, and promising for cooking loss but not for Warner–Bratzler shear force. Results of this work show that Vis-NIR spectroscopy may be a useful tool for on-line prediction of some physical beef quality traits when applied directly on the carcass surface.

Doping single-walled carbon nanotubes with surfactant peptides containing electron-donor substituents and nitrogen heterocycles

We examined the potential of a series of aromatic moieties with different electron-donating ability to alter SWCNT electronic properties. The selected aromatic moieties included p-amino-phenylalanine, 3-[4-(dimethylamino)phenyl]propanoic acid, tyrosine, tryptophan, and 4-pyridylalanine. These moieties contained exocyclic electron-donor substituents, such as amine, dimethylamine, and hydroxyl groups, as well as nitrogen-containing heterocycles, including indole and pyridine. We attached these aromatic molecules to the N-terminus of an amphiphilic surfactant peptide and obtained stable aqueous peptide/SWCNT dispersions. Atomic force microscopy images and ultraviolet–visible–near-infrared spectra revealed that, under the conditions used, all peptides, except for the one with 4-pyridylalanine, disperse SWCNTs well in solution. The local electron density of states of the peptide-coated SWCNTs was examined using scanning tunneling spectroscopy dI/dV plots. All spectra showed an additional peak in the conduction band side very close to the Fermi level, indicating n-type doping of the SWCNTs. The doping behavior was further verified by Raman spectroscopy G-band peak downshifts. Additionally, we found that the location of the heteroatom contributes significantly to the interaction between the aromatic moieties and the SWCNTs.

Near-Field Interactions between Metal Nanoparticle Surface Plasmons and Molecular Excitons in Thin-Films. Part II: Emission

We systematically study the effect of near-field plasmon–exciton interactions between metal nanoparticles (NPs) and molecular thin-films. In part I, we discussed the effect of these interactions on the absorption of the organic molecules, whereas in this part, we probe their impact on the emission of the organic molecules by time-resolved and time-averaged photoluminescence measurements. Through carefully designed experiments, we are able to exclusively probe the effect of the presence of metal NPs on the emission of organic molecules while eliminating their influence on molecular excitation. For this study, we utilize Ag NPs in a size regime where absorption dominates light scattering, as well as several organic materials that emit at different wavelengths throughout the ultraviolet–visible–near-infrared spectrum. By introducing a transparent, nonemissive spacer layer, we are able to accurately probe the range of plasmon–exciton interactions. We find a good qualitative agreement between the photoluminescence measurements and calculations based on an analytical model. In general, we observe a strong reduction of the emission quantum yield in the presence of the Ag NPs within a distance of <30 nm. However, for poor emitters in combination with embedding media with a high permittivity, we find an emission enhancement due to the Ag NPs. Finally, we compare the results obtained from both parts of this study to discuss the potential application of near-field plasmon–exciton interactions in thin-film opto-electronic devices and sensors.

Mineralogy and chemistry of altered Icelandic basalts: Application to clay mineral detection and understanding aqueous environments on Mars

We used a suite of techniques, including those emulating compositional data sets obtained from Mars orbit and obtainable at the Mars surface, to examine aqueous alteration of basaltic rocks from Iceland as a mineralogic and geochemical analog for Noachian environments on Mars. A sample suite was collected for laboratory measurement of (1) whole‐rock visible/near‐infrared (VNIR) reflectance and thermal infrared (TIR) emission spectra; (2) VNIR and TIR reflectance spectra of particle‐size separates derived from the bulk rock and from materials extracted from fractures/vesicles; (3) X‐ray diffraction (XRD) patterns for determination of quantitative modal mineralogy; (4) major element chemistry using flux fusion of whole‐rock powders; and (5) electron microprobe analyses of minerals in thin sections. Conclusions about aqueous alteration can be influenced by technique. For these basalts, whole‐rock chemical data showed scant evidence for chemical fractionation, but TIR, VNIR, and XRD measurements identified distinctive assemblages of hydrous silicate minerals, differing by sample. XRD provided the most complete and accurate quantitative determination of sample mineralogy. However, VNIR spectroscopy was the technique most useful for determining composition of low‐abundance smectite clays, and TIR spectroscopy was the most useful for recognizing hydrated silicates in thin surface coatings. High spatial resolution mineralogical and chemical data sets were useful for understanding the texture and distribution of alteration products and variations in fluid chemistry. No single approach provides a complete assessment of the environment of alteration, demonstrating the importance of employing multiple, synergistic mineralogical and geochemical techniques and instruments in exploration of rock strata from aqueous paleoenvironments on Mars.

Soil total carbon analysis in Hawaiian soils with visible, near-infrared and mid-infrared diffuse reflectance spectroscopy

Accurate assessment of total carbon (Ct) content is important for fertility and nutrient management of soils, as well as for carbon sequestration studies. The non-destructive analysis of soils by diffuse reflectance spectroscopy (DRS) is a potential supplement or alternative to the traditional time-consuming and costly combustion method of Ct analysis, especially in spatial or temporal studies where sample numbers are large. This alternative technique has been utilized in several locations in the United States and elsewhere, but has not been tested on the unique and diverse tropical soils in Hawaii. This study investigated the feasibility of DRS for Ct prediction of Hawaiian agricultural soils by creating visible, near-infrared (VNIR) and mid-infrared (MIR) spectral libraries and developing chemometric models with partial least squares regression (PLSR) and random forests (RF) ensemble tree regression. The sample set contained 305 soils from across the five main Hawaiian Islands, representing 10 soil orders and over 100 soil series. The Ct of these samples measured by dry combustion ranged from <1% to 56%, greater than that of most previously published studies. The VNIR spectra of the soils commonly exhibited features associated with OH− and H2O, iron oxides, phyllosilicates, and organic molecules. The numerous features in the MIR spectra can be attributed to OH−, organic molecules, and a variety of silicate minerals, such as phyllosilicates and quartz. Excellent results were obtained from both PLSR models using VNIR and MIR spectra. The models had R2 values of 0.95/0.94 (VNIR/MIR), root mean squared error (RMSE) values of 2.80%/3.08%, residual prediction deviation (RPD) values of 4.25/3.91, and ratio of performance to inter-quartile distance (RPIQ) values of 3.66/3.38. The RF VNIR and MIR models were also of high quality, with R2=0.95/0.96 (VNIR/MIR), RMSE=2.82%/2.28%, RPD=4.27/5.28, and RPIQ=3.68/4.56. An evaluation of the different wavelength ranges and chemometric methods determined that for this sample set all were capable of robust predictions of Ct. These results indicate that DRS is an alternative technique for analysis of Ct in Hawaiian soils.

High Efficiency Si/CdS Radial Nanowire Heterojunction Photodetectors Using Etched Si Nanowire Templates

p-Si/n-CdS radial heterojunction nanowires have been grown by pulse laser deposition of CdS on vertically aligned Si nanowires fabricated using a room temperature wafer-scale etching of p-type Si. Temperature-dependent photoluminescence characteristics have been studied in detail in the blue–green–red regions from these p-Si/n-CdS core–shell nanowires. The photocurrent spectra of the nanowire heterojunctions have been investigated at room temperature to study the spectral responsivity and detectivity of the core–shell nanowire diodes. The peak responsivity (1.37 A/W) and detectivity (4.39 × 1011 cm Hz1/2/W) at −1 V show the potential of the nanoscaled devices for the high efficiency photodetectors in the visible–near-infrared spectrum.

Application of Particle Swarm Optimization (PSO) Algorithm to Determine Dichlorvos Residue on the Surface of Navel Orange with Vis-NIR Spectroscopy

There is increased interest in the investigation and implementation of rapid, non-destructive methods for detection of pesticide residues. This study reports development of a method for determination the pesticide concentration on the surface of fruit samples by using visible near-Infrared (Vis-NIR) spectroscopy. A total of 330 navel oranges were selected for the calibration (n=220) and prediction (n=110) sets, and sprayed dichlorvos solution diluted with tap water on the surface of fruit samples. After 12 h, the spectra of fruit samples were acquired in the range of 350-1800 nm, then pesticide residue was measured using gas chromatography (GC). Chemometrics analysis using Partial Least Squares (PLS) was performed in order to determine the concentration of pesticide residue. The particle swarm optimization (PSO) was used for wavelengths selection of Vis-NIR spectra data in order to develop the PLS prediction model. The model using optimal intervals for predicting the concentration of dichlorvos residue selected by PSO algorithm yielded better results than the PLS model using full wavelengths. This PSO-PLS model could predict dichlorvos residue of 220 samples of prediction set with correlation coefficient of 0.8732. The wavelength selection through a PSO algorithm would enhance the predictive ability when applying PLS model.

Quantification of Total Soluble Solids and Titratable Acidity For Citrus Maturity Using Portable Vis-NIR Spectroradiometer

Soluble solids contents (SSC) and titratable acidity (TA) are the two major components in the citrus industry that have a direct impact on the crop value and the juice quality. It is crucial for growers to know the stage of maturity in-field before harvesting and to know the fruit quality for the juice processing in the processing plant. This study presents an application of a commercially available visible-near infrared (Vis-NIR) spectroradiometer for estimating SSC and TA of Hamlin and Valencia orange, and Thompson Red grapefruit. The Vis-NIR spectra of the juice were acquired using a portable spectroradiometer within the wavelength range of 350 to 2,500 nm in transmission mode. The absorbance spectra were compared to the values of SSC and TA as analyzed using a bench top refractrometer and chemical titration, respectively. Calibration models relating the Vis-NIR spectra of the juice to its SSC and TA were developed based on two different regression analyses: principal component regression (PCR) and partial least squares regression (PLS-R). Selected combinations of spectral preprocessing such as derivatives, mean centering, multiplicative signal correction, standard normal variate, and detrending were applied to evaluate the performance of the calibration models. Performance of five different models in predicting the SSC and TA using the Vis-NIR ranges were compared. The best R2 and root mean square error of prediction (RMSEP) for SSC were 0.91 and 0.47Brix developed by Valencia orange absorbance spectra (Model 5), respectively. Meanwhile, the combination of Hamlin and Valencia absorbance spectra (Model 2) predicted the maximum TA with an R2 of 0.73 and an RMSEP of 0.25%. The results concluded that portable Vis-NIR spectroradiometer-based technique can provide basic information on the levels of SSC and TA in-field as a preliminary analysis before harvesting.

Discrimination of Australian soil horizons and classes from their visible-near infrared spectra

Summary A soil’s reflectance spectrum in the visible and near infrared is rich in information. It is an integrative property of the soil that measures its colour, the abundance of iron oxides, clay minerals and carbonates, the amount of water and organic matter and its particle size. We explored the merit of discriminating between soil horizons, of which we had 13 654 samples, and soil orders from the Australian Soil Classification, of which we had samples from 1697 profiles with designated horizons, by analysing quantitatively their diffuse reflectance spectra in the visible-near infrared (vis–NIR) range (350–2500 nm). We re-sampled the spectra to 10-nm intervals and converted them to logarithms of deviations from their convex hulls. We then transformed them to canonical variates, which we display as scatter diagrams in loworder canonical planes. The minerals, colour and organic constituents thought to be responsible for their discrimination are identified. Each spectrum was re-examined and allocated to the group whose centroid was nearest in the canonical space. Topsoil horizons (A, A2 and transitional AB and AC horizons) were distinguishable from subsoils (B, C and transitional BC horizons). Vertosols, Ferrosols, Podosols (Podzols), Organosols, Calcarosols, Rudosols, Sodosols, Hydrosols, Kandosols and Kurosols were in general well separated from other soil orders and were assigned to their own orders such that 80% or more were correctly allocated. These orders possess characteristics that are easily distinguished by vis–NIR spectroscopy. Re-allocations to other orders were interpretable and could be related to identifying features of the ASC classification. Our results show that spectra distinguish soil horizons and soil type. They suggest that vis–NIR spectroscopy could make an important contribution to the definition and identification of classes in an effective system of soil classification.

Spectral response of rice (Oryza sativa L.) leaves to Fe2+ stress

In the management of lake eutrophication, the regulation effect of Fe is considered, in addition to the controlling nitrogen- and phosphorus input. Based on the "Fe hypothesis", this paper tentatively applied plant spectral response to the remote sensing early-warning mechanism of lake eutrophication. A laboratory water culture experiment with rice (Oryza sativa L.) was conducted to study Fe uptake by plants and the chlorophyll concentration and visible-near infrared spectrum of vegetable leaves as well as their interrelations under Fe(2+) stress. Three spectral indices, i.e., A (1) (integral value of the changes of spectral reflectivity in the range 460-670 nm under Fe(2+) stress), A (2) (integral value of the changes of spectral reflectivity in the range of 760-1000 nm under Fe(2+) stress) and S (blue-shift range of red edge curve under Fe(2+) stress), were used to establish quantitative models about the relationships between the rice leaf spectrum and Fe(2+) stress. With the increase of Fe(2+) in a culture solution, the Fe content in rice plants increased, while the chlorophyll concentration in vegetative leaves decreased. The spectral reflectivity of vegetable leaves increased in the visible light band but decreased in the near infrared band, and the blue-shift range of the red edge curve increased. The indices A (1), A (2) and S all had significant correlations with the Fe content in rice leaves, the correlation coefficient being respectively 0.951 (P < 0.01), -0.988 (P < 0.01) and 0.851 (P < 0.01), and simulated (multiple correlation coefficients R (2) > 0.96) and predict the Fe level in rice leaves.

Simulated in situ characterization of soil organic and inorganic carbon with visible near-infrared diffuse reflectance spectroscopy

Visible and near-infrared (VisNIR, 400–2500 nm) diffuse reflectance spectroscopy (DRS) is a rapid, inexpensive sensing method that has shown promise for lab-based soil characterization. However, little has been reported on how DRS will work in a field setting on intact soil cores. Seventy-two soil cores, representing 21 soil series and four parent materials, were extracted from six fields in Central Texas. Each soil core was scanned with a VisNIR spectrometer with a spectral range of 350–2500 nm in four combinations of moisture content and pre-treatment, including field-moist intact, air-dried intact, field-moist smeared intact, and air-dried ground. Visible near-infrared spectra were then used to predict soil organic and inorganic carbon (C) using partial least squares (PLS) regression. The PLS model was validated with 30% of the original soil cores that were randomly selected and withheld from the calibration model. The organic C validation had a root mean squared deviation (RMSD) of 5.4 g kg − 1 and 4.1 g kg − 1 for the field-moist and air-dried intact scans, respectively. The RMSD values for inorganic C were 8.7 g kg − 1 and 7.8 g kg − 1 for the field-moist and air-dried intact scans, respectively. Smearing the samples had minimal effects on prediction accuracies for organic and inorganic C. Variable soil moisture did reduce prediction accuracies. Soil color, pH, and soil reaction to 1 N HCL were added as auxiliary predictors. Soil color improved organic C predictions by 0.2 to 0.4 g kg − 1 . The field-moist intact inorganic C model improved with soil color (RMSD = 8.0 g kg − 1 ), soil pH (8.3 g kg − 1 ), and soil reaction to HCl (6.5 g kg − 1 ). These results show that in situ spectroscopy can measure organic and inorganic C with some loss of accuracy compared to dried ground samples. In inorganic C predictions, an easy-to-measure auxiliary variable, like soil reaction with 1 N HCl, can improve in situ predictions compared to dried ground predictions.

Direct Comparison between Visible Near- and Mid-Infrared Spectroscopy for Describing Diuron Sorption in Soils

Both visible near-infrared (VNIR) and mid-infrared (MIR) spectroscopy have been claimed to better predict pesticide sorption in soils than other methods. We compared the performances of VNIR and MIR spectroscopy for predicting both organic carbon content (foc) and the sorption affinity (Kd) of diuron in 112 surface soils from South Australia. Separate calibration models were developed between VNIR and MIR spectra, and foc and Kd using partial least-squares (PLS) regression. MIR clearly outperformed VNIR for predictions of both foc and Kd in soils. Correlation (R2) and accuracy (RPD) indices were 0.4 and 1.3 for the VNIR-PLS model versus 0.8 and 2.3 for the MIR-PLS model, respectively, for Kd prediction. PLS loadings for sorption prediction were compared in terms of the soil information they contained. While VNIR loading did not include any direct spectral information regarding soil minerals, MIR loading included peaks associated with sand, clays, and carbonates. Perhaps by better predicting foc and integrating the effects of OC as well as minerals, the MIR-PLS model provided a better prediction for diuron Kd values in our calibration set.

Relationship between wine scores and visible–near-infrared spectra of Australian red wines

Sensory analysis of wine involves the measurement, interpretation and understanding of human responses to the properties perceived by the senses such as sight, smell and taste. The sensory evaluation of wine is often carried out by wine judges, winemakers and technical staff, and allows characterization of the quality of the wine. However, this method is lengthy, expensive, and its results depend on panel training and the specific vocabulary used by the panel. A robust, rapid, unbiased and inexpensive method to assist in quality assessment purposes will therefore be beneficial for the modern wine industry. This study aims to investigate the relationship between sensory analysis, visible (VIS) and near-infrared (NIR) spectroscopy to assess sensory properties of commercial Australian wine varieties. For the purposes of this study 118 red wine samples (Cabernet Sauvignon, Shiraz, Pinot Noir, Tempranillo, Nebbiolo and blends) graded by a panel of experienced tasters and scored according to the Australian wine show system were scanned in transmission in the VIS and NIR range (400-2,500 nm). Partial least squares regression models were developed between the overall score given by the judges and the combined VIS-NIR spectra, using full cross validation (leave-one-out method). The results showed that NIR spectroscopy was able to predict wine quality scores in red wine samples (R = 0.61 and standard error of prediction of 0.81). The practical implication of this study is that instrumental methods such as VIS-NIR spectroscopy can be used to complement sensory analysis and can facilitate the task at early stages of product development, making high-throughput screening of novel products feasible or maintaining the consistency of the product.

The Effect of Scanning Angle on Measured Reflectance and Tenderness Prediction in Beef Longissimus Thoracis Steaks

Near-infrared (NIR) reflectance spectroscopy has previously demonstrated potential to quantify tenderness and sensory attributes of meat. The objective of this study was to test the effect of scanning angle on visible/near-infrared (VNIR) spectral reflectance and the ability of reflectance spectra to predict tenderness of longissimus thoracis (LT) steaks. Two experiments were conducted to evaluate the effect of scanning angle configuration on steak reflectance. The first experiment included 60 steaks from two quality grades, aged for either 7 or 14 d, which were randomly assigned to two treatments in which the steak angles with respect to the light and sensor were changed. In the second experiment, 16 steaks from the two quality grades were scanned at six different angles. The Warner-Bratzler shear force (WBSF) was measured as a quantitative measure of tenderness. There were significant amplitude differences of up to 50% in the average reflectance spectra between the two treatments in experiment 1, and up to 200% difference between the reflectance spectra when the steak was tilted from 30 counterclockwise to 30 clockwise. First derivatives of the reflectance spectra also showed significant differences with respect to scanning angle. Chemometric models on VNIR spectra could predict only 19% of the variability in tenderness when the scanning angle was not considered. However, when individual models were developed for each scanning angle-aging group, the models accounted for 20% to 86% of the variability in tenderness. This study, with its limited data and tenderness ranges, showed that scanning angle is an important factor affecting the accuracy of reflectance-based tenderness models.

Highly c-Axis Oriented Al-Doped ZnO Thin Films Grown in Premixed H2/Ar Sputtering Gas

Al-doped ZnO (AZO) thin films were grown on Corning 1737 glass by RF Magnetron Sputtering under premixed hydrogen-argon (H2/Ar) sputtering gas. It is found that the introduction of various H2 concentrations during sputtering deposition altered the properties of Al-doped ZnO films. The presence of H2 during AZO growth at low deposition temperature leads to the growth of a-axis preferential orientation crystal whereas c-axis preferential orientation occurred only at higher deposition temperature. Highly oriented c-axis (002) crystal has been successfully grown under 3% H2 concentration at 200°C deposition temperature. Film’s resistivity is appeared to be a function of H2 concentration. Additional H2 concentration in sputtering gas increased of film’s transmittance up to 85% at visible-near infra red spectra while it caused the Burstein-Moss shift toward the blue region at 350 nm wavelength.

The prediction of total anthocyanin concentration in red-grape homogenates using visible-near-infrared spectroscopy and artificial neural networks

This study compares the performance of partial least squares (PLS) regression analysis and artificial neural networks (ANN) for the prediction of total anthocyanin concentration in red-grape homogenates from their visible-near-infrared (Vis-NIR) spectra. The PLS prediction of anthocyanin concentrations for new-season samples from Vis-NIR spectra was characterised by regression non-linearity and prediction bias. In practice, this usually requires the inclusion of some samples from the new vintage to improve the prediction. The use of WinISI LOCAL partly alleviated these problems but still resulted in increased error at high and low extremes of the anthocyanin concentration range. Artificial neural networks regression was investigated as an alternative method to PLS, due to the inherent advantages of ANN for modelling non-linear systems. The method proposed here combines the advantages of the data reduction capabilities of PLS regression with the non-linear modelling capabilities of ANN. With the use of PLS scores as inputs for ANN regression, the model was shown to be quicker and easier to train than using raw full-spectrum data. The ANN calibration for prediction of new vintage grape data, using PLS scores as inputs, was more linear and accurate than global and LOCAL PLS models and appears to reduce the need for refreshing the calibration with new-season samples. ANN with PLS scores required fewer inputs and was less prone to overfitting than using PCA scores. A variation of the ANN method, using carefully selected spectral frequencies as inputs, resulted in prediction accuracy comparable to those using PLS scores but, as for PCA inputs, was also prone to overfitting with redundant wavelengths.

Chain of Custody for Samples of Live Crude Oil Using Visible—Near-Infrared Spectroscopy

In order to design oil production facilities and strategies, it is necessary to acquire crude oil samples from subsurface formations in oil wells in so-called openhole prior to production. In some environments, such as deepwater production of oil, decisions of huge economic importance are based on such samples. To date, there has been little quality control to verify that the crude oils collected in the sample bottles and analyzed up to a year later in the laboratory have any relation to the actual crude oils in the subsurface reservoirs. These high-pressure samples can undergo myriad deleterious alterations. Here, we introduce the chain-of-custody concept to the oilfield. The visible-near-infrared spectrum of the crude oil is measured in situ in the wellbore at the point of sample acquisition. This spectrum is compared with the spectrum measured on putatively the same fluid in the laboratory at the start of laboratory sample analysis. First, quantitative assessment is made of whether the fluid in the (high-pressure) sample bottle remains representative of formation fluids. Second, any specific changes in the spectrum of the fluid can be related to possible process control failures. Here, the entire process of chain of custody is proven. The chain of custody process can rapidly become routine in the petroleum industry, thereby significantly improving the reliability of any process that depends on fluid property determination.

Finding mineralogically interesting targets for exploration from spatially coarse visible and near IR spectra

Spectroscopic studies of analog terrestrial mineral assemblages are necessary to develop criteria to identify similar environments on Mars. We use visible/near infrared (VNIR) laboratory, field, and remotely acquired spectral data to identify the iron-bearing and hydrous minerals of Rio Tinto, Spain, an astrobiological analog. Mineralogy evolves from iron sulfate- and oxide-rich (jarosite, rozenite, gypsum, schwertmannite, copiapite, goethite, and hematite assemblages) in young sediments to hydrated iron oxides in preserved terraces.Using spectra from the Rio Tinto, we examine one of the key challenges of extraterrestrial exploration: how to identify promising targets from spatially coarse data for in situ investigation. We apply an index to quantify the expression of spectral diversity as a function of spatial scale from hand sample to landscape. To validate this method for use at the decimeter orbital scale, we apply the index to cm-scale point spectra and meter-scale gridded spectra collected in the field. This exercise in spatial scaling gives increased confidence in the ability of the Spectral Variance Index (SVI) method to locate regions with increased mineral diversity from remotely sensed data. We divide the remotely sensed data into 25×25 pixel (200 m×200 m) cells and calculate the average mean (albedo) and spectral variance over all wavelengths for each cell. We next calculate the expected variance for each cell with a linear regression between mean and spectral variance. The number of standard deviations of each cell's spectral variance is from the expected variance is the SVI value. We locate ∼20 areas with high SVI values within the tailing piles and along wide riverbanks downstream of the active mine. This method uses spatially coarse VNIR spectra to recognize areas in Rio Tinto that would be ideal targets for future field exploration, and could also be applied to Mars orbital spectral datasets, such as OMEGA and CRISM.