P.P. Subedi

Assessment of Potato Dry Matter Concentration Using Short-Wave Near-Infrared Spectroscopy

The utility of short-wavelength near-infrared spectroscopy (over the wavelength region 750–950 nm), used in a partial transmittance optical geometry, was assessed as a means of estimating the dry matter concentration of potato tubers. The sampling optics did not involve contact with the sample, and could be used on a moving stream of product. A prediction accuracy of R2 (correlation coefficient of determination) of 0.85 with a root mean square error of prediction (RMSEP) of 1.52% for intact, whole tubers and R2 = 0.95 and RMSEP = 0.50% for sliced tubers was achieved. We conclude that short-wavelength near-infrared technology using a partial transmittance optical sampling geometry can be a useful tool for rapid assessment of tuber dry matter concentration prior to processing.

Assessment of Titratable Acidity in Fruit Using Short Wave near Infrared Spectroscopy. Part B: Intact Fruit Studies:

Interactance spectra (700–1100 nm) of intact fruit possess features interpreted as being due to the dilution and perturbation of water OH stretching and combination bands (with the second and third overtones of OH stretching at 960 nm and 740 nm, respectively), while the contribution from citric acid CH2 and OH stretching and combination bands are not obvious. A model developed using interactance spectra collected from the cut surfaces of lime fruit (mean ± SD of 7.3 ± 0.51 g citric acid equivalents 100 mL−1, units subsequently presented as% w/v) achieved prediction results of RMSEP=0.16%, rp2 = 0.79, bias = −0.03%. However, for intact lime fruit, model calibration results (RMSECV=0.16, rcv2 = 0.85) were markedly better than prediction results (RMSEP=0.30, rp2 = 0.49). For a low total titratable acidity (TTA) product, peach (with spectra collected across fruit maturity stages; mean ± SD of 0.88 ± 0.17%), calibration results were relatively poor (RMSECV=0.09%, rcv2 = 0.79) and the model failed in prediction (RMSEP=0.10%, rp2 = 0.00, bias = 0.02%). It is concluded that interactance geometry shortwave (700–1100 nm) near infrared spectroscopy using diode array instrumentation and an interactance optical geometry suited to on-line or field-portable instrumentation used for Brix and DM assessment is not appropriate for assessment of the acidity of intact low TTA fruit and has limited use for high TTA fruit.

Robustness of partial least-squares models to change in sample temperature: II. Application to fruit attributes

Partial least-squares regression models were developed using spectra of tomato fruit collected at 15°C and tested on spectra of an independent set of fruit at higher sample temperatures. The influence of sample temperature on the model used to predict fruit dry matter(DM) was manifested primarily in terms of bias, not standard error of prediction. For example, a model for DM created with samples at 15°C had a bias of −0.9% DM and −1.9% DM when used to predict DM in fruit at 25°C and 35°C, respectively. The addition of spectra of a relatively small number of samples collected at different temperatures to the calibration set can create a model that is robust to temperature; however, continued addition of samples at a uniform temperature overwhelms this compensation effect, resulting in a model that is not robust to temperature. For a model that included spectra of fruit at a range of temperatures, the prediction bias increased as the ratio of samples at 15°C to samples at other temperatures increased beyond 200:1. The use of orthogonal scatter correction, external parameter orthogonalisation, generalised least-square weighting, global model development and repeatability file were compared for the development of a temperature-robust DM model. The use of a repeatability file is recommended on the basis of the lowest root mean square error of prediction and bias. Selection of wavelength regions to avoid water absorption features is recommended for an attribute not associated with an OH feature (such as skin-colour prediction).

Robustness of Partial Least-Squares Models to Change in Sample Temperature: I. A Comparison of Methods for Sucrose in Aqueous Solution

Sample temperature is well known to impact model performance for prediction of chemical attributes in high-moisture-content samples when using short-wave near infrared spectroscopy. A number of methods proposed to reduce this effect were compared in this study. A short-wave near infrared spectroscopy system operating in transflectance geometry was used to record spectra of sucrose solutions (mean = 4.16% and SD = 5.8% w/v) at different temperatures. Partial least-squares regression models were developed using spectra of sucrose solutions collected at 15°C and tested on spectra of an independent set of sucrose solutions at a sample temperature of 35°C. As sample temperature impacts the water peak, the performance of a model of sucrose content is perturbed, mainly through an increase in bias. Addition of a relatively small number of spectra of the same set of samples at different temperatures facilitated a model robust to temperature, but continued addition of samples at 15°C beyond the ratio of 1:125 overwhelmed the compensation effect, resulting in a model that was not robust to temperature. The use of orthogonal scatter correction (OSC), generalised least square weighting (GLSW), external parameter orthogonalisation (EPO) and repeatability file were considered. Of these methods, OSC corrected bias but impacted bias corrected root mean square error of prediction (SEP) and r2, EPO performed better but still with some bias, GLSW gave the best r2 and SEP result but still with bias, while use of the repeatability file method gave the best overall result.

Robustness of Tomato Quality Evaluation Using a Portable Vis-SWNIRS for Dry Matter and Colour

The utility of a handheld visible-short wave near infrared spectrophotometer utilising an interactance optical geometry was assessed in context of the noninvasive determination of intact tomato dry matter content, as an index of final ripe soluble solids content, and colouration, as an index of maturation to guide a decision to harvest. Partial least squares regression model robustness was demonstrated through the use of populations of different harvest dates or growing conditions for calibration and prediction. Dry matter predictions of independent populations of fruit achieved R2 ranging from 0.86 to 0.92 and bias from −0.14 to 0.03%. For a CIE a⁎ colour model, prediction R2 ranged from 0.85 to 0.96 and bias from −1.18 to −0.08. Updating the calibration model with new samples to extend range in the attribute of interest and in sample matrix is key to better prediction performance. The handheld spectrometry system is recommended for practical implementation in tomato cultivation.

Automated mango flowering assessment via refinement segmentation

An automated flowering assessment system for mango orchards was proposed. Segmentation of flowers from a complex background (i.e. leaves, branches and ground) was achieved based on (i) colour correction via adjustment of the brightness and contrast to a reference level, to rectify the illumination variability spatially within and between images; (ii) colour thresholding with fixed thresholds to separate flowers, although with some branches and trunks; and (iii) SVM classification to refine the segmentation results, removing the branch and trunk errors. Mango tree canopy images (n=160) were acquired during a five-week flowering period, with 15 of the images used in calibration and 145 used in validation. The proposed method had a good correlation with human scoring, with coefficient of determination (R2) of 0.87.

Spectrophotometer ageing and prediction of fruit attributes

Deterioration of lamp output quality over time and degradation of detector signal-to-noise ratio are issues associated with ageing of a spectrophotometer. To document the effect of instrument ageing on short wavelength near infrared spectroscopy-based assessment of internal attributes of fruit quality prediction (total soluble solids, TSS, of juice), an assessment was conducted of several handheld photodiode array-based spectrophotometers over several years, with repeated spectra of a reference Teflon (PTFE) tile and spectra of 20 apple fruit acquired at yearly intervals. The repeatability of each instrument was assessed as the standard deviation of absorbance of repeated measures of a reference, typically around 0.2 mAbs. Instrument changes were identified in performance and in principal components analysis plots, but performance (apple TSS model) was not related to instrument repeatability. A piecewise direct standardisation method is recommended to maintain a multivariate calibration model across spectrometers.

Quality estimation of Agave tequilana leaf for bioethanol production

Agave tequilana is a potential biofuel crop, for which the characters of juice total soluble sugar content (TSS), dry matter content (DM), cellulose, hemicellulose and lignin content are quality criteria. Spectra of leaves were obtained using a hand-held silicon photodiode array (Si PDA)-based spectrometer with a wavelength range of 300–1100 nm and an InGaAs-based Fourier transform near infrared (FT-NIR) spectrometer with a wavelength range of 1100–2500 nm. Fresh leaves were harvested at different maturity stages, in different seasons and from two locations in Queensland during 2012–2014. Partial least square regression models were developed for DM and TSS of fresh leaf, and for cellulose, hemicellulose and lignin of dried material, with models tested on populations of independent samples collected in different years, seasons and locations. Prediction statistics for DM of fresh leaf using the Si PDA spectrometer (729–975 nm) were r2 = 0.49–0.87 and root mean square error of prediction (RMSEP) = 2.36–1.44%, while with the use of the FT-NIR spectrometer, the prediction statistics were r2 = 0.53–0.66 and RMSEP = 2.63–2.18% (across different years, seasons and locations). Prediction statistics for TSS in fresh leaf using the Si PDA spectrometer (729–975 nm) were r2 = 0.53–0.69 and RMSEP = 1.70–1.91%, with poorer results obtained using the FT-NIR spectrometer (r2 = 0.33–0.56; RMSEP = 1.88–2.45%). With increased sample diversity in the calibration set, NIR technology is recommended for estimation of DM and TSS in fresh Agave leaves. FT-NIR-based prediction of cellulose, hemicellulose or lignin of independent sets (of different years or cultivars) was unsatisfactory, with r2 < 0.75 and bias >10% of mean. These results may be improved with increased sample range, and attention to laboratory (reference method) error. However, leaf cellulose and hemicellulose content may be more easily estimated through correlation to leaf DM level (R2 of 0.77 across all sampling events).