An extensive summary of publications reporting radial specific gravity (SG) variation for North American conifers is presented. SG varies from pith to bark in three distinct patterns before reaching ...asymptomatic values: Type 1 (SG increases with cambial age), Type 2 (SG is initially high at the pith, then decreases with cambial age, before increasing), and Type 3 (SG decreases with cambial age). Pinaceae have either a Type 1 or 2 pattern (majority are Type 2). Cupressaceae have a Type 3 pattern, but radial SG data do not exist for some species. We reviewed publications in table 7-1 of Panshin and de Zeeuw (the 1970 edition) which reports SG variation by species and Type (the 1980 edition has an equivalent table but does not reference publications), examining sampling strategies and radial patterns. For a small number of species, Type was either incorrectly interpreted or sampling made a conclusion regarding pattern of radial variation impossible. Specific examples of mislabeled Types reported by Panshin and de Zeeuw for Douglas-fir (Pseudotsuga menziesii), lodgepole pine (Pinus contorta), red pine (Pinus resinosa) and shortleaf pine (Pinus echinata), and western redcedar (Thuja plicata) are described. We conclude for nearly all species the pattern of radial SG variation is consistent; however, different patterns have been reported for eastern white pine (Pinus strobus) and subspecies of Douglas-fir and ponderosa pine (Pinus ponderosa).
In this study, we used both nondestructive and destructive methods for assessing solid wood properties in six Vietnamese grown Eucalyptus clones at 6 years after planting. We measured stress wave ...velocity in standing sample trees (SWVT), logs (SWVL), and small clear specimens (SWVS) obtained from the trees and logs, and to measure static properties, we used MOE—modulus of elasticity and MOR—modulus of rupture. The highest average MOE and MOR were detected in clones 3 and 5, suggesting that these clones might be more appropriate for breeding programs focused on improving wood quality of Eucalyptus grown in Vietnam. Mean MOE and MOR of the lumber had significant (p < 0.001) relationships with SWVT (r = 0.61 and 0.53, respectively) and SWVL (r = 0.76 and 0.71, respectively). Stress wave velocity measurements of both standing trees and logs can be useful for further segregating Vietnam’s Eucalyptus timber resource based on MOE and MOR. For the small clear specimens, the best prediction of stiffness (dynamic modulus of elasticity (MOEd)) was obtained when both SWVS and air-dry density (AD) were used. The coefficient of correlation between MOE and MOEd was 0.93.
Near-infrared wavelengths selected by genetic algorithm were used to optimize partial least squares (PLS) regression models for loblolly pine (
Pinus taeda
L.) from the southeastern United States. ...Wood properties examined included density (D), microfibril angle, modulus of elasticity and tracheid coarseness (C), radial diameter (R), tangential diameter (T), and wall thickness (w)—measured by SilviScan. The optimization process was run for each property with Agenda 2020 samples utilized for PLS model development and the other sets used for prediction. The number of variables (i.e. wavelengths) varied from 10 to 100 with an optimum number identified by genetic algorithm. When compared to a full data set model (based on 700 wavelengths), calibration and prediction performance of optimized PLS regression models were superior for all properties. Importantly, representative wavelengths for each property were consistently related to recognized bond vibrations observed in specific wood components demonstrating that optimization targets wavelengths directly related to changes in wood chemistry within the examined loblolly pine samples.
The objective of this work was to provide a rapid and nondestructive imaging method for evaluating the hygroscopic behavior of thermally modified lignocellulosic materials (softwood and hardwood). ...The difference in the hygroscopic behavior was explained by moisture content (MC) mapping results and molecular association characteristics of absorbed water (i.e. weakly, moderately, and strongly hydrogen-bonded water molecules) with wood at various relative humidities (RH). To achieve this goal, near-infrared (NIR) spectral images in the wavelength range 1816–2130 nm (covering the combination of stretching and deformation vibrations for OH) were used to visualize MC distributions over the surface of Japanese cedar and European beech samples which had been thermally treated at different temperatures. A curve fitting method was utilized to explore changes in water-wood structure characteristics based on shifts to longer wavelength in spectral signals caused by increasing MC. The curve fitting results support the recent nuclear magnetic resonance (NMR) studies that different bound water stabilities may pool in different compartments of the wood cell wall. Furthermore, water was firmly bound to wood at low RHs and H-bonds gained mobility as the number of absorbed molecules increased. It is concluded that NIR hyperspectral imaging also has the potential to be a complementary methodology for studying the transient changes of wood-water interactions before equilibrium.
An optimization problem was developed by using a genetic algorithm to select wavelengths for establishing multivariate calibration models based on partial least squares (PLS) regression. Two near ...infrared (NIR) data sets represented by untreated and second derivative spectra were used to predict
Eucalyptus globulus
pulp yield. The optimization process was run with the number of variables (i.e., wavelengths) varied from 10 to 100 to determine the optimum wavelengths and number of latent variables for PLS regression model. A linear function of R-squares for calibration and prediction sets was utilized as the objective function of the optimization problem. The optimum wavelengths selected by genetic algorithm helped to considerably improve the performance of the PLS regression model, not only for the calibration sets but also for the prediction sets. The optimum number of latent variables varied over a wide range, from the maximum allowed (20) to a lower limit of six. Representative wavelengths for each data set were also statistically determined and assigned to corresponding wood components through a band assignment process, which showed strong agreement.
Near-infrared (NIR) spectroscopy is widely used as a nondestructive evaluation (NDE) tool for predicting wood properties. When deploying NIR models, one faces challenges in ensuring representative ...training data, which large datasets can mitigate but often at a significant cost. Machine learning and deep learning NIR models are at an even greater disadvantage because they typically require higher sample sizes for training. In this study, NIR spectra were collected to predict the modulus of elasticity (MOE) of southern pine lumber (training set = 573 samples, testing set = 145 samples). To account for the limited size of the training data, this study employed a generative adversarial network (GAN) to generate synthetic NIR spectra. The training dataset was fed into a GAN to generate 313, 573, and 1000 synthetic spectra. The original and enhanced datasets were used to train artificial neural networks (ANNs), convolutional neural networks (CNNs), and light gradient boosting machines (LGBMs) for MOE prediction. Overall, results showed that data augmentation using GAN improved the coefficient of determination (R
) by up to 7.02% and reduced the error of predictions by up to 4.29%. ANNs and CNNs benefited more from synthetic spectra than LGBMs, which only yielded slight improvement. All models showed optimal performance when 313 synthetic spectra were added to the original training data; further additions did not improve model performance because the quality of the datapoints generated by GAN beyond a certain threshold is poor, and one of the main reasons for this can be the size of the initial training data fed into the GAN. LGBMs showed superior performances than ANNs and CNNs on both the original and enhanced training datasets, which highlights the significance of selecting an appropriate machine learning or deep learning model for NIR spectral-data analysis. The results highlighted the positive impact of GAN on the predictive performance of models utilizing NIR spectroscopy as an NDE technique and monitoring tool for wood mechanical-property evaluation. Further studies should investigate the impact of the initial size of training data, the optimal number of generated synthetic spectra, and machine learning or deep learning models that could benefit more from data augmentation using GANs.
Prediction of pulp yield of
Eucalyptus globulus
wood samples based on partial least squares (PLS) regression can be optimized by utilizing specific near infrared (NIR) wavelengths. A critical feature ...of this approach is the weighting of constraint conditions. Equal weighting balances optimization in terms of calibration and prediction; however, there is a lack of knowledge regarding prediction performance of wood property models when different weight factors are used. In this study, pulp yield models were developed using two
E. globulus
data sets characterized by narrow (5%) and extreme (22.6%) yield ranges and represented by untreated and second derivative NIR spectra. The global optimization solver pySOT was used to optimize the performance of a PLS regression model in terms of wavelengths selected and number of latent variables. A linear function of R-squares for calibration (
R
c
2
) and prediction (
R
p
2
) sets was utilized as the objective function with the aim of maximizing
α
R
c
2
+
1
-
α
R
p
2
for all values of
α
between 0 (maximizing
R
p
2
without concern for
R
c
2
) and 1 (only maximizing
R
c
2
). Values of
α
≤
0.8
provided good predictive performance, whereas
α
≥
0.9
tended to overfit the calibration data indicating that models are robust for values of
α
from 0 to 0.8. Representative wavelengths for each data set were identified and assigned to corresponding wood components through a band assignment process. Strong agreement was observed for
α
≤
0.8
; however, for
α
≥
0.9
,
identified wavelengths generally occurred in regions unrelated to vibrations arising from specific wood components.
Maps developed using Akima’s interpolation method, and representing average data for trees aged 13 and 22 years, were used to compare patterns of within-tree variation for
Pinus taeda
L. (loblolly ...pine) tracheid properties: coarseness (
C
), specific surface (
S
), radial (
R
) and tangential (
T
) diameter and wall thickness (
w
). SilviScan-calibrated near-infrared (NIR) spectroscopy provided data for the analysis with
C
(
R
c
2
= 0.85,
R
p
2
= 0.85),
S
(
R
c
2
= 0.83,
R
p
2
= 0.76), and
w
(
R
c
2
= 0.89,
R
p
2
= 0.93) models having very good calibration / prediction statistics, while those for
T
and
R
diameter were moderate (
R
c
2
= 0.79,
R
p
2
= 0.57) and poor (
R
c
2
= 0.64,
R
p
2
= 0.19), respectively.
C
,
S
, and
w
maps were similar to the density maps for
P. taeda
and indicate the properties increase radially at all heights. The
T
diameter map was similar to maps reported for microfibril angle except that
T
diameter increased radially and with height whereas microfibril angle decreased radially and with height. The map for
R
diameter increased with height and was unlike the other properties examined; caution is recommended regarding any interpretations based on the
R
diameter map owing to the weak statistics observed for the NIR model. Changes observed between the two ages are consistent with the asymptotic progression of properties associated with maturation.
To maximize utilization of our forest resources, detailed knowledge of wood property variation and the impacts this has on end-product performance is required at multiple scales (within and among ...trees, regionally). As many wood properties are difficult and time-consuming to measure our knowledge regarding their variation is often inadequate as is our understanding of their responses to genetic and silvicultural manipulation. The emergence of many non-destructive evaluation (NDE) methodologies offers the potential to greatly enhance our understanding of the forest resource; however, it is critical to recognize that any technique has its limitations and it is important to select the appropriate technique for a given application. In this review, we will discuss the following technologies for assessing wood properties both in the field: acoustics, Pilodyn, Resistograph and Rigidimeter and the lab: computer tomography (CT) scanning, DiscBot, near infrared (NIR) spectroscopy, radial sample acoustics and SilviScan. We will discuss these techniques, explore their utilization, and list applications that best suit each methodology. As an end goal, NDE technologies will help researchers worldwide characterize wood properties, develop accurate models for prediction, and utilize field equipment that can validate the predictions. The continued advancement of NDE technologies will also allow researchers to better understand the impact on wood properties on product performance.
Near-infrared (NIR) spectra or NIR-hyperspectral images obtained from radial strips or wood discs provide a cost-effective methodology for examining wood property variation within trees. The ...calibration used for wood property prediction is critical and can be obtained using two fundamentally different approaches. One involves using a spatial-specific model where wood property data and corresponding spectral data are measured at the same resolution for calibration and prediction, e.g. 10-mm radial increments. The other provides a spatial-interpolated model and involves measuring a property on a broad-scale, e.g. whole-tree, calibrating this data against NIR spectra representing the equivalent scale and then using the calibration to predict the property at higher resolution. To understand the impact of these approaches on subsequent patterns of within-tree variation, whole-tree air-dry density (ADD) and coarseness maps, based on data obtained using the two different approaches, were compared. Patterns of ADD and coarseness variation were comparable indicating that both approaches can be utilized to examine within-tree variation. Spatial-interpolated models have a distinct advantage; being based on whole-tree (or disc) samples, they greatly reduce the cost of wood property analysis and allow the development of maps for properties that are costly and difficult to measure, for example, pulp yield.