As mentioned in the previous section, the availability of a reliable tracking service discloses several new opportunities in timber supply chains. Among the most promising applications can be found the synergy between tracking technology and timber grading systems based on fast and nondestructive sensors. At present, these sensors find little application in forest operations: timber assortments can be sorted with machine support (e.g., using the StanForD data and dedicated software in the harvest machine), but quality grading of the logs is performed only visually. Nevertheless, this is a costly and time-demanding operation, which is potentially biased by the subjective judgment of the expert evaluator.
Once in the industry, a large set of sensors is used to precisely quantify the volume of the logs delivered and its quality. Additionally, an increasing number of sensors or analytical techniques are used to optimize the following steps of product transformation in the sawmill or pulp mill. According to Gergel et al. (2019), the sawmilling optimization made possible by a detailed knowledge of the properties and defects of single logs leads to an increased value recovery of 15% and 23%, respectively, for conifer and broadleaf timber. The same authors also highlighted how the status and quality of the standing trees prior to harvest plays a role of the utmost importance in the efficiency and economic balance of the timber supply chain. Nevertheless, in real practice, just empirical sorting models are used for inventorying trees according to their economic value. These return the share of predefined timber assortments with an intrinsic inaccuracy due to the impossibility of detecting internal wood defects from external features. A wide range of sensors could contribute to provide this service early in the supply chain. Their potential is maximum when different sensors work in synergy, contributing to generate a wide and comprehensive picture of the timber quality, for instance, integrating stem shape characteristics as returned by a TLS system (e.g., taper, straightness, crown insertion, etc.) with internal properties as detected by optical sensors on the felled trees (e.g., rot, decay, eccentricity, etc.). In the short term, the fusion of such data provides a more detailed definition of quality of single logs. In the long term, optically sensed characteristics may be used to tune the models interpreting 3D data provided by laser technology, gradually but constantly decreasing the gap between estimated properties and actual characteristics of the assortments. The very same data and approach can also disclose the “memory” of timber, being this the result of the growing conditions that the tree endured along its life. Widening the scope of the data analysis and the inputs of the models, it will be possible to support CSF studies and forecasts by relating the captured wood characteristics with the physiology of the trees and their response to the past stresses.
A recent trend in timber characterization is to deploy sensors as early as possible in the timber supply chain. Although application of timber quality gauges in forest machinery is mostly at the experimental stage, it arises great interest from the industry, since quality determination early in the supply chain would allow a great reduction of procurement costs by delivering just the desired quality to each end user, optimizing the logistics and further increasing the profitability of the industrial transformation. At the same time, the application of sensors for timber quality along the timber production process would bring additional benefits also to CSF applications, providing more data and with higher level of detail.
In addition to UAVs, forest technologies for monitoring tree processes include sensor
networks deployed on the tree stems or embedded in the soil layers. The proliferation of these technologies generates a flow of data, which needs to be appropriately investigated through machine learning for automating or responding to disturbance events. The digitalization of forest stands allows forests to operate as technological platforms so that trees function as technical instruments informed by data that are meant to enable precision forestry and practices oriented toward high timber quality.
Wood Properties Relevant for CSF
The industrial conversion of timber highly relies on the intrinsic characteristics and technical properties of this resource, which may be advantageous (or not) depending on the downstream process applied. The set of properties defining the quality of wood must be determined against the industrial requirements, local policy, and environmental constraints to assess its market value.
In a theoretical approach, a healthy tree results in production of the “perfect wood” defined here as a bioresource without wood defects. Perfect wood corresponds to the cylinder with a pith positioned in its geometrical center and containing regular concentric structures corresponding to yearly rings. In practice, perfect wood does not exist and always includes some defects. Here, the “defect” is considered as an undesired imperfection of the regular wood tissues that is a result of diverse stresses and factors affecting tree growth and the morphological constitution of the plant. Such natural features may be highly undesirable, downgrading the industrial value of timber. An example could be a knot that is an imprint of the branch positioned in the trunk. Branches cause the presence of knots that deviate the fiber direction, having a tremendous effect on the mechanical properties of timber, as well as tree stability.
In temperate forests, the life of trees covers a sequence of several seasons that are recorded in the tree rings. These are structural wood tissues of different properties in spring/summer (early wood) or autumn/winter (late wood). The ring width, its chemical composition, microfibril structure, as well as the ratio of late to early woods may vary, depending on the age of tree, meteorological history, or presence of diverse factors stimulating or inhibiting the tree growth. The natural yearly sequence of air temperature, solar light photoperiod, as well as water stress
levels are expressed as the dynamical changes of the phenological events determining specific xylogenetic sequences. Therefore, the tree-ring structure may be considered as a “fingerprint” unique for each plant. This can be used for dating of wood (dendrochronology), determination of the wood origin (dendro-provenance), analysis of the local climate changes (dendro-climatology), or identification of catastrophic events, occurring during the life span of the tree, among others. As such, tree-ring analysis can be considered as one of the most important inputs for the tree growth and health models developed in CSF (see Chap. 7 of this book: Bosela et al. 2021).
Traditionally, tree-ring analysis was performed by visual (microscopic) assessment of tree core samples extracted from the living tree or on cross sections of the log after felling. For that reason, the amount of available information was relatively limited. However, several modern technologies, especially based on the tomography approach (Van den Bulcke et al. 2014), allow mapping of the tree-ring structure without necessity for its cutting or extracting samples. Mobile X-ray computed tomography technology was applied in wood density measurements and moisture content monitoring on standing trees (Raschi et al. 1995; Tognetti et al. 1996). Computer 3D tomography allows locating internal log features that include pith, sapwood, heartwood, knots, and other defects. With the appropriate techniques, it can also return a detailed analysis and densitometry of annual growth rings (Van den Bulcke et al. 2014). Additionally, spectroscopic methods as well as its evolution by means of hyperspectral imaging (Sandak et al. 2020; Schraml et al. 2020) provide a possibility for extraction of till-now not accessible information regarding chemical and physical properties of wood resolved spatially to the tree-ring level.
Nevertheless, from a technical perspective, it is crucial that both the quality of the acquired images and the capacity of the interpretation software provide sufficient information regarding tree rings of the logs (Subah et al. 2017; Cruz-García et al. 2019). The sensors installed on the processor head and operating in the forest on unprepared surfaces (the chain saw cutting surface is relatively rough for this use) may be incapable of returning the required quality for a deep and reliable analysis. On the other hand, they would potentially provide a large dataset, based on all the logs produced in the forest, including those with defects. This contraposes to the higher potential of industrial timber analysis, where more powerful sensors, deployed in a controlled environment, can provide extremely detailed data. Once set up, an automated tree-ring analysis, linked with an effective traceability system, would provide a very large volume of data. Such information may prove extremely valuable for understanding the past dynamics of relatively large areas of forests or to refine and better elaborate the information provided by dendrometers installed in monitoring plots (Cruz-García et al. 2019). For instance, the availability of a large dataset reporting growth pattern of trees of the same species growing in different areas allows to better understand the seasonal and site-specific growth response to drought (Mina et al. 2016) or human-driven factors, such as pollution (Innes and Cook 1989) or wildfires (Walker et al. 2017). This information can be further elaborated, helping to understand the response to stress
of forests with different characteristics, such as density (Sun et al. 2020) and elaborate guidelines for CSF implementation. Although measurements of stem growth characteristics in temporal detail (e.g., radial increment, slow vs. fast growing trees) are important to understand wood properties, these properties change during wood formation in response to changing environmental conditions. Indeed, vessel conduit dimensions, cell wall thickness, and the relative proportions of different xylem cell types vary during the growing season. Changes in stem size detected through dendrometers can be associated with tree water status (Zweifel et al. 2007), with daily fluctuations being related to physiological parameters (e.g., leaf water potential, whole plant transpiration) and environmental conditions (e.g., evaporative demand and air temperature) (Giovannelli et al. 2007; Tognetti et al. 2009). These stem radius changes provide a sensitive indicator of the combined effects of actual radial growth and stem water storage and release (Drew and Downes 2009). Stem size variation provides indications of water stress
thresholds in tree species and is potentially useful in threshold analysis for binary classification and determining the influence of thermal or moisture cycles on elastic shrinkage (Cocozza et al. 2009, 2012). Coding of the dendrometer
signal helps quantify stem (or log) sensitivity to environmental fluctuations (moisture, temperature), as well as synchronize time series related to wood properties and climatic events, and to identify time lags of environmental effects on wood traits (Cocozza et al. 2018).
Ideally, the study of tree reactions to stresses should be based on the analysis of tree-ring development of trunk sections located at different heights. In fact, while the common dendrochronology focuses on basal sections, where all the rings are represented (since the early development of the tree), in physiology studies, the rings grown higher in the tree’s crown may better describe the stresses suffered by the plant. For instance, in the case of Picea abies growing in the Alps, the phloem’s growth at the base of the trunk is completed already in July, while at higher levels of the tree, it keeps growing for the whole summer. Thus, the rings developed higher on the stem provide more effective evidence of the growing conditions of the mid-late growing season. If the tracking system records the order of production of each log, their original position in the trunk can be located and the sawmill’s data (e.g., tomography) can be related to sections corresponding to different heights of the tree.
Density of wood (or specific gravity) is a metaparameter determining several technical characteristics of natural resources, considered as the most relevant wood quality descriptor (Zobel and Jett 1995). It expresses the amount of wood substance contained in a given volume. Even if density is not directly affecting wood properties (it is a property quantifier), it is highly correlated to the majority of wood assets (chemical, physical and mechanical), the yield of production, and the overall “wood quality” in general. The density distribution differs within a single plant, both along the tree height and the trunk diameter. Despite being a property of remarkably high native variance and heritability, it can be related at stand level to growth conditions or silvicultural interventions (Briggs et al. 2008), thus providing valuable information if appropriate reference values can be defined. Density is often measured in sawmill, but the interest to discriminate the timber products with the desired properties as early as possible within the supply chain stimulated the development of portable instruments. These can be deployed both for assessing the characteristics of standing trees (Paradis et al. 2013) and for automatically measuring each log with gauges installed directly in the processor head (Walsh et al. 2014a).
The effect of tree-ring width on tree-ring wood density depends on the species (conifers vs. broadleaves, fast- vs. slow-growing species), the timing of climatic events influencing growth throughout the growing season and the general fertility of the site. In particular, intra- and interspecific interactions may affect the radial growth and wood density of individual species growing in mixture when compared to its monoculture (Zeller et al. 2017). The acoustic sensing technology, such as the modulus of elasticity of wood (MOE) and the dynamic modulus of elasticity (MOEd), allows the estimation of intrinsic wood properties for standing trees, stems, and logs. These parameters depend on wood density and are fundamental for the evaluation of wood quality, providing information related to wood anatomy and tree physiology (Russo et al. 2020).
Chemical Composition of Wood
From the chemical viewpoint, wood is a natural composite of three biopolymers, including cellulose, lignin and hemicellulose. These are major constitutive chemical components, with their specific ratio varying between wood species, forest types, and within individual trees. In temperate areas, the chemical composition differs substantially along the tree height and its radius following the natural lifetime sequence of the periods when the tree grows fast (spring) or forms more mechanically resistant morphological structures (autumn). The variation of chemical composition can also be noticed at the level of the yearly ring that reflects the combined effects of the season and plant development stage as well as any other stresses for the tree due to biotic or abiotic factors. In addition to cellulose, lignin, and hemicellulose, small amounts of minerals and extractive components are present in natural wood. The latter are particularly relevant despite their low concentrations. In fact, extractives may significantly affect the suitability of wood resources for a given conversion process or affect the durability of wood-derived products.
All the chemical components form larger macromolecules, such as microfibrils, that are combined at different scales as fibrils, cells, and yearly rings constituting a hierarchical structure. The specific physical properties of wood are, therefore, highly dependent on the scale of observation (nano, micro, macro). It implies the necessity for adjusting measurement procedures and instrumentation for determination of desired chemical/physical characteristics and material properties. Such information has not yet been used for CSF applications. This is probably due to the cost and time delay of wet chemistry analysis. The availability of fast and nondestructive sensors, such as hyperspectral cameras, may disclose a new source of information to understand the health and growing conditions of trees.
In contrast to the perfect wood, the real trunk contains diverse imperfections recording all the lifelong-related growth conditions, perturbations, or stresses. In the timber industry, these are defined as wood defects and may include numerous features differentiating defected from the perfect wood (Kimbar 2011). The European standard EN 1927-1:2008 provides a systematic methodology for identification and quantification of the log/wood defects that are later used for determination of the quality class. The following sections report the most relevant defects from the perspective of CSF.
Resin pockets are small gaps within the structure of the xylem filled with resin. Wood development due to tree growth usually occludes them within a few years after their formation. Resin pockets are a significant technological defect for the timber industry, particularly for joinery and furniture applications, due to the release of resin over time from the finished products. Resin pockets are common in conifers with resin canals (such as Picea, Pinus, Larix, and Pseudotsuga genera) but may also be the consequence of stress
. In the latter case, they are commonly related to animal or insect attacks, sites exposed to strong winds and storm damages. Research demonstrated that water stress
is to be regarded as the most relevant factor leading to the development of this wood defect (Seifert et al. 2010; Jones et al. 2013), although other factors, such as excessive growth rate or share of defect core over total diameters at breast height (DBH), may contribute to pocket formation (Woollons et al. 2008). Availability of data on resin pockets may be a useful tool for identifying historical occurrence of water stress
, areas with frequent wind gusts, and, possibly, the impact of forest management on certain growing conditions. Clearly, the interpretation of resin pocket presence is much more significant if associated with other parameters detected by the sensors on processed and sawn timber, namely, tree-ring development. Furthermore, tree rings are present throughout the whole trunk, but their size increases from the base upward and from the core outward (Gjerdrum and Bernabei 2007). Thus, for a correct interpretation of their occurrence, it is important to know the position in the stem of the timber sample considered. This is possible only with an accurate traceability system capable of relating each log sourced from a tree with each other according to their sequential order.
Like resin pockets, resinous wood is a zone within trunk volume with exceptionally high content of extractive components and resins. It is present only in conifer species with resin channels. The usual causes for resinous wood formation are responses to the microorganism activity (especially parasite fungus) or to the damage of wood induced by mechanical actions. The analogy for self-protection by the resin release in some broadleaves is a gum production that can be triggered by the frost, wound, or microorganism attack.
Checks, splits, and shakes (commonly defined as cracks) are separations or ruptures of the wood tissue in the longitudinal plane that normally occur along fibers in the radial or tangential sections of the trunk. These may appear on the cross section of the tree or on the log side circumference. Checks are separations of the fibers that do not extend through the timber from one face to another. Splits, however, extend the material discontinuity from the one log face to another. Shakes are separations or weaknesses of fiber bond, between or through the annual rings. There are diverse sources of stresses occurring to wood during life cycle. The most relevant are growth and drying stresses, beside thermal, frost, wind-, solar-, or lightning-induced tensions. Shakes may originate from causes other than drying stresses, e.g., from careless felling, where internal stresses existing in the living tree are released when the tree is felled. Checks, shakes, and splits are present although they may not be visible (closed checks and closed splits). Cracks may close up if the dry timber is subsequently exposed to damp conditions, but once the fibers have separated, they cannot join together again. A great threat for the tree after a crack occurring is elevated risk of decay fungi spore access to the unprotected wood tissue. The presence of cracks is an important limitation for the downstream conversion, especially if combined with other defects, such as spiral grain. Due to specific properties, cracks are relatively easily detectable even if not visible on the surface of log. This is due to the discontinuity of the material and related change of the material stiffness (natural frequency) or elasticity (stress
wave propagation velocity).
Reaction wood is a type of defect that tends to form in trees growing in a leaning posture. This may be caused by exposure to strong winds or because the tree grows on a slope. Reaction wood in coniferous trees is formed on the lower side of the lean and is called “compression wood.” It is often characterized by a dense hard brittle grain that contains very high content of lignin that increases wood resistance for compression. On the contrary, the broadleaves create reaction wood referred to as “tension wood” that is positioned on the upper side of the lean. It contains higher content of cellulose that increases wood resistance to tension stresses. Properties of compression wood are considerably different from those of normal mature wood. Compression wood tracheids, for example, are about 30% shorter than normal. In addition, compression wood contains about 10% less cellulose and 8–9% more lignin and hemicelluloses than in normal wood. These factors reduce the desirability of compression wood for pulp and paper manufacture. It is also less suitable as sawn timber since it shows a lower strength, stiffness, and dimensional stability, resulting in a decrease in yield of high-quality end products.
Data related to reaction wood can be a particularly useful tool in the frame of CSF in mountain
areas, as it records specific reaction of trees to environmental conditions. According to Łszczyńska et al. (2019), data regarding the frequency and characteristics of reaction wood in forested plots may be used to assess the landslide hazard risk. In fact, in case of slight land movements, trees are tilted causing leaning and thus the formation of reaction wood. The same phenomena also lead to the formation of eccentric piths, which can be detected by tree-ring analysis. The vertical stability of a tree can be assessed using an automatic accelerometer (gyroscopic sensor
), which measures the position and oscillation in three axes (Matasov et al. 2020), therefore, providing useful information on the effect of wind exposure on the tree aerial architecture and species-specific biomechanics.
Fungi decay is considered as the most problematic biological threat degrading wood at all stages of its life cycle, including postharvesting. Due to its chemical composition, wood is an optimal source of nutrition for fungi and, therefore, it will be a subject of extensive degradation whenever favorable conditions for growth of fungi occur. These include temperature ranges from 20 to 30 °C with a wood moisture content ranging from 20 to 50%. Diverse species of decaying fungi are specialized in degradation of specific wood polymers resulting in different degradation results. This led to the classification of fungi into three major groups: white rot, brown rot, and grey rot. The fungi spores may access wood by several ways, including root, broken branch, damaged leader, or scar on the stem. Cracks, wounds, or any other exposed surfaces of the wood. The presence of birds nesting in hollows in the tree is a certain sign of the progressing decay deterioration. Rot is the wood defect for which the quality grade reduction is obligatory. Logs cut out from older trees are more likely to contain developed rot. The final stage of rot is a complete or extensive material loss forming internal cavities. Plants have developed several mechanisms to defend themselves against decaying fungi. For instance, diverse chemical substances synthetized by trees are natural biocides, such as tannins, resins, or gum.
A knot in the tree is the portion of a branch or limb that has been surrounded by subsequent xylem growth during the tree life. Knots form morphological structures starting from the pith and by following the radial direction reaching the log surface. There are more than 50 types of knots that are classified according to the size, decay presence, and location and distribution within the stem.
The size, type, and distribution of knots have the most important impact on log quality and are the main consideration when applying grading rules. The severity of the grain deflection caused by the knot is correlated with its size. In any case, the presence of knots changes the anatomical structure of surrounding wood (reaction wood presence and grain deviation), as well as its chemical composition (extremely high content of extractive components). The distribution of knots in logs depends on the species and characteristics of the growing site. It is also determined by the growth characteristics of the tree and the tree age. The detailed knowledge regarding knots in trees is highly relevant for CSF. Fortunately, there are several scanning techniques for automatic detection and classification of knots that are implemented during forest operations, log sorting/grading, as well as downstream conversion.
Any deviation of the tree form from the perfect cylinder is considered as a shape imperfection, reducing the yield of product that can be obtained in the sawmill. Sweep, excessive taper, bulges, swell, flanges, and out-of-roundness are some of the most important shape imperfections. All these defects can be measured both on the standing tree and on the processed log by means of laser triangulation scanners, photogrammetry, and TLS. In some cases, imperfections can be reported also by the standard measurement equipment installed on most modern timber processors.
Sweep is a bow-like bend in the trunk of a tree diverging the trunk from the straight and vertical theoretical axis of the tree. The presence of sweep is a result of diverse factors, including slope of the terrain, temporary loading due to snow or wind, mechanical damages, or insect activity.
Tree taper is defined as the gradual reduction of the log diameter along its length. It is a natural feature of each living tree, even if logs with a high degree of taper are considered as having a poor form. Although taper cannot be eliminated, it is possible to minimize it by means of appropriate silvicultural activities. In fact, the extent of taper depends not only on the tree species, local climate, age of the tree, soil fertility, and terrain irregularity but also on the density of trees within the surrounding forest stand.
Bulge is an enlargement of the tree diameter forming a barreling shape. It is a natural feature when occurring at the bottom of the tree, assuming reasonable progress of the diameter changes. On the contrary, when occurring in the higher part of the trunk it is frequently associated with fungi or bacteria attack.
Canker is a defected wood in a form of gnarls or volume losses, both attributed to the phytopathological changes triggered by an activity of microorganisms (fungi or bacteria). In contrast to swollen wood, the tissue of the canker is abnormal and sick.
Out-of-roundness is a shape imperfection appearing as an elliptical cross section of the log. It is frequently associated with the eccentric or double pith. The usual consequence of the pith shift is the presence of reaction (tension or compression) wood. Another reason for the out-of-roundness is the partial damage of the cambium due to mechanical or phytopathological injury.
Such stem defects can be related to the influence of various environmental factors, supporting CSF studies to better understand tree growth dynamics. According to Schneider (2018), the hydraulic and biomechanic theories are the most widely used. Both theories underline the importance of crown dimensions in determining tree form, as confirmed by Kidombo and Dean (2018). Yet, climatic variables such as total summer precipitation and mean winter temperature may have a higher influence on tree taper than the average wind speed (Schneider 2018). This underlines that the complex tree growth dynamics require a holistic analysis of all climatic factors and physiological processes to understand stem formation.
Sensors for Timber Quality Assessment
Once the timber reaches the mill, a large array of possible sensors can be deployed to analyze the quality and volume of the logs to be processed. There are several technical solutions available for wood defects detection or quality grading of logs. Some of the most compatible with CSF requirements are presented in Fig. 9.10.
The length of log combined with its diameter is a basic merchantable property. As mentioned in Sect. 9.4.2, all the processors used for trees harvesting and delimbing are equipped with a measurement wheel and optical encoder (Mederski et al. 2018) (Fig. 9.10a). Sensors used for log diameter measurements are usually absolute encoders configured as protractors, which are integrated with both delimbing knives and feed rollers and measure these rotation angles (Fig. 9.10b). An advantage of such a measurement system is a possibility for a continuous determination of the diameter change trends along the three when passing the processor head. Combined length and diameter measurements provide not only highly precise information regarding a single log volume but can also quantify taper. Such a measurement approach is capable of fast data acquisition and straightforward integration with a database when converted to StanForD file format.
The light curtain is a simple optical measurement system, where the dimensional information regarding the object is determined by illumination (or shadowing) of electronic photodetectors (Fig. 9.10c). It was a very popular solution for the size sorting of logs supplied to the sawmill. However, nowadays it is replaced by the 3D laser triangulation systems relying on the image analysis of the structured light profiles (usually laser) that appear as deformed when observed from an angle (Siekański et al. 2019). The advanced analysis of the surface texture allows identification of some wood defects appearing as particular textures of the bark.
Implementation of X-ray scanners for monitoring of logs is the industrial solution for detection of the majority of wood defects not visible on the log surface (Fig. 9.10d). It allows for identification of deviations of the wood properties along the log, without the possibility to localize the depth position of each feature. For that reason, this setup is frequently duplicated to provide a possibility for better recognition of the defect location within the log section. The ultimate solution for the X-ray imaging of logs is X-ray computed tomography (CT) scanner (Fig. 9.10k). This technology allows straightforward detection and mapping of internal defects, such as butt decay, voids, cracks, or inclusions. The resolution of images allows dendrometric analysis of tree rings and other refined measurements (Van den Bulcke et al. 2014; Stängle et al. 2015; Rais et al. 2017; Gergel et al. 2019). Appropriate data analysis may allow for the information collected on the logs to be related back and integrated with the information on the original tree. An example of this application is provided by Stängle et al. (2014), which compared data from TLS stem and branch scar analysis with X-ray computed tomography (CT), and Uner et al. (2009) using X-ray CT to highlight the effect of thinning on timber density. Thanks to the presence of several X-ray CT operatives in sawmills around the globe, and to their high-speed analysis (up to 180 m/min), this technology is a highly promising data source for CSF applications.
As in the case of drilling resistance, it is possible to indirectly assess mechanical strength of wood by measuring cutting forces occurring when crosscutting logs with a chain saw (Fig. 9.10f). This can be implemented by integrating load cells with the cutting unit or by measuring other effects, such as electrical power consumption or oil pressure changes in the hydraulic circuit of the saw motor (Sandak et al. 2019). Another possibility for adopting cutting resistance analysis for log characterization is to measure the forces required for delimbing. It is clear that a healthy and big branch results in much higher cutting resistance than delimbing of the small and dry branch. In any case, value obtained for a chain saw or delimbing knife can be considered only as an estimate and indirect quantification of the wood suitability; however, it is useful to identify some critical wood defects, such as decay or butt rot.
wave propagation velocity is a highly useful tool for identification of the stiffness and modulus of elasticity (Walsh et al. 2014b) of logs derived from harvested trees (Fig. 9.10g). In that case, the sensor
can be installed directly on the processor head and perform analysis before a crosscutting operation. A similar approach is used for determination of the mechanical properties of logs by measuring these natural frequencies. In that case, the log is induced to vibrate by an impact, and the vibrational or acoustic response is measured with specific detectors (Fig. 9.10h). These systems can be implemented both, as a part of the processor head configuration or component of the log sorting line in the sawmill.
The advanced algorithms used in image analysis enable mimicking of human vision and thus more effective automatic detection of wood imperfections or presence of defects. Other uses of the images is to implement a fingerprinting approach for log traceability and authentications (Schraml et al. 2020). The scanning with cameras can be performed on the log crosscut end (Raatevaara et al. 2020) (Fig. 9.10i) or on the side of the log (Shenga et al. 2015) along its circumference (Fig. 9.10j). The spectral information collected may include monochromatic, RGB color, multispectral, or hyperspectral images.
Emerging Scanning Technologies
The scanning technologies presented in the previous section represent the industrial state-of-the-art solutions and are commercially available on the market. However, there are several promising techniques with great potential for migration from the laboratory testing into practical applications both in forest and in timber transformation industries. Some examples of the most suited for the CSF are briefly described below.
Spectroscopy, including visible, near-infrared and mid-infrared ranges is an analytical technique quantifying interaction of the electromagnetic radiation (light) with the matter. The light can be scattered, absorbed, transmitted, reflected, or trans-flected from the measured surface. The specific pattern of that interaction is recorded as a spectrum. Light absorbance, especially in the infrared range, is highly related to the chemical composition (functional groups with dipole momentum) of the sample. Therefore, by assessing light spectra reflected from the wood sample, it is possible to determine its chemical composition and some physical properties. An extensive reference dataset and advanced multivariate data analysis are indispensable to assure the high reliability of prediction models (Sandak et al. 2016b). However, if properly calibrated, spectroscopic evaluation of wood can provide a rapid and very low-cost assessment of a broad range of the properties, including species, provenance, chemical composition, physical properties, or suitability for diverse conversion paths (Sandak et al. 2017). Portable NIR spectrometers even allow scanning of standing trees or fallen logs directly in forest (Sandak et al. 2016a).
An evolution of the spectroscopy toward a space resolved map of spectra is implemented as multispectral or hyperspectral imaging. The difference between both is in the number of spectral bands constituting the spectrum, that is, <10 in the case of multispectral imaging. This technology allows rapid mapping of the chemical properties of the material over the object surface (Thumm et al. 2010). There are several properties of wood that can be assessed with the help of this technique, such as moisture content, chemical anisotropy of the constitutive polymers at diverse heights of trees (Meder and Meglen 2012), and fiber angle direction (Ma et al. 2017), among others. The limitation of hyperspectral imaging is relatively high investment cost, fragility of the optical instruments when integrating with forest operations, and very high amount of generated data that requires refined IT systems and algorithms.
Only a few wood scanning techniques allow scanning of the bulk interior. In contrast to harmful ionizing radiation of X- and gamma rays, microwaves are considered as safe and easy to apply in the scanning systems. A great advantage is that, assuming sufficient power of emitter, microwaves penetrate wood bulk, and these interactions with the matter can be interpreted as attenuation, phase shift, or polarity change. These wave properties are directly correlated to the wood density, wood moisture content, and grain angle direction (Schajer and Orhan 2005). For that reason, it is possible to simultaneously measure all the above wood properties. Microwave scanners can be implemented as an array that, in consequence, allows spatially resolved maps of wood assets (Table 9.3).
Intelligent Forest Machines, the Way Ahead
As described in the previous sections of this chapter, the availability of digital forest data makes it possible to optimize forest works from the operational cost and environmental impact point of view. Traceability tools further increase the precision of the operations and guarantee a full control of the woody products, whose value is effectively assessed and maximized by an array of sensors deployed in the mills. From the management (and CSF) point of view, the most effective solution is to link the databases of the forest inventories and the derived tracking systems with the log scanners operating at the sawmill. Nevertheless, a further significant improvement could be achieved with the determination of timber characteristics early in the supply chain, enhancing the sorting of logs and increasing the overall value of the derived timber products (Taube et al. 2020). The fast development of sensors makes this challenge possible, as it had been demonstrated by the EU
project SLOPE (https://cordis.europa.eu/project/id/604129), funded under the seventh Framework Program. Within the project, the potential of the Virtual Forest, the tracking systems, and the timber sensors had been combined in a technological showcase, proving the feasibility of this concept.
The virtual forest was generated integrating information from satellite, UAVs, and TLS surveying systems, combining macro and local analysis in the characterization of the forest resources in mountain
areas. Data was stored in a dedicated database provided with a 3D interface, which allowed users to navigate into the virtual forest, estimate the volume, and value of timber in a selected area as well as to plan accordingly the harvest operations by cable yarding (Fig. 9.11).
The cloud point generated by TLS surveys was used to characterize the trunks in high detail. A dedicated software matched the shape characteristics of the trees with the timber assortments and locally accepted their value, returning the maximum value recovery conditions. Those were appended in the database as bucking instructions to be transferred from the forest to the processor head. For this purpose, standing trees were georeferenced and marked with RFID UHF tags whose ID was linked to the data and bucking instructions of each tree (Fig. 9.12).
All forest operations were performed by prototypes of intelligent machines, namely, a cable yarder and a processor head operating at landing. The former detected the RFID and weight of the load, while the latter by reading the ID of the tree could acquire the crosscutting positions for optimal bucking. Additionally, the processor head installed several sensors for timber quality assessment as described in Sandak et al. (2019).
In detail, the processor deployed the following systems (Fig. 9.13):
Load cells and hydraulic pressure sensors (1) for estimate of a branch index and the approximate position of knots on the trunk
Stress wave and free vibration measurement systems (2) for timber density assessment
Near-infrared (NIR) and hyperspectral imaging systems (3) for characterization of crosscut section and detection of defects (e.g., rot, resin pockets, etc.)
RFID-UHF reader (4) for acquisition of tree ID and retrieval in the VF database of the cutting instructions elaborated for maximum value recovery
Additionally, the processor could mark each new log with an RFID tag, providing a complete tracking system capable of linking the original standing tree (through the first tag attached on it) to each log delivered to the sawmill.
The whole system proved effective in assessing the quality of timber at roadside. The optical sensors (not installed in the prototype) could identify decay spots and would warn the operator to recalculate the cutting instructions, which were based on the shape of the standing tree. From the point of view of a CSF application of this data, the system provides a further source of information which, when fully integrated with other sensor
data, adds to the tool kit available for CSF planning and operations. Currently, only the logs with sufficient market value reach the sawmill and can be further analyzed with industrial sensors. However, the presence of sensors on the processor head would provide data of lower quality (due to the work conditions), but as this is data from all the trees and logs produced in the stand, it is thus more representative of the general health of the trees. Once a track and trace system are fully integrated within the timber supply chain, it is possible to relate the information regarding each standing tree (and the logs produced from it) with the analysis performed in the sawmill. If the same trees were included in a network of sensors, the historical data collected by microcontrollers up to the tree felling could be related and integrated with all the information provided by all the sensors installed on the forest machines or the sawmilling facilities (Fig. 9.14).
The availability of an infrastructure for sensing, wireless transfer, and cloud-elaboration of data developed for forest operations (and maintained by its revenues) is a clear opportunity for CSF. In fact, the network of sensors deployed for forest monitoring and management purposes could rely on this infrastructure for data transfer, storage, and elaboration. Furthermore, by accessing the databases of the virtual forest, it will be possible to integrate, validate, and broaden the data provided by the climate-smart sensors. The data used for operations planning and generated by machine sensors can be integrated with a unique, flexible system serving different long-, medium-, and short-term purposes, such as in situ forest monitoring (see Chap. 10 of this book: Tognetti et al. 2021), large-scale surveying of CSF indicators (see Chap. 11 of this book: Torresan et al. 2021), establishing a CSF network (see Chap. 5 of this book: Pretzsch et al. 2021), or implementing climate-smart silvicultural operations (see Chap. 8 of this book: Pach et al. 2021).
In turn, the data gathered for forest monitoring can be used for planning close-to-nature forestry operations. This improves the capacity to meet multiple goals, such as economic value, biodiversity, and forest resilience. Martín-Fernández and García-Abril Martín (2005) proved that if extensive and high-quality data is available, such planning can be performed at the level of individual trees by means of appropriate algorithms (iterative conditional mode), which otherwise could be applied only on small, intensively managed forest properties.