1. INTRODUCTION

Economically, the forestry industry would benefit significantly from a nondestructive way to gain precise information about the quality grade of wood. Currently, special attention is paid to the early detection of wood-destroying insects with regard to their occurrence and propagation in trunk wood. Due to climate change and globalization, insect infestation and the spread of new species have increased significantly in recent years. Thus far, however, the insect-infested area of a forest is often only estimated without an exact identification of infested and noninfested trees. In order to prevent a far-reaching infestation, healthy trunk woo, which is not yet intended for harvesting, is often felled without necessity. The tree to be felled is currently determined only by a visual inspection. In higher areas of the tree, where the infestation usually takes place first, this visual assessment is not possible.

The typographer breed in the area directly below the bark in the bast of deadwood and living trees. Typographer females are lured by pheromones of the male to the host tree. There they mate in a so-called Rammelhöhle (nuptial chamber) under the bark of spruce trees. The females then drill a tunnel and place the eggs at the end of it. After hatching, each larva eats itself through the bast, which leads to the species-typical boreholes. Finally, the larvae create a chamber in which they pupate. The finished typographer drill through the bark into the open air. Beetles and larvae live off the so-called bast. Within the bast are the vessels of the tree, which transport water and nutrients. If a tree is severely affected, the flow of water and nutrients between the crown and the roots is disturbed so that the tree eventually dies [1]. Typical characteristics of an infestation are the feeding paths on the inside of the bark, small brown traces of sawdust on the trunk, needle drop and discoloration, woodpecker tees and resin leakage [2].

At a daytime temperature from 18°C to 20°C, the first beetle flight starts. Another swarm out usually takes place in July/August. Usually, there are one to two generations per year, but with long-lasting warm weather, a third generation can develop in September or October. A female typographer has about 60 progeny per generation. Thus, one female can produce in purely mathematical terms more than 100,000 descendants per year. The early detection of an infestation could therefore significantly reduce the spread of insect infestation. Up to the present, there has been no technical method by which the presence or spread of insect tunneling can be qualitatively assessed without removing the bark. Therefore, we have explored the possibilities of using terahertz (THz) radiation as a noninvasive method. THz time-domain spectroscopy (THz TDS), with power to access the elusive band of the electromagnetic spectrum between 100 GHz and 10 THz (30 mm and 30 mm), is a technology that arose about three decades ago and has made substantial progress since then [3,4]. Increasingly, the THz radiation is being used for interdisciplinary studies to determine different forms of material cavities in various fields, including art, medicine, or quality control [5–7].

In recent years, scientists in the THz field as well as researchers in the field of wood analysis have seen the potential of combining their expertise in order to reveal information about density and moisture of wood which was not accessible before [8,9]. Previous studies have also shown the potential of THz radiation in the nondestructive testing of wooden samples [10–13]. Until recently, however, THz studies were limited to flat samples. Due to the recent development of a robotic-based THz system [14], the THz radiation can now also be used for samples with complex 3D surfaces [15]. In this article, we present tomographic THz time-domain measurements acquired with a robotic THz system of spruce wood sections infested with and without typographer. Furthermore, we compare the results with X-ray computed tomography (CT) images of the same spruce samples to evaluate the potential and the limitations of the THz measurements.

2. SAMPLES AND METHODS

The sample material from spruce with infestation by typographer was obtained by kind permission by the forest warden’s office of Lower Saxon State Forests in Altenau, on June 9, 2017. At that time, spruces that had infestations had been freshly felled and stem sections were taken. For comparative studies, material with an unaffected area was obtained from the same tree. The samples are shown in Figs. 1(a) and 1(b), respectively. The dimensions of the prepared samples were about $10\mathrm{cm}\times 7\mathrm{cm}\times 0.5\mathrm{cm}$ for the infested and about $15\mathrm{cm}\times 6\mathrm{cm}\times 0.9\mathrm{cm}$ for the uninfected piece. Bark and sapwood in their entire thickness had a dimension of about 1 cm at the widest point. First, the samples were shrink-wrapped in foil, frozen, and dried at room temperature before the actual measurement took place (May 2, 2018). The THz properties of uninfected spruce samples are investigated in [16]. For the THz measurements, areas with aluminum foil were marked in order to locate the measurement positions more easily. After the measurements were taken, the bark was removed from the sapwood to obtain a visual impression of the degree of destruction by insect infestation [c.f. Figs. 1(c) and 1(d)]. The insect-infested sample shows a significantly more inhomogeneous wood structure due to the large number of insect burrows. This inhomogeneity is also reflected in the determined density of the measured noninfested ($0.401\mathrm{g}/{\mathrm{cm}}^3$) and infested ($0.350\mathrm{g}/{\mathrm{cm}}^3$) wood samples without bark. It is known the wood density is correlated with the THz properties of wood [8]. The density of the unbranched spruce approximates our determined values within the scope shown in [16].

 

Fig. 1. Investigated spruce samples. (a), (b) The measurement area is indicated by the rectangle out of aluminum. (c), (d) Samples after removal of the surface show the different structure below the bark caused by the infestation of the typographer.

Download Full Size | PDF

A. Robotic-based THz System

For the THz measurements, we combined a THz time-domain spectrometer developed at the HHI in Berlin [17] with a robotic arm. By doing this, it is possible to measure samples with a complex 3D shape in reflection geometry. The whole system is described in detail in [14]. The measurement process can be divided into three steps. In the first step, the 3D surface of the sample is acquired by using a fringe projection method. Based on this surface data, the measurement path for the THz sensor is calculated in a second step ensuring that the THz sensor is always perpendicular and at a defined distance to the measurement point. To avoid collisions between the sample and the THz sensor, a simulation of the raster scan measurement is performed before the real THz measurement starts. For the measurement of the spruce samples, we measured an area of about 6 cm by 4 cm with an amount of 364 points for the noninfested sample and 298 points for the infested sample. The lateral resolution of the THz scan is about 2 mm for both samples, resulting in a slightly different size of the measurement area due to the different sample size. For each measurement point, 100 waveforms are averaged to improve the signal-to-noise ratio. The data received in that way are further processed using a sparse deconvolution algorithm [18] to evaluate thin layers up to a thickness of 20 μm. Afterward, the temporal position of the reflected THz pulses is detected in the impulse response function and used for a tomographic reconstruction based on the surface model [17]. To translate the temporal information of the THz data into a spatial one, we used a refractive index of 1.38 of spruce in the THz frequency region. A previous THz study of spruce in transmission configuration has shown that the dispersion of the refractive index of spruce is quite low in the THz region [16]. Thus, the use of an averaged refractive index of about 1.38 for the tomographic reconstruction is possible.

B. X-ray Computed Tomography

This technique allows the 3D reconstruction of many materials, as long as they have enough contrast at X-ray wavelengths. The insect tunneling should be recognizable in black in the CT images, while other materials will appear in darker gray tones depending on their particular composition. Given that X-ray CT is a standard imaging technique in medical and material science, we do not consider that an extensive technical description is required. It should be noted, though, that the equipment used was a Philips multidetector spiral CT type Brilliance 64 with 64 detector rows, each having an average thickness of 0.625 mm.

3. RESULTS AND DISCUSSION

The results of the THz tomography analysis of the spruce samples are summarized in Fig. 2. At the top, the 3D surface model of each sample is shown. This was acquired with the fringe projection method. The area measured with the THz system is highlighted by the colored rectangle. The white dashed line indicates the position of the corresponding cross-section for each sample, as shown at the bottom of Fig. 2. Each line represents the interface of two layers at which the THz pulse was reflected. These lines are reconstructed based on the temporal position of the peaks in the impulse response function. The peaks detected in each measured signal are assigned to an interface according to their temporal order, resulting in the lines shown below. Sometimes, the sparse deconvolution algorithm fails, and one THz pulse is reconstructed by two or more peaks in the impulse response function. This leads to lines that are close to each other. Nevertheless, the totality of all reconstructed lines shows the internal structure of the samples. The cross sections exhibit several interfaces beneath the first layer. In the case of the noninfested spruce sample, these interfaces are much more homogenous than for the infested sample. The inhomogeneous structure can be attributed to the burrows caused by insect infestation. The cross sections exemplify many other cross sections all over the measurement area, which show a similar result. It therefore can be assumed that the sample has been completely destroyed by the insect infestation. The subsequent invasive removal of the bark confirms this assumption [c.f. Fig. 1(d)].

 

Fig. 2. Surface model of the (a) noninfested and (b) infested spruce sample with the measurement area highlighted by the colored rectangle. White dashed line indicates the position of the corresponding cross section through the THz data presented in (c) and (d), respectively.

Download Full Size | PDF

 

Fig. 3. Comparison between the THz cross sections and the CT images.

Download Full Size | PDF

To verify the THz measurement results in a noninvasive way, we compared them with CT images of the same samples. The comparison shows good agreement between the course of the THz and CT data, as shown in Fig. 3. However, the depth resolution and the contrast of the CT images are limited. It seems that also in the CT images the structure of the infested sample is more inhomogeneous than that of the noninfested sample. Overall, the differences are much more clearly presented in the THz images.

4. CONCLUSION

We have shown that THz tomography with a robotic THz system is able to detect damages on wood caused by insects like the typographer, whose burrows are close to the surface of the wood. It is not possible to evaluate the exact geometry of the insect tunneling. However, based on the THz data, it is possible to distinguish between infested and noninfested wood if the burrows are not deeper than 1 cm. Burrows of insects, which are deeper in the wood, cannot be detected by THz tomography because the penetration depth of THz radiation is limited to 1 cm in most wood species, assuming a signal-to-noise ratio of the THz system of 60 dB. The depth resolution of the THz measurements exceeds that of CT images. Furthermore, the usage of a THz time-domain spectrometer does not require any industrial safety measure like for X-ray-based techniques. We have proved that it is therefore suitable for mobile use to identify infested areas on spruce caused by typographer. Thus, the THz technique presented here could help to detect insect infestation on trunk wood, imported woods, or on wood products in an early infestation stage. Especially with regard to the spread of new wood-destroying insects, this technique can profitably be used for checking.