This paper presents the results of an experiment measuring the tree root system architecture using ground-penetrating radar. Despite the importance of this issue, the knowledge of the root system of trees remains incomprehensive due to various reasons including the limitations of research methods. The basic approach to the exploration of the root system of plants is the root zone excavation method. However, this method inflicts substantial damage to plants and can lead to their destruction. In some cases, excavation is impossible or prohibited, i.g. in cities or protected natural areas. Studying the tree root systems is an urgent issue at the moment and new methods are systematically used to tackle it. The non-invasive approach to study the root architecture includes using the methods of geophysics, ground-penetrating radar (GPR) in particular. The root system of plants tends to adapt to the environment, which includes interaction with the accommodating soil. One of the natural factors influencing the growth is the level of the impenetrable soil horizon, which changes both the structure of the root system and the aerial part of a plant. The aim of this paper is to study the root architecture of mature trees using GPR, rather than root excavation method, to determine its configuration depending on the bedding of bedrocks and to justify how changes in the bedrock depth affects the development of the aerial parts of trees. The principle of GPR is to examine the propagation of an electromagnetic wave in a medium by emitting a probing signal and receiving a reflected signal. Underground objects such as roots are distinguished by diffracted waves (See Fig. 1). The summary of literature data shows that roots up to 0.5 cm in diameter can be found by a high frequency antenna (See Table 1). We carried out a field experiment on an apple tree (Malus domestica Borkh) which grows on the territory of the Botanic Garden of Petrozavodsk State University (See Fig. 2). This territory is characterized by a thin soil cover and underlying moraine soils. Below are metatuff bedrocks. We studied two separate samples. The choice was made due to external differences (See Fig. 3). The trees growing on Plot 2 (Sample 2 of the experiment) have a conspicuous compound curvature of the trunk and a lower height than plants from Plot 1 - sample 1 (See Table 2). The two trees underwent annual average phenological stages simultaneously. They are 29 years old. No significant differences in the annual number of ripening fruits was revealed. Thus, the experiment was carried out on the two samples growing in equal conditions but having differences in appearance and biometrics. Literature data show that an apple tree usually has 4 perpendicular skeletal roots and 4 diagonal skeletal roots 1-2 cm in diameter in horizontal slice. As for vertical slice, tree roots usually lie at a depth of 20-60 cm (See Fig. 4). The measurements were carried out with OKO-2 GPR by "Logis-Geotech" (Russia) with antenna unit (center frequency of 1700 MHz) on a regular rectangular grid with a space between profile lines of 50 cm. As a result, 26 profiles (13 horizontal and 13 vertical ones) were obtained for each sample (See Fig. 5). We processed the obtained radargrams using signal filtering operations, searching for the diffraction hyperbola, migration procedure to localize the source object and the Hilbert Transform to examine the energy of the reflections (See Fig. 6). The analysis of the hyperbolas has determined the relative permittivity of the studied soil and underlying metatuffs. It is 9 and 6, respectively. The electromagnetic wave length in the soil body is 5 cm, which implies the possibility to distinguish objects sized 1.5-5 cm. This range is consistent with the expected size of the apple tree taproots. After the processing of each profile, the roots were localized using hyperbolas and the surface of bedrocks was determined (See Fig. 7). We found out that the soil profiles of the two experimental samples were different. In the first case, the average depth of bedrocks bedding is 60 cm, and there is a v-shaped decrease in the identified hyperbolas with depth. In the second case, the average depth is 45 cm and most of hyperbolas are located in the first 30 cm. The examination of the location of the roots in horizontal slices within an interval of 10 cm shows a change in the density of the roots distribution as we go deeper (See Fig. 8). The quantity of roots of Sample 1 gradually decreases with depth as well as the occupied area. In addition, the roots tend to concentrate near the trunk. Sample 2 has many roots in the upper part of the profile (0-30 cm) and they occupy a significant area. However, the quantity of roots in the lower part decreases sharply. In both cases, we detected a predisposition of the root locations to the areas of the immersion of the roof of the bedrocks. In addition, a study was conducted concerning the intensity of the reflection amplitudes of both samples at different depths (See Fig. 9). At the near-surface level (20 cm) there are high-energy reflections. They occupy most of the studied area, and this corresponds to a horizon full of roots. A different picture emerges at a depth of about 40 cm. The samples vary in both the reflection intensity and the occupied area. The revealed changes are due to the decrease in the roots quantity and apparently to the medium size of the roots. We examined the correlation between the morphology of the bedrocks top surface and the spatial root architecture of the samples with three-dimensional models made in the form of a cloud of points as a set of established root locations in the XYZ-coordinate system (See Fig. 10). Our studies have demonstrated the possibility to describe the tree root architecture by means of GPR. Using the two samples of the apple tree as an example, the structure of the root system in the soil-ground massif was determined. Also, we established the influence of the bedding roof depth of crystalline rocks on the ontogenesis of a tree. The results obtained prove that GPR is a promising method for studying the underground areas of trees, which provide the possibility to analyze the root system without excavation. Such studies may prove useful in the future for both solving agricultural technology and forest management tasks, as well as for green spaces control in urban areas.
