Advanced application of ground-penetrating radar in underground tree root systems detection and mapping

• Main Author: Luo, Wenhao
• Other Authors: Abdulkadir C. Yucel
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2023
• Subjects: Engineering::Electrical and electronic engineering
• Online Access: https://hdl.handle.net/10356/170919
• Institution: Nanyang Technological University

Tree roots absorb water and nutrients and maintain the trees’ health and growth. Trees with unhealthy roots can easily fall, resulting in loss of lives and damage to properties. Therefore, it is of critical importance to be able to map the tree root systems in order to monitor their health. The healthy root structure ensures its strong support to the tree and prevent the collapse of the tree and thus, avoiding the loss of lives and properties [1],[2]. The spatial distribution of the roots is investigated recently not only because of its critical impacts on the ecosystems but also to reduce the risk of conflicts between the root system and the infrastructure elements, such as sidewalks, curbs, and building foundations [3], [4]. Many methods have been developed to investigate tree roots. Traditional methods, such as excavation and uprooting, are destructive, labor-intensive, and time-consuming, thus are not preferred in long-term studies [5, 6]. The utilization of ground-penetrating radar (GPR) has experienced a growing trend in the field of imaging subterranean tree root systems. This surge in popularity can be attributed to its inherent merits, including non-destructive capabilities, portability, and straightforward deployment. For now, there are some limitations when applying GPR in tree root systems detection and reconstruction. 1) There is always a trade-off between penetrating depth and resolution of GPR; 2) Inhomogeneity of the realistic soil condition always reduces the GPR detection accuracy and results in high noise level; 3) Undulating terrain over the realistic ground causes the wrong estimation of the position and dimension of the roots; 4) The accuracy of reconstruction of the distribution of root system deteriorates when it comes to complex root systems. In order to enhance the GPR detection capability, a balance between the high resolution achieved via higher frequency samples and the deep penetration depth achieved via lower frequency samples needs to be achieved. Researchers have proposed several multi-frequency GPR data fusion algorithms for the classical pulse center frequency GPRs [7] [8] [9, 10] [11]， the center frequency for these pulse GPRs are 100 MHz, 200 MHz and 400 MHz for permafrost subgrade condition assessment and 205MHz, 500MHz and 800MHz for limestone detection. Other researchers have proposed methods that vary the bandwidth with time/depth for step-frequency continuous wave (SFCW) GPRs. SFCW GPRs have attracted much interest because of its portability and accurate energy level controllability over the whole frequency band[12, 13] [14] [15]. In this thesis, we address these four main limitations of the GPR. Firstly, we propose a depth-adaptive time-frequency filtering technique. A time-frequency filter is derived based on the time-frequency properties of the response signal. This proposed technique is applied to data collected by the SFCW GPR system from actual sites with different scenarios. In our approach, we employ a combined time-frequency domain analysis technique known as Short Time Fourier Transform (STFT). This method enables us to monitor the evolution of frequency spectrum density over a given period. The weighted linear regression (WLR) method is applied to derive the filter window. Through this approach, we can filter out frequency samples that are not of interest. At the same time, the proposed method is adaptive to the soil characteristics. The filter window derived changes base on the soil properties such as permittivity and conductivity. The proposed technique is validated using data collected within and out of monsoon season. By adopting a conversion method that is a variant of the chirp Z-Transform, the proposed method provides highly improved radargrams of the subsurface roots systems. The conventional GPR data processing for tree roots detection ignore the random and complex nature of the heterogeneous soil and assume the soil’s relative permittivity to be constant (homogenous) throughout the survey region. This yields an inaccurate position estimation of the tree roots. Furthermore, the soil’s spatial heterogeneity introduces unwanted environmental clutter in the mapping of the tree root. To address these issues, a data processing framework is proposed for the accurate mapping of tree roots in heterogeneous soil environments. The proposed framework combines four techniques to be applied consecutively: 1) A hyperbola extraction method based on a column-connection clustering algorithm is used to extract individual hyperbolae in B-scans, eliminate mutual influence in the process, and suppress noise. 2) An improved Hough transform (HT) technique is adopted to estimate the relative permittivity of each root’s surrounding soil environment for each extracted hyperbola. 3) A retrieval method is employed to restore roots scenarios by dealing with each hyperbola individually. 4) Finally, individually restored features are combined in the final image. The images obtained via the proposed framework show the horizontal and vertical positions of tree roots accurately with less background noise. Next, we address the issue of the ground surface. In current literature, the ground surface above the tree roots is assumed to be flat, and standard processing methods based on hyperbolic fitting are applied to the hyperbolae reflection patterns of tree roots for detection purposes. When the surface of the land is undulating (not flat), these typical hyperbolic fitting methods becomes inaccurate. This is because, the reflection patterns change with the uneven ground surfaces. When the soil surface is not flat, it is inaccurate to use the peak point of an asymmetric reflection pattern to identify the depth and horizontal position of the underground target. The reflection patterns of the complex shapes due to extreme surface variations cause difficulties for subsequent analysis. Furthermore, when multiple objects are buried under an undulating ground, it is hard to judge their relative positions based on a B-scan that assumes a flat ground. In this thesis, a root fitting method based on electromagnetic waves (EM) travel time analysis is proposed to take into consideration the realistic undulating ground surface. Both the wheel-based (WB) GPR and the antenna-height-fixed (AHF) GPR System are presented, and their corresponding fitting models are proposed. The effectiveness of the proposed method is demonstrated and validated through numerical examples and field experiments. Precedent 3D reconstruction methods are found to be effective in mapping simple, smooth root structures. However, repetitive and dense B-scans are needed, otherwise, the retrieved roots' spatial distribution and growth extension trend accuracy would deteriorate with the increase in the root systems’ complexity. The Figures of representative of simple tree root and complex tree root system are shown in Fig. 1.1. To address these issues, for the first time, we explore the possibility of integrating the horizontal angle information of the tree roots and a slice-relation-clustering (SRC) algorithm to reconstruct the complex tree root systems in a 3D manner. The proposed framework, which takes the roots’ horizontal angle as an analyzing condition instead of biological properties that are similar among neighboring branches used in existing methods, clusters pre-processed and focused 2D reflection patterns from the same single root together. The whole roots system is the combination of every single root cluster. Real measurement results show that our proposed method achieves a high efficiency in accurate root system reconstruction. In summary, in this thesis, we have studied and worked on applying GPR more effectively to the detection of tree roots. We have obtained an adaptive frequency band filter along the time according to the spectrum analysis of the reflection signals. We have also taken the changes in the dielectric constant of the heterogeneous soil environment and the variation of the land surface's topography into consideration when mapping roots under the 2D scanning. Furthermore, the horizontal angles of the roots have been considered when making a 3D reconstruction of tree root systems.