Research status and prospects of ground-penetrating radar in the detection of termite nests in embankments

Termites build nest systems within embankments, damaging the integrity of the soil structure, which is one of the main causes of major engineering accidents such as piping and landslides, seriously threatening the safe operation of China’s water conservancy projects. According to statistics, the direct economic losses caused by termite damage in China exceed 2.5 billion yuan annually. Moreover, with climate warming of the climate and the northward shift of the rain belt, the activity range of termites continues to expand, affecting 18 provinces including Jiangsu, Zhejiang, Anhui, and Henan. The prevention and control of termite infestation in embankments is urgent. Ground-penetrating radar technology, with its rapid, non-destructive, and high-resolution imaging capabilities, demonstrates significant advantages in detecting termite nests and has become an important research direction in this field. However, this technology still faces many challenges in practical applications. This paper systematically explains the technical principles of ground-penetrating radar for detecting termite nests, introduces the core detection method of reflection and the basis for equipment selection, and analyzes the structural characteristics (including the main nest, secondary nests, and tunnel network) and geophysical properties (conductivity, dielectric properties, density differences) of termite nests, clarifying their physical property basis for detection. Based on this, the paper comprehensively reviews the research and application progress from four aspects: detection influence mechanisms, data processing and analysis, integrated geophysical exploration techniques, and intelligent detection systems. In terms of detection influence mechanisms, existing studies have revealed the constraining relationship between antenna frequency and detection depth, clarified the rule that high water content leads to intensified signal attenuation, and summarized the differences in signal characteristics between nests and tree roots, cavities, etc.; however, understanding of the combined effects of multiple factors remains insufficient, and the fidelity of numerical simulations is limited. In terms of data processing and analysis, techniques such as singular value decomposition have been applied for signal denoising, and indicators such as reflection coefficient and peak frequency can be used for nest identification, but the processing still relies on experience and lacks dedicated algorithms and a multi-parameter joint discrimination system. In terms of integrated geophysical exploration technology, the joint application of ground-penetrating radar and the high-density resistivity method has been preliminarily implemented, effectively reducing the misjudgment rates, but there is a lack of standardized technical procedures are still lacking. In terms of intelligent detection systems, deep learning algorithms have achieved classification and recognition of nests and interference objects, but the model generalization remains limited, and the recognition accuracy for deep and small-sized nests needs improvement. In response to the above technical bottlenecks, this paper proposes the following five future research directions and corresponding solutions: first, promote coordinated innovation in detection equipment and methods, develop a composite antennas integrating low-frequency penetration and high-frequency resolution, and enhance detection capability for medium- and shallow-depth nests; second, deepen the research on environmental coupling mechanisms, use CT scanning and 3D printing technologies to construct a high-fidelity physical models of nests, and establish a quantitative relationships between nest parameters and radar spectral characteristics; third, improve data processing and multi-parameter analysis capabilities by extracting multi-dimensional features such as amplitude, phase, and frequency, and establish a dedicated discrimination system for termite nests; fourth, advance integrated geophysical exploration technologies by establishing a “general survey–detailed survey–precise survey” three-level detection system, and formulating data fusion and interpretation standards; fifth, develop end-to-end intelligent detection systems by integrating deep learning to achieve automation and intelligence from data acquisition to risk assessment. The coordinated advancement of these research directions is expected to overcome existing technical bottlenecks and provide a theoretical basis and technical reference for achieving precise and efficient detection of termite nests.