Enhanced GPR detection via HHT for discontinuous permafrost interface identification

Discontinuous permafrost degradation poses significant challenges to the stability of highway infrastructure in cold regions, highlighting the importance of precise interface detection. The attenuation of electromagnetic waves in traditional ground penetrating radar (GPR) limits its detection performance. To overcome this limitation, a three-layer discontinuous permafrost model was developed using measured temperature data. GPR surveys with various antenna frequencies were conducted over the model, and the acquired signals were analyzed using the Hilbert-Huang Transform (HHT). The influences of permafrost thickness, boundary depth, and the dielectric constant of the surrounding clay on the first intrinsic mode function (IMF1) peaks were then systematically investigated under different antenna frequencies. The results demonstrate that the instantaneous frequency peaks of IMF1 derived from HHT closely correspond to the discontinuous permafrost interfaces. IMF1 successfully identified interfaces for permafrost layers at depths of 3 m with thicknesses between 3 m and 6 m. With increasing permafrost thickness, the error associated with the upper interface increased, whereas the error at the lower interface initially rose to 1.8 m and subsequently decreased to 1.5 m. Moreover, as the dielectric constant of the surrounding clay increased, the errors at both the upper and lower interfaces exhibited an increasing trend. Nevertheless, IMF1 was less effective in detecting permafrost boundary depths than thickness, and it could only identify layers with boundary depths of 3–4 m. The proposed method enhances interface clarity in GPR images and provides quantitative guidance for practical applications, supporting low carbon and intelligent cold region engineering.