Abstract: River embankments often suffer from weak foundation conditions, making them susceptible to seepage, piping and slope failures during flood seasons. Regular, rapid and accurate inspections are essential for ensuring the safe operation of embankments. As an effective non-destructive geophysical technique, Ground Penetrating Radar (GPR) has been widely applied in potential hazard detection in embankments. However, raw GPR signals often contain considerable high-frequency noise and background interference, making it difficult to extract reliable features under complex field conditions. Traditional spectral analysis methods, such as the Fourier Transform, typically suffer from spectral smearing and unclear dominant frequency components, limiting their practical application. To address these limitations, a three-dimensional spectral energy ratio imaging method was proposed for potential embankment hazard detection based on combined wavelet-fourier transforms. In this approach, wavelet transform is first applied to analyze non-stationary and transient components of the radar signal, allowing for multiscale decomposition of its frequency characteristics. Subsequently, a short-time Fourier transform is used to calculate the temporal evolution of frequency-domain information and extract dominant frequency components and energy distribution across time windows. Based on this, a classification criterion of hazard types is developed using the spectral energy ratio, enabling the effective separation of loose zones and water-enriched zones within the embankment. This approach facilitates high-resolution, efficient and non-destructive identification of internal embankment anomalies. Finally, a field investigation was carried out on the triangular joint polder embankment in Yongxiu County, Jiujiang City, Jiangxi Province. The results demonstrate that the proposed method successfully detects and delineates loose and water-rich areas within the embankment, indicative for hazard assessment and routine inspection of earthen dams.