Ground-penetrating radar (GPR), due to its efficiency and non-invasive nature, has become an important tool for detecting the permafrost table, overcoming the limited spatial coverage and high costs associated with drilling and in situ temperature monitoring. Compared with the commonly used 50–100 MHz antennas, the potential of high-frequency antennas to improve detection accuracy and interface resolution has not been fully explored. To address this gap, this study introduces a multi-strategy interface identification method incorporating envelope analysis. Field experiments were conducted in the island-like permafrost zone of the Da Xing’anling Mountains, Heilongjiang Province, using shielded GPR systems operating at 250 MHz and 500 MHz to detect the permafrost table. Potential interfaces were extracted using centroid and edge-detection algorithms and validated against ground temperature observations. The results indicate that: (1) integrating GPR with multi-source data enables accurate estimation of active layer thickness, and the envelope-based multi-strategy approach is effective for interface identification; (2) the 250 MHz antenna is better suited for capturing broader subsurface structures, while the 500 MHz antenna provides higher resolution for shallow layers—combining the two enhances overall interpretive quality; and (3) snow cover significantly affects electromagnetic wave propagation, reducing the accuracy of radar detection. This study provides valuable guidance for engineering investigations, site selection, and foundation design in permafrost regions, contributing to improved precision and efficiency in GPR-based detection of the permafrost table.

Isolated permafrost is widely distributed in freeze–thaw transition zones, characterized by blurred boundaries and strong spatial variability. Traditional methods such as drilling and electrical resistivity surveys are often limited in achieving efficient and continuous boundary identification. This study focuses on a typical isolated permafrost region in Northeast China and proposes a boundary detection strategy based on multi-frequency electromagnetic (EM) measurements using the GEM-2 sensor. By designing multiple frequency combinations and applying joint inversion, a boundary identification framework was developed and validated against borehole data. Results show that the multi-frequency joint inversion method improves the spatial identification accuracy of permafrost boundaries compared to traditional point-based techniques. In areas lacking boreholes, the method still demonstrates coherent boundary imaging and strong adaptability to geomorphological conditions. The multi-frequency joint inversion strategy significantly enhances imaging continuity and effectively captures electrical variations in complex freeze–thaw transition zones. Overall, this study establishes a complete non-invasive technical workflow—“acquisition–inversion–validation–imaging”—providing an efficient and scalable tool for engineering site selection, foundation design, and permafrost degradation monitoring. It also offers a methodological paradigm for electromagnetic frequency optimization and subsurface electrical boundary modeling.

Permafrost maps are important for regional planning in cold regions, but binary products do not represent gradients in thermal conditions, seasonal thaw response, or mapping confidence. This study developed an uncertainty-aware regional permafrost-condition screening framework for the Daxing’anling region during 2003–2022 by integrating TTOP-derived mean annual ground temperature (MAGT), reconstructed permafrost-occurrence probability, and Kudryavtsev-model-derived active-layer thickness (ALT). MAGT ranged from −4.11 to 5.28 °C, with a mean of −0.28 °C, and comparison with 23 published borehole records from 17 reported locations yielded r = 0.756 and RMSE = 0.419 °C. Explicit measurement depths were available for 12 of the 23 records and ranged from 10 to 15 m. Model-derived ALT ranged from 1.480 to 1.864 m. The complete point-scale evaluation using all 12 valid maximum depth of seasonal thaw (MDST) records yielded an RMSE of 0.92 m. Adding ALT changed 21.05% of valid-pixel assignments. Cold–low-response permafrost, Cold–moderate-response permafrost, Warm–enhanced-response permafrost, Near-thaw transitional permafrost, Marginal/low-confidence permafrost, and Non-permafrost occupied 7.3%, 17.1%, 14.9%, 4.1%, 13.9%, and 42.8% of the domain, respectively. Full Monte Carlo uncertainty propagation retained 78.65% modal agreement with the deterministic baseline, and 56.65% of the domain had a maximum class-membership probability below 0.60, whereas threshold-only perturbation retained 98.92% agreement. The framework is therefore suited to regional investigation and monitoring prioritization; project-level engineering decisions require direct geotechnical and deformation-based evidence.