Seasonal snow density is a key determinant of snow water equivalent (SWE) and melt dynamics. However, characterizing its spatiotemporal variability is challenging, particularly in the shallow, ephemeral snowpacks of the eastern United States. This study evaluates the potential of integrating Ground Penetrating Radar (GPR) and LiDAR onboard an Uncrewed Aerial System (UAS) to assess the applicability of dielectric permittivity models for density retrieval in a transitional melt environment. Three field surveys were conducted at Kingman Farm, New Hampshire, where we collected UAS-GPR and in situ measurements of snow depth, density, and liquid water content (LWC) across open and sheltered fields between late February and early March, 2025. LiDAR‑derived snow depth and GPR travel time were used to retrieve snow permittivity; these values were compared against theoretical estimates derived from seven dielectric models, including the physically-based Complex Refractive Index Model (CRIM) and six empirical formulations that account for snow wetness. Field observations revealed a rapid transition from dry to wet snow, with GPR-derived permittivity exhibiting a non-linear increase as LWC exceeded 3–4 %. Comparisons showed that the physically based CRIM and the empirical Lundberg and Denoth models achieved comparably close agreement with GPR-derived permittivity (r = 0.70–0.71, RMSE ≈ 0.35). The remaining empirical models showed moderately larger error (RMSE = 0.40–0.48), while the Webb model showed weaker agreement (r = 0.54, RMSE = 0.79). These findings suggest that while empirical relationships performed reasonably well in transitional snowpacks outside the regions where they were established, applying physically-based models to UAS-GPR data offers an alternative and promising pathway for characterizing bulk snow properties in hydrologically complex environments.