[Objective] In response to the urgent need for preventing and controlling seepage disasters in shallow gassy sandy soil during major engineering construction projects in the Pearl River Estuary, this study aims to develop an efficient prediction method for unsaturated hydraulic properties based on physical mechanisms, thereby overcoming the technical bottlenecks of long testing cycles and high costs in traditional experiments. [Methods] Using typical gassy sandy soil from shallow gas reservoirs in the Pearl River Estuary, soil-water characteristic curve (SWCC) tests using a pressure plate apparatus and unsaturated permeability tests were conducted. According to the abstract conceptualization of the AP model, the soil internal structure was assumed to consist of a vast number of interconnected pore channels with different pore sizes. Based on the equivalent capillary tube model theory, under matric suction increasing from low to high, pore channels drained sequentially from larger to smaller sizes. Then, by assuming that the pore radius and matric suction satisfied the Young-Laplace equation, the relationships between sandy soil particle size distribution data, SWCC, and unsaturated permeability coefficient were established. Consequently, a prediction method for the SWCC and unsaturated permeability coefficient of sandy soil based on particle size distribution was proposed. [Results] Experimental results showed that the particle size distribution of the sandy soil was closely related to its water retention characteristics. Its air entry value (AEV) was generally below 10 kPa, and the residual volumetric water content was typically higher than that of pure quartz sand, approximately 8%. The residual matric suction corresponding to residual volumetric water content ranged from 10 to 20 kPa and was influenced by the fine particle content. The scaling factor α had a critical impact on the predictive performance of the AP model and showed a better correlation with the normalized volumetric water content (θ/θ_s). The optimal fitting equation was α=1.243 1+0.369 4exp(-4.9θ/θ_s). [Conclusions] The pore channels in the soil are categorized into n grades (r₁, r₂, r₃, …, r_n) from largest to smallest pore radius. Under unsaturated conditions, water in the larger pore channels is drained, and only channels from grades m to n (r_m, …, r_n, where m<n) remain water-filled, serving as the main pathways for unsaturated seepage. Based on the particle size distribution curves of sandy soil samples, the mass fraction of the i-th component (with a spherical particle radius of R_i) in the samples is w_i, from which the relative permeability coefficient of the sandy soil can be directly calculated without conversion via the SWCC. A comparison with a typical statistical model—the Jackson model—is conducted, which verifies that the method proposed in this study for predicting the unsaturated relative permeability coefficient of sandy soil based on the particle size distribution curve is reasonable and feasible.