Static seals are crucial for maintaining the integrity of mechanical systems. However, the complex coupling between surface topography and tribological effects often leads to the formation of irregular textured interfaces, making accurate prediction of static sealing performance a significant challenge. To address this issue, this study develops an enhanced three-dimensional percolation lattice model incorporating two key innovations: an eight-connectivity percolation criterion that captures diagonally connected pore clusters, yielding a more realistic estimate of effective porosity compared to conventional four-connectivity methods; a breadth-first-search-based shortest-path algorithm that quantifies tortuosity while explicitly accounting for the directional coupling between surface texture and macroscopic leakage flow. Hydraulic oil leakage experiments conducted on mated rough surfaces demonstrate strong agreement between the predicted and measured leakage rates.