The measurement velocity of water flow along gravel-rich hill slope is of great interest for hydrological simulation and prediction in alpine watersheds. The alpine terrain is often covered by rock debris of different origin, size, and type, here summarized as ‘gravel’ for the sake of simplicity. To do this, water flow velocity within a gravel layer was measured using the electrolyte tracer method under virtual boundary conditions, an improved method over pulse electrolyte tracer method. Models employing two virtual boundary conditions, a sine function and a normal distribution function, are suggested as preferable for determining flow velocity by simulating the transport processes of electrolytes along a gravel-covered flume at different flow rates (3, 6, and 12 L min−1) and slope gradients (4°, 8°, and 12°). Flow velocities as well as determination coefficients and mean square errors were calculated at five measurement locations (0.3, 0.6, 0.9, 1.2, and 1.5 m) using model simulation. The feasibility and accuracy of velocity measurements for water flow within gravel layers are justified to the extent that the two models fit well with the experimental data, the velocity values predicted by the two models are almost identical and, finally, the relations between velocity and discharge and between velocity and slope gradient are rationally explained. The estimated velocity results also agree well with those previously measured by commonly-used methods such as the pulse boundary method and the conventional dye tracer method. Our results show that the virtual boundary conditions method, when used to measure water flow velocity within a gravel layer, predicts the flow velocity more rationally than the pulse boundary and dye tracer methods, and with greater precision. This study suggests that this method is not only readily applicable for obtaining water flow velocity measurements within gravel layers, but also for understanding hydrodynamics in typical watershed.