Fluid flow in a single fracture is commonly simulated using the Reynolds equation. Recent work suggests that this depth‐averaged approach underestimates head loss in regions of changing aperture. Implementing an ad hoc correction in the numerical formulation of the Reynolds equation allows us to modify local head loss, and calibrate simulation results to existing experimental data. Calibrated flow fields provide an improved estimate of longitudinal dispersivity, demonstrating the importance of adequately describing local head loss.