Initializing spatially distributed ecohydrological models with soil biogeochemistry is computationally expensive, especially when lateral fluxes must be resolved. We developed a hybrid initialization framework that combines 1D flux-tracking spin-up simulations with random forest extrapolation to generate spatially heterogeneous, topography-informed initial conditions. The approach captures the effects of topography and lateral transport while reducing computational costs by up to 90 %.