Injected at the Sleipner storage site is migrating into several thin layers. Using a tuning relationship, two different layer thicknesses can give the same reflection amplitude, and it is then not possible to go from amplitudes to layer thicknesses without further constraints. Exploiting spatial and time-lapse dependencies in the reflection amplitude maps makes it possible to resolve this ambiguity and create layer thickness maps when the flow is gravity dominated. The topography of the sealing cap rock was used as an optimization parameter. Tests were done on synthetic data and real data from the Sleipner injection. The resulting topography map for the Sleipner case deviated by 5.3 m on average from simple time-depth mapping, which is within the mapping uncertainty. Although the predictive power is limited, outputs of the method can be used to check if the flow matches a gravity-dominated model or if other flow mechanisms are needed to explain the observations.