To calculate the direct problem, control parameters must be defined during inverse modelling. Some restrictions must be observed:

• In general, a coupled flow and transport model can be calculated. Dispersivity and other model data that is only required for transport modelling cannot be optimised yet, and measured concentrations cannot be included as observation data. Therefore, the transport calculation is currently not scheduled for inverse modelling. One exception is the optimisation of flow parameters for a density-dependent model. In this case, the transport problem must also be calculated.

• In general, it is possible to save the calculated potential heads only for certain time steps or hydrographs of individual nodes in a transient calculation in order to reduce the required storage space. With inverse modelling, however, the potential heads are required for each time step of the transient calculation. For this reason, the potential heads for all time steps must be stored temporarily for a transient inverse modelling run.

• A "warm start", i.e. the continuation of a transient calculation (p. 380), with the null-file is not possible with an inverse modelling run. Saving the data of the transient calculation for a continuation of the iteration (out66 file) is therefore also not useful.

• For a transient three-dimensional or vertical flow model that is to be calculated in saturated/unsaturated mode, the relative K values can be calculated iteratively within a time step. This is not possible with inverse modelling. For such problems, the relative K values must be updated from one time step to the next (i.e. with only one iteration per time step and without damping of the iteration: damping factor = 1.0).

• As a further limitation compared to a normal flow calculation, it is not possible to work with maximum infiltration/exfiltration rates (MXKI, MXKE, MXEI, MXEE) in inverse modelling. However, LKNO observation data can be used to keep leakage rates within a certain order of magnitude.