Preparation in the model file
The parameters to be calibrated must first be defined in the model file. Thus, for example, water course sections are combined or the hydraulic permeabilities are divided into zones.
Within a zone, either a single K-value or storage coefficient (number of parameter values within the zone = 1) or a field of K-values or storage coefficients (number of parameter values within the zone > 1) can be calibrated.
For the calibration of leakage coefficients, the surface waters must be defined as LERA. A river or stream can be divided into several sections, these sections must be equivalent in LERA and Z-LR. Within a section only one value can be specified and optimized.
Structure of the parameter file
The parameter file must be in *.csv format. The columns are separated by a semicolon, as decimal the point is predetermined. The structure of the file looks like this:
Example of a parameter file
In column A the type of the parameter is defined. The following types of parameters are currently implemented in the inverse modelling: KWER, KSPE, LERA, FLAE.
In column B the zones of Z-KW and Z-SP are taken over. The parameter LERA in column B is always 0. The assignment of Z-LR is made in column C.
In column C is the element number of the base defined.
Column D is the starting value for each parameter. In the first inverse calculation, this value is read.
Column E contains the standard deviation. This corresponds approximately to the expected errors of the size. For potential measurements, this can be 5-10 cm, for quantities are 10% good values. In logarithmic variables such as KWER the standard deviation is given in orders of magnitude (eg K-value = 1e-03 + / - 0.5 magnitude).
Column F contains the estimated value of the parameter for the generation of the residual.
Columns G / H contain the lower and upper limit of the estimated value of column F.
Column I: 0 for variable / 1 for fixed (making sure that the desired value is defined in column D!) 
Shown in the following figure.
Column J contains the name of the parameter that appears in the output.
Left: Bases and initial K-values. Right: The variable K-values are circled (column I=0). The K-values at the other bases are constant (column I=1), but have to stay in the parameter file for the interpolation.
Structure of the observation file
The existing measurements are saved in a file without a special format (e.g. text file). The measured values can be present steady state or transient. The structure of the file is as follows:
Line 1-3 comments
Line 4 empty
Line 5 ZEITEINHEIT MENG SEKUNDE (MINU STUN TAG_ JAHR)
Line 6 ZEITEINHEIT ZEIT SEKUNDE (MINU STUN TAG_ JAHR) OR
Line 6 BEZUGSDATUM XX.YY.ZZZZ
Line 7 empty
Line 8-x measured values:
POTE Name_1 // comment (Attention: no blancs in the name!)
x-coordinate y-coordinate [z-coordinate in 3D]
Zeitpunkt Measurement Standard deviation OR
DATUM Measurement Standard deviation
KNOT Name_2
Number of nodes n
n node numbers (10 per line)
Zeitpunkt Measurement (Sum over all n nodes) St.dev. OR
DATUM Measurement (sum over all n nodes) St.dev
LKNO Name_3
Number of nodes n
n node numbers (10 per line)
Zeitpunkt Measurement (Sum over all n nodes) -St.dev. OR
DATUM Measurement (Sum over all n nodes) -St.dev.
In line 5 and 6, the time units for the quantity and the time unit for the time scale are set. For steady state measurements, the time unit is defined as ZEIT with the time 0.0 in the file. For transient measurements can be selected if an absolute time scale in a time unit (e.g. hour: 0.0, 1.0, 1.5, 2, 20) is used or if it is worked with a reference date and a date time scale.
From line 8 the measured values follow. There are 3 possible categories: POTE for potential measurements, KNOT for point withdrawal / infiltration amounts and LEKN for measured infiltration and exfiltration rates at a water course section. The control word for the category is followed by the name of the measuring point. It must be ensured that there are no blanks in the name.
A tab is not allowed between the x- and y-coordinate.
After the point of time follow the measured value and its standard deviation. The standard deviation is used for the weighting of this value. This is described in detail in chapter “Minimization of the target functions“.
Examples of a steady-state and transient observation file follow:
Observation data for a steady state calibration
Observation data for a transient calibration