Parameterization is an important step in setting up an inverse model for automated parameter estimation. A parameter is an input value such as hydraulic conductivity or recharge. Each parameter is altered by the inverse model in an iterative fashion until a set of parameters are found that minimize the residual error.
In order to ensure that the parameter estimation process is stable, the number of parameters should be less than the number of observations unless regularization is being used in conjunction with PEST. Parameterization is the process of selecting input values, and grouping them into parameter sets to reduce the number of independent parameters.
The most common form of parameterization is zonation. With this approach, an array of input values is divided into zones where all of the cells in each zone share a single parameter value. This concept is illustrated in the figure below. The entire array of values is represented by three parameters.
Zonation is most easily accomplished in GMS by assigning key values to polygons in the map module. It can also be accomplished by assigning key values to a selected set of cells in the 3D grid module using the Cell Attributes command.
The following MODFLOW input values can be defined as parameters in GMS:
|
Package |
Option |
|
LPF/HUF |
HK - Horiz. hydraulic conductivity |
|
Recharge |
RCH - Recharge value |
|
ET |
ET - Max ET flux |
|
Well |
WELL - Pumping rate |
|
River |
RIV - River conductance |
|
General Head |
GHB - Conductance |
|
Drain |
DRN - Conductance |
Related Links:
Automated Parameter Estimation