Parameter Weighting

Use the options Constraints > Add Constraint > Parameter Weighting or Parameter Weighting > Modify to define a confidence level associated with various regions of the Parameter Reference Model.

Modify Parameter Weighting Constraint dialog box options

 

Constraint type

By Default, the gradient reference model values for the voxel elements is set to zero for all voxel elements chosen appropriately by the inversion. If the inversion model type is MVI, then all three components of the vector voxel model elements are set to zero.

Alternatively you may have a compelling reason to provide a Constant physical gradient value to be used for the entire model. If you have a Voxel model derived through other processes that represents the voxel gradient model, you may choose to provide it instead.

For all model types other than MVI, you may provide a Constant other than zero for the entire model. You may have a prior knowledge of the characteristics of the study area, where, if the geophysical response of the area is generally high, then setting this value to >0 may speed up the process. Alternatively, if you have derived a plausible Voxel model for the study area through earlier processes, you may choose to provide it instead.

For the MVI model types, the Constant option is available but must be specified as a vector. The vector components correspond to the U, V, and W components of the weighting vector within the rotated coordinate system. You may also input a vector voxel where a weighting vector is specified for each cell within the mesh.

Filename

If the Voxel option has been chosen, you are prompted to provide the weighting voxel coinciding with the parameter reference model. This voxel model is re-gridded to the mesh interval and extrapolated to the extents of the padded mesh. This voxel should be in the range of 0 to 1, indicating less or more confidence in the parameter reference model.

Weighting

During constrained inversion, the need often arises to vary the global weight of the parameter reference weighting. Rather than requiring you to generate a new target weight with scaled values, the inversion allows you to globally scale the 0 to 1 range of the target weighting by a constant value. This value is referred to as the global target weighting.

Application Notes

The target weighting allows you to provide a confidence level associated with various regions of the target model. For example, surrounding a logged drill hole, you may have high confidence in the physical property values in the target model, however, the further from the drill hole, that confidence decreases. Setting the target weight to near 1 in the vicinity of the drill hole and decreasing from 1 further from the drill hole will direct the inversion to more closely honour the target model in the vicinity of the drill hole.

Access Seequent Online Learning and select the VOXI guided paths to learn more about effective workflows and key concepts.