3.2.6. IP Inversion Input File

The inverse problem is solved using the executable program ipinv3d_55.exe. The lines of input file are as follows:

Line #	Description	Description
1	mode	inversion mode
2	par tolc	stopping criteria
3	IP Observations File	path to IP formatted observations file
4	Sensitivity Matrix	path to sensitivity matrix file
5	Initial Model	initial model
6	Reference Model	reference model
7	Active Model Cells	active model cells
8	Lower Bounds	lower bounds for recovered model
9	Upper Bounds	upper bounds for recovered model
10	alpha_s alpha_x alpha_y alpha_z	weighting constants for smallness and smoothness constraints
11	Hard Constraints	use SMOOTH_MOD or SMOOTH_MOD_DIFF
12	Weights	weights

Fig. 3.6 Example input file for the IP inversion program (Download ).

3.2.6.1. Line Descriptions

• mode: approach used to determine the trade-off parameter

  • mode=1: the program chooses the trade off parameter by carrying out a line search so that the target value of data misfit is achieved (e.g., \(\phi_d = N\)). par is the chi-factor for the target misfit (usually 1.0).

  • mode=2: the user solves the optimization problem for a user-defined trade-off parameter given by par.

• par tolc: set the mode for stopping criteria for the inversion. If *mode*=1, then the inversion will terminate when the data misfit is equivalent to a chi factor of \(par \pm tolc\)

• IP Observation File: path to an IP formatted observations file

• Sensitivity Matrix: file path to sensitivity matrix file (.mtx) that is output by ipsens3d.exe

• Initial Model: On this line we specify the starting model for the inversion. On this line, there are 2 possible options:

  • Enter the path to a conductivity model

  • If a homogeneous conductivity value is being used, enter “VALUE” followed by a space and a numerical value; example “VALUE 0.01”.

Warning

DO NOT use a starting value of 0 otherwise the inversion cannot compute a step direction at the first iteration. It acceptible to start with something like 1e-5 for all cells.

• Reference Model: The user may supply the file path to a reference conductivity model. On this line, there are 2 possible options:

  • Enter the path to a conductivity model

  • If a homogeneous chargeability value is being used, enter “VALUE” followed by a space and a numerical value; example “VALUE 0.01”.

• Active Model Cells: Here, the user can choose to specify the model cells which are active during the inversion. There are two options:

  • null: this flag is used if all cells below the surface topography are active

  • active cells model: the user can enter the path to an active cells model where 1 denotes cells below the surface and 1 denotes active cells and 0 denotes inactive cells

• Lower Bounds: Lower bound constraints on the recovered model. There are 2 options:

  • Enter the flag VALUE followed by the lower bound value that will applied to all cells

  • Enter the path to model file which contains a lower bound value for every cell in the mesh. Bounds for inactive cells are ignored

• Upper Bounds: Upper bound constraints on the recovered model. There are 2 options:

  • Enter the flag VALUE followed by the upper bound value that will applied to all cells

  • Enter the path to model file which contains an upper bound value for every cell in the mesh. Bounds for inactive cells are ignored

• alpha_s alpha_x alpha_y alpha_z: Alpha parameters . Here, the user specifies the relative weighting between the smallness and smoothness component penalties on the recovered models.

• Hard Constraints: SMOOTH_MOD runs the inversion without implementing a reference model (essential \(m_{ref}=0\)). “SMOOTH_MOD_DIF” constrains the inversion in the smallness and smoothness terms using a reference model.

• Weights: Apply cell and/or interface weights to the inversion. There are two options:

  • null: enter this flag if no additional weights are applied

  • weights file: enter the file path to a weights file to apply weights

3.2.6.2. Format for v5.0

The input file format for ipinv3d.exe v5.0 is almost identical. The only difference are:

• the idisk flag which allows the user to write the sensitivities to disk or store in RAM

• the irest flag which allows the user restart the inversion from a previous iteration

• the ability to change the role of the reference model in the model objective function