Inversion Configuration

This is used in the name of the inversion folder and to track different inversion runs. Data Type: String.

The amount of GPU memory. Data Type: Positive Float.

The distance between source-receiver pairs sampled from the data, in meters. Data Type: Positive Float.

Whether to use topography in the inversion. If False, the earth is approximated to have a flat surface at z=0. Data Type: Boolean.

The scheduling strategy for the inversion process. Options are: “minimum” (uses the minimum number of GPUs required) or “fixed” (uses the number of GPUs specified in the num_gpu parameter). Data Type: String.

The uncertainty floor in the weighting function, given by the formula:

$\frac{1}{data\_eps + data\_std \times \|data\|}$

Data Type: Positive Float.

Uncertainty fractional percentage, given by the formula:

$\frac{1}{data\_eps + data\_std \times \|data\|}$

Data Type: Positive Float.

The noise threshold value. Data points below this value will have a weighting of 0 and will not contribute to the inversion. Data Type: Positive Float.

The index of the first time channel used in the inversion. Data Type: Positive Integer.

The index of the last time channel used in the inversion. Data Type: Positive Integer.

The optimizer used for the inversion. Options: “PGNCG” (Projected Gauss-Newton Conjugate Gradient) and “IGN” (Inexact Gauss-Newton). Data Type: String.

Constants for the optimizer:

• maxIter: Maximum number of iterations. Data Type: Positive Integer.
• LSshorten: Line search shortening factor. Data Type: Positive Float.
• maxIterLS: Maximum number of line search iterations. Data Type: Positive Integer.

Settings for the beta schedule, controlling the trade-off between data misfit and model regularization. For example, if the cooling rate is 3 and the cooling factor is 2, the beta value is divided by 2 every 3 iterations.

• cooling_factor: Cooling factor. Data Type: Non-negative Float.
• cooling_rate: Cooling rate. Data Type: Non-negative Integer.

The beta value for the inversion process. Data Type: Positive Float.

Alpha values for the inversion:

• alpha_s: Smoothness. Data Type: Non-negative Float.
• alpha_x: X-direction. Data Type: Non-negative Float.
• alpha_y: Y-direction. Data Type: Non-negative Float.

The reference model sigma value for the regularization. Data Type: Positive Float.

Whether to apply depth weighting in the inversion. Data Type: Boolean.

Constants for the weighting:

• type: Weighting type. Data Type: String.
• alpha: Weighting alpha value. Data Type: Non-negative Float.
• penetration_limit_start: Start value for penetration limit. Data Type: Non-negative Float.
• penetration_limit_end: End value for penetration limit. Data Type: Non-negative Float.
• limit_weight_max: Maximum weight limit. Data Type: Non-negative Float.

Whether to apply an upper bound on conductivity values. Data Type: Boolean.

The upper bound value in Log10(mS/m). Data Type: Positive Float.

Whether to apply sparse norm regularization. Data Type: Boolean.