Regrid (Horizon)

Use Regrid to convert a horizon to convert between surveys, or to a different spatial sampling. 

Regrid is commonly used to convert:

  • From X/Y horizons to 3D IL/CL horizons
  • From 2D horizons to X/Y
  • From 3D horizons to 2D lines
  • Multi-survey horizons onto a single survey

Note: The Regrid button is disabled if a horizon is selected in the map view that is waiting for another operation (such as propagation). Insight notifies you to complete the previous operation before continuing.

Where is the regrid operation?

Configuring Regrid

Natural Neighbour/IWD Settings

Splines in Tension Settings

Completing the operation

Where is the regrid operation?

Regrid is found in the Map View, under the Operations tab (along with other horizon operations).

  1. Open the Map View.
  2. Expand the left panel and open the Operations tab.
  3. At Operation, select Regrid.

Tip: Click the Help icon to read more information about this operation.

Configuring Regrid

  1. In the navigation bar, select the Horizon to regrid.
  2. In the Survey field, select the output survey:
    • To create X/Y grids, select <<X/Y>>.
    • To create IL/CL grids, select the 3D survey.
    • To create horizons on 2D lines, select the 2D survey.
      • The input horizon must overlap or intersect the 2D survey
    • For the gather horizon, enter the gather domain values.
  3. In the Area field, select a polygon/probe to constrain the operation.
  4. Interpolation Settings: (see How Interpolation Settings Work)
    • Interpolate:
      • If disabled: Only output results where input values exist
      • if enabled: Output results near input values (use interpolation)
    • Algorithm: Select one of the two available algorithms:
      • Natural Neighbour/IWD: This algorithm is a blend of natural neighbour interpolation and inverse distance weighted interpolation, using the natural neighbour set of input points for each output location.
      • Splines in Tension: This is an implementation of the algorithm described in "Gridding with Continuous Curvature Splines in Tension" by Smith & Wessel (Geophysics, 3, 293-305, 1990). This algorithm requires a regular grid and cannot be used to interpolate or extrapolate onto a 2D survey.
  5. Extrapolate: If selected, enable extrapolate to completely fill a specified region according to the settings.
  6. Use trend: Select to interpolate/extrapolate the input horizon relative to the best-fit plane through the existing horizon points.
  7. Click Regrid.

Natural Neighbour/IWD Settings

Faults: Select to prevent the regridding calculation from crossing fault planes.

Fault Polygon: Set a polygon to avoid interpolating the area inside the selected fault polygon. Insight honours any enabled faults in the Control Panel. Unassigned fault sticks are not used.

Max search: When interpolating, a point further than this distance from its nearest neighbour will not be interpolated. More precisely, no region will be interpolated if the smallest radius circle through any three points around exceeds this value. This has the effect of filling small gaps and holes but not larger ones. When extrapolating, this parameter has no effect and all holes are filled.

Spreading: Natural neighbour weighting. Spreading power and Smoothing power control the relative weighting of the natural neighbour and inverse distance portions of the calculation. The value is the power applied to the natural neighbour interpolation weight and inverse-distance weight respectively. A value of 0 means that weight is not applied. The following are some examples:

Spreading power Smoothing power
1 0 Pure natural neighbour interpolation (the default)
0 1 Inverse-distance weighted interpolation
0 2 Inverse-distance-squared interpolation
0 50 Approximates nearest neighbour interpolation

Smoothing: Inverse-distance weighting.

Strike/dip: Define the weight in the strike direction.

Direction (degrees): The azimuth of the strike axis, in degrees (clockwise from North).      

Strike direction (in degrees) and Strike vs. dip weight bias the weighting in a particular direction. For example a strike direction of 45 and a strike vs. dip weight of 2 means the weight of an input point to the north-east or south-west of an interpolated location will be doubled compared to the weight of an input point to the north-west or south-east.

Splines in Tension Settings

Tension: A value between 0 and 1. When tension is 0 the result is a natural bicubic spline, or minimum curvature interpolation. This may result in oscillations and overshoots. When tension is 1 the result is equivalent to a steady-state temperature field with heat sources or sinks at the input points. This surface will have no oscillations or overshoot with local maxima or minima (relative to the planar trend if used) only at the original horizon points.

Expand grid: The algorithm works on a rectangular grid and all input points must be within the bounds of that grid to be used. If an area constraint is supplied and does not cover the entire horizon the rectangular grid will be expanded by this distance on each side to ensure that nearby input points are included.

Advanced Settings

Most of these settings are for additional control of the solver. You should not generally need to change these.

Boundary tension: By default, the tension at the boundary of the grid will be the same as the internal Tension parameter. You may specify a different tension here.

Constraint weight: The algorithm finds a least-squares solution for two conditions: a curvature constraint for each output point and an interpolation condition for each input point. This setting controls the relative weight of the curvature constraint compared to the interpolation, with a higher value giving more weight to the curvature constraint.

Damping: The damping parameter for the LSQR solver. It may be used to prevent overfitting.

Tolerance: The tolerance parameter for the LSQR solver. The default value works well for almost all surfaces. In rare cases such as very small grids or very sparse input, it may be helpful to reduce this value.

Max iterations: Limits the maximum iterations allowed for the LSQR solver. Note the solver is run several times at increasingly fine grid resolutions and this is the maximum number of iterations for each of those stages individually, not a total.

Fitting: The algorithm improves on the one published by fitting all input points rather than discarding those which are not the closest to an output location. It is possible to emulate the original behaviour of the algorithm by changing this setting from the default Fit all input points to Use only one input point per output grid cell.

Completing the operation

After running the operation:

  • At Class: Choose the Display Class for the results
  • Use min and max from class settings (checked)
    • Use amplitude ranges from the class. Update the range in the display class.
  • Use min and max from class settings (not checked)
    • Set specific a min and max value for this result.
    • Click Adjust to set the Min and Max from the extracted amplitude.
  • Click Discard to cancel this calculation and delete the results.
  • Click Create to create a new horizon containing the extracted values.
  • Click Add property to store the result as a horizon property.

Note: By default, the new horizon will be named: “Well Marker: marker_name (domain survey)”.  For example, the operation above will yield a new horizon called "Well marker: Sand_B5_base (TVDSS (m) IL/CL)". To rename the horizon, see Horizon Details and Configuration.