A customer commented recently that the kriging algorithm was taking a lot of time to complete for a large dataset. Kriging is computationally expensive and because his dataset had several million points, we suggested he try IDW, direct gridding or minimum curvature instead.Here is a list of the gridding algorithms available and a brief description of why they might be chosen:
Minimum Curvature can be used when data is sparsely sampled and the surface is expected to be relatively smooth or continuous between data points.
Kriging is a geostatistical method that determines the most probable value at each grid node based on a statistical analysis of the entire data set. Because of this, it is computationally expensive. Kriging can be used if the data is variable between sample locations, known to be statistical in nature, or for poorly sampled/clustered data. This algorithm is available for 2D and 3D gridding.
Bi-directional line gridding is designed to rapidly interpolate roughly parallel line-based data, especially if there is a high sample density down the lines relative to the line separation. The interpolation uses linear, minimum curvature or Akima splines. It is only available in Oasis montaj and cannot be used to interpolate randomly distributed XYZ data.
Tin gridding results in output grid cell values that closely match the magnitude of the original data at known XY positions. The interpolation is entirely local and every point will be influenced either by its nearest or natural neighbours depending on the parameter chosen. Tin gridding can be used for irregularly sampled data. This algorithm is available in Oasis montaj and Target. In Target for ArcGIS, this algorithm is only available when gridding drillhole data for a plan maps, section maps or plan grids in 3D maps.
Inverse distance weighted gridding (IDW) can be used when data is sparsely sampled and the surface is not expected to be smooth or continuous between data points. The data points are weighted so that the influence of one point relative to another declines with distance. Three key parameters that can be set that will influence the interpolation are search radius, weighting power and weighting slope. IDW can be used to create 2D grids and in the April 2012 release, IDW will also be an option for 3D gridding.
Direct gridding is designed for highly sampled or oversampled data such as LiDAR. The output value will be determined based on the minimum, maximum or the mean of the data points that fall within the grid cell. This algorithm is available in Oasis montaj and Target for 2D gridding. In the April 2012 release, direct gridding will also be an option for 3D gridding in Oasis montaj.