Sometimes the value of the measurement position has a directional effect that can be quantified statistically but may not be explained by a known identifiable process. This directional effect is called anisotropy. Anisotropy is used to analyze whether the sample shows a different range in different directions. The tolerance angle is used to determine at which angle the proximity point is included or excluded until the bandwidth is reached. Bandwidth is used to specify the search width that should be used when determining which pairs of points will be plotted in the semivariogram.
The points in a bin are position pairs that are located in a particular distance and direction. Conceptually, you can think of a direction grouping as a restriction on the point pairs that will be drawn during the grouping process, or as drawing all the point pairs and considering only the part of the graphic that represents a particular direction. The following scenario depicts a 90-degree direction grouping with a bandwidth of 5 meters, an angular tolerance of 45 degrees, and a step distance of 5 meters from a single sample point (blue).
The direction search continues for each sample point on the surface. The following scenario shows the direction grouping of three points.
Then, the point pairs are grouped according to common distances and orientations, the bars are averaged, and the point-to-mean averages of each bar are plotted on the Semivariogram graph. If the cell center on the semivariogram surface is contained within the search direction, a bar is drawn on the semivariogram.
ArcGIS Tutorial: Empirical Semivariogram in different directions