Available with Spatial Analyst license.
Reclassifying values into intervals or by area groups the input values by dividing the value range into an equal number of specified intervals (equal interval) or by distributing the number of cells into a defined number of groups until each group has the same number of cells (equal area).
To divide the range of values into equal intervals, the Slice tool allows you to reclassify an input raster into intervals or area. For instance, if the range of values in an input raster is 1 to 200, and the number of intervals to slice is 10, the output raster will have 10 values ranging from 1 to 10. Cells with the value 1 to 20 on the input raster will be assigned 1, 21 to 40 will be assigned 2, and so forth.
The following example reclassifies the original values from base raster into 10 equal intervals to assign new reclassified values. The values range from 1 to 20 on the base raster and from 1 to 10 on the output raster.
Slicing a raster into groups with equal areas, as with slicing into equal intervals, requires defining the number of groups into which to slice the output raster. Once the number of groups has been defined, the Slice tool attempts to distribute an equal number of cells in each group based on the count of cells in each zone. The number of values and the number of cells in each zone in the input raster and the specified number of groups will determine how close each output value or grouping is to containing the same number of cells.
The following example reclassifies the original values from base raster into five equal zones, each with the same number of cells (as close as possible).
The reclassification of values into intervals or by area considers all values and their distributions in a raster simultaneously and reclassifies the values into a specific number of groups. In a hypothetical analysis of deer habitat, an input raster to the suitability model may be based on the preference of the deer for locations far from roads. A distance map is created from the existing roads. Instead of individually reclassifying each of the thousands of distance values to a 1-to-10 deer preference scale, the values can be sliced into 10 groups. The group farthest from the roads receives the highest deer preference value, a value of 10, and the group nearest to the roads, a value of 1.
You would reclassify by intervals when a defined number of classes makes sense and the output class values are based on a similar relative scale as the input values. This would be the case if the input data is continuous, because with continuous data, the values are relative to a phenomena or point of reference. Thus, the resulting output classes from the slice reclassification will correspond to the original relative scale of the input values. The distance from roads example above demonstrates continuous data that can be reclassed on a relative scale. Usually, you will not reclassify categorical data (for example, land-use types) by intervals.
You could reclass categorical data by area when it represents similar types of features. The values do not need to be on a relative scale, since the zones will be allocated based on number of cells in each zone, not on their value. For example, the input raster may represent various zones of different types of conifer trees in a managed forest. You may want to divide the landscape into 10 study sites with equal numbers of conifers.