When should the discrete values data type be used for an image?
You should use the discrete values data type only when your image values are not uniformly distributed. This usually occurs with scanned maps. For instance, let’s say that you have 4 values: 0, 1, 2, and 191. If this image band was used as a reflectance band and the values stretched, then the values that are very close together (0, 1, and 2) would be lumped into one category and 191 would be lumped into another category. Instead, if discrete values were used, we would get four separate categories: (a) 0 –> 0; (b) 1 –> 0.33; (c) 2 –> 0.67; and (d) 191 –> 1. Note: You cannot use discrete values with floating point (decimal values) raster images. Instead, you can only use discrete values with whole numbers, such as a 0-255 range of values.
