Soil scientists sometimes are not able to give optimality values, but they know where the typical location is (a typical location is the place where soil is almost the typical case of a certain soil type). The value of environmental variable in the typical location can be viewed as the value that has the highest optimality value for the soil type. By providing typical locations, soil scientists describe their knowledge implicitly. Point rule is designed to support this kind of knowledge. The point was referred to as tacit point in earlier versions of SoLIM software.
Point rule is very similar to range rule. The basic function underlying them are the same:
f(x): the optimality value
x: the environmental variable value at the location to be inferred
b: the environmental variable value at which the optimality for the soil type is the highest (the most ideal)
d: the difference between b and the environmental variable value at which the optimality value is 0.5 (soil is less typical of the specific soil type)
The only difference is how b is determined: b of an range rule is directly specified by the user, whereas b of a point rule are identified by the inference engine according to the coordinate of the point specified.
There are no low unity, high unity, low cross, high cross for a point rule. Instead, it has four parameters: central X, central Y, left width and right width.
Central X and Central Y:
The coordinates of the point where the soil is the most typical. They will decide the value of b. It must be noted that in a point rule, b can only be a single value.
Left Width and Right Width:
Left width is the parameter d in the function when x<b, whereas right width is the parameter b when x > b.
Like range rule, point rule also has three types of curves.
| • | bell-shape curve: For a bell-shaped curve, optimality values decrease away from the idea condition. If the left width and right are the same, the curve is symmetric. Otherwise, it is unsymmetrical. |
| • | S-shaped curve: S-shape curve can be created by setting the optimality value to 1 for x greater than b. It is a "the-higher-the-better" shape. Right width is not needed for S-shape curve. |
| • | Z-shaped curve: Z-shape curve can be created by setting the optimality value to 1 for x less than b. It is a "the-lower-the-better" shape. Right width in not needed for Z-shape curve. |
Point rule records the x-y coordinates (spatial location) of the typical point, and the left width and right width. . Its optimality value is determined by the x-y coordinates rather than the direct input.