Agriculture Reference
In-Depth Information
WHY CLASSIFICATION?
ClassiÝcation (Smith, 1963) is a system contrived by humans to organize ideas, as they know
them, according to their concepts of order (relationships). There is uncertainty associated with the
facts that are obtained, in the degree of association among the facts, and in the ideas and concepts
whose relationships to each other indicate patterns of order. ClassiÝcation is not a truth that can
be discoveredÐit is only a set of relationships, an abstraction of knowledge, a scientiÝc model of
the subject of interest.
As an abstraction of knowledge, a classiÝcation relies on facts. The facts are data that have
been derived from observations and measurements that are known in terms of the operations by
which they were acquired. As a set of relationships, a classiÝcation relies on the correlations
(relationships) among various facts that are determined empirically, without the necessity of reasons
for them. And as a model, it deals with concepts that describe and explain the elements and
relationships of our current understanding. Insofar as the relationships are believed to be cause-
and-effect ones, and not merely empirical ones, they form the basis of explanations.
We need classiÝcation because there are too many objects (soils) to remember and too many
relationships that may exist (Cline, 1963). We need a way to organize information into patterns
that we can recognize and use more efÝciently and effectively. This means there will be groupings
and separations in such a system. The adequacy of a classiÝcation system is evaluated against its
purpose and not the expectations of the evaluator. If a purpose is implied or assumed but not stated,
then any evaluation becomes somewhat tenuous. There are options; there are choices.
WHAT OBJECTIVES?
We expect a classiÝcation to achieve our objectives, which are conveniently grouped as applied
and scientiÝc objectives.
Applied objectives commonly require direct interpretations. These have been called technical
groupings (Bartelli, 1978), and each one is a type of classiÝcation. Some are based on a single
property, such as stony soils, wet soils, or organic matter content. Others are based on simple
interpretations, such as capability groupings or land-use potentials. There are as many technical
groupings as there are objectives (reasons) for grouping soils. Most soil information systems, such
as USDA-NRCSÔs National Soil Information System (NASIS), contain information and classes of
numerous technical groupings. The general public and users of soil-related information are more
familiar with these objectives, because the groupings can be applied directly to answer speciÝc
questions of interest to them.
ScientiÝc objectives are expectations that concepts of order (in nature) will be revealed and, as
such, a system needs to (a) give identity to otherwise unidentiÝed individuals in the population,
(b) provide identiÝed groups of individuals, and (c) organize classes in such a manner that they
will abstract concepts of order. Most hierarchical systems (Allen and Starr, 1982) provide identity
to the objects of interest, however, they require at least one step of reasoning to reveal either genetic
relationships or applications to practical problems. There are many options and many choices.
THE ARCHITECTURE OF THE SYSTEM
The decade of the 1950s was a remarkable period in the development of soil science. Many
national soil classiÝcation systems were being conceived, and the Food and Agriculture Organiza-
tion of the United Nations was embarking on a major effort to produce the soil map of the world.
National and global soil survey programs required an acceptable classiÝcation system, but soil
databases were sparse. The path of least resistance (for acceptance by peers) was to enhance existing
