Американский Научный Журнал SOIL CLASSIFICATION HAS YET TO BE CREATED (40)

The thesis is put forward that all the classifications of soils existing in the world do not possess the attributes of such a concept accepted in science and its technical applications but are only lists of author's names of soils. The names are substantiated mainly by factors of soil formation (natural zoning, vegetation, series, etc.), on weakly reasoned genetic horizons and small unformalized aggregates of soil properties. At the Soil Institute Named After V.V. Dokuchaev of the Russian Academy of Science, a mathematical apparatus for the development and use of digital soil classifications was created. Скачать в формате PDF
40 American Scientific Journal № ( 37 ) / 2020
СЕЛЬСКОХОЗЯЙСТВЕННЫЕ НАУКИ

SOIL CLASSIFICATION HAS YET TO BE CREATE D

Рожков Вячеслав Александрович
DOI: 10.31618/asj.2707 -9864.2020.1.37.4
Abstract . The thesis is put forward that all the classifications of soils existing in the world do not possess the
attributes of such a concept accepted in science and its technical applications but are only lists of author's names
of soils. The names are substant iated mainly by factors of soil formation (natural zoning, vegetation, series, etc.),
on weakly reasoned genetic horizons and sm all unformalized aggregates of soil properties. At the Soil Institute
Named After V.V. Dokuchaev of the Russian Academy of Scien ce, a mathematical apparatus for the development
and use of digital soil classifications was created.
Key words : classification attributes, existing soil classifications, digital classification.

The most important problems of worl d soil science
are soil classifications, which are considered the
philosophy and language of science. However, the
existing diverse structures describ ing national and
international sets of soils, strictly speaking, are not
classifications. These are nothin g more than lists of
names of soils arranged in a certain order: by natural
zones, vegetation, series, etc. Further research of such
structures is aim ed at finding new soils, additional
indicators, and clarifying the intervals of their values.
The necessar y provisions of the scientific concept of
classifications are not used: indicators of similarity —
soil differences, evaluations of the information co ntent
of attributes, quality criteria and comparison of
classifications, formal rules for recognizing new soils,
etc. Higher school: scientific research. The existing soil
classifications from Sibirtsev to Soil Taxonomy are
lists of soils by natural zones, series, etc. These lists of
soils should be considered as the source material for
creating these classifi cations. They are absolutely
necessary for understanding the real diversity of soil
profiles and taking into account the opinions of
different scienti sts and practitioners, determining soil
characteristics for the subsequent formation of the
required attri butes of formalized classifications. A
meta -algorithm for constructing a formal classification
of soils as an addition and completion of traditional
constructions was developed. The meta -algorithm
software includes software modules that have been
tested ov er many years of use in solving real problems.
A hierarchical or ordinate classification based on
equivalence classes is provided. At the same time,
evaluations of the information content of attributes are
new, calculations of similarity -differences betwee n
objects and classes, a weighted pair -group criterion of
grouping by the principle of “nearest neighbor” are
formalized. For the first time, criteria for the quality of
classifications and their comparison with each other
were used. Recognition of new soi ls is carried out by
discriminant analysis and/or by the values of the
similarity of their descriptions with established objects.
Numerical classifica tion methods (multivariate
statistics and cluster analysis) allow us to solve these
problems. The result o f their application is a “digital”
classification of soils. At the same time, it is a question
of formalizing the creation, analysis and use of
classi fications by introducing mathematical methods,
and traditional lists are the source material for
construct ing classifications.