4. Effective Multivariate Description of Soil and Its Environment

When we wish to characterize soil, it soon becomes very clear that one or two properties of soil materials, horizons, profiles or pedons will not suffice to give an adequate description. Soil classification, land capability, soil quality, condition and health assessments often involve the observation of tens or scores of soil properties on a single soil entity; e.g. the new soil microbial DNA descriptions involve hundreds or thousands of attributes. For analysis of such high-dimensional data, multivariate statistical techniques are most appropriate, particularly ordination techniques which help to reduce the dimensionality down to a few (typically 2 or 3) which can be graphed simply and the relationships between soil entities, and between observed soil attributes on those entities, displayed.