Structure is an important physical feature that affects many other aspects of the soil. Accurate maps of top- and sub-soil structure are desirable for a wide range of models that aim to predict erosion, solute transport, or flow of water through the soil. Such maps would also be useful for land management to determine where to apply nutrients and pesticides, to target sub-soiling, and to stabilize soil structure.

Soil structure in the field is typically inferred from bulk density or penetrometer measurements. Recently, soil resistivity and conductivity have also been used, but these methods are affected by many spatially varying soil properties; in particular the moisture content at the time of sampling, soil texture, and stones. We suggest a return to traditional methods for observing soil structure and using geostatistics to map these observations. Indicator kriging has been used to map nominal data such as water table class therefore it seems an appropriate approach for mapping soil structure.

Observations of top- and sub-soil structure were made at four sites using the descriptive system of Hodgson (1974), which takes into account the degree of ped development and type of ped. The observations were then ranked in three different ways (Hodgson, 1976; Peerlkamp, 1967 and Kerry, 2003) for soils of different texture.

Geostatistical maps were produced for each site by indicator kriging. These show the probability of encountering soil with the structure represented by a particular rank and could be incorporated into soil models.

Observations are straightforward to make for a single observer and the broad textural grouping could be assessed at the same time by hand texturing. The observations are not affected by soil moisture or stones. A method for choosing between ranking methods and an economical protocol for mapping will be presented.