Agriculture Reference
In-Depth Information
such as the Australian Soil ClassiÝcation, will continue to evolve and have a place in the develop-
ment of landscape models with spatial soil attributes.
Innovative methods for broad-scale mapping of land resources have been developed during
the last decade using digital terrain analysis, geophysical remote sensing, direct Ýeld measurement
and pedological knowledge (e.g., Cook et al., 1996; Fitzpatrick et al., 1999; McKenzie and Ryan,
1999; Davies et al., 2001; Merry et al., 2001). The new methods overcome many of the problems
of conventional survey by providing predictions of individual soil and land properties (i.e., predict
Ýrst, classify later). Substantial improvements to the methods are possible, and these will be
necessary before they can be generalized to the broad range of landscapes occupying the
productive zones of Australia. A critical requirement is the development of more process-based
models of soil and regolith distribution. Current methods rely on statistical correlations between
soil properties and various attributes computed from terrain models and related sources. While
these are always assessed for their process signiÝcance, it would be much better to have a set
of predictor variables and models that reÞect processes active in landscape evolution at scales
that can be applied across regions.
Powerful PC-based geographic information systems and web technology have created much
better ways for communicating the results of landscape analysis for a range of purposes. Effective
systems have been developed that provide simple access to these:
¤
Primary survey data
¤
Digital images of soil proÝles and sites
¤
Summaries of laboratory data
¤
Spatial predictions of individual soil properties and land qualities
¤
High resolution digital surfaces of environmental variables
¤
Access to publications
A logical extension would be the integration of simulation modelling outputs. It is within this
environment, driven by information technology, remote sensing, and conceptual models of land-
scape development, that Ýeld-based and practical soil classiÝcation schemes must operate if they
are to be relevant.
SUMMARY
Soil classiÝcation systems in Australia are brieÞy reviewed. The recent Australian Soil
ClassiÝcation is a signiÝcant improvement over older systems, but it remains weak for soils from
non-agricultural areas where few data exist, such as for acid sulfate, saline, and hydromorphic
soils. Transfer of technology has been traditionally advocated and achieved by pedologists relying
heavily on morphologically deÝned soil typesÐthe morphological criteria have pedological
signiÝcance, but their relevance to practical land management is less clear. The rationale is not
necessarily true that general-purpose soil classiÝcation systems stratify the soil population in
ways that are relevant to agronomic and engineering applications, and in sufÝcient detail to
facilitate information transfer.
Transfer of soil information has become less pedocentric and is tailored more to suit client
needs. This has involved improvements in the communication of general-purpose classiÝcation
systems and the development of several special-purpose schemes tailored to particular problems
(i.e., infocentric). Case studies are presented to demonstrate innovations incorporating these phi-
losophies. They include improved communication of general-purpose classiÝcation systems through
the development of a CD-based interactive key, and user-friendly soil classiÝcations at state or
regional levels in Western Australia (Soil Supergroup Groups), South Australia (Soil Groups and
Classes), and NSW (Broad Soil Groups and Special Soil Groups) to assist with the communication
