Page 39 - Modern Spatiotemporal Geostatistics

20 Modern Spatiotemporal Geostatistics — Chapter 1

natural symmetry. In some cases the use of a specific system depends on math-
ematical convenience. The local properties of space and time (e.g., curvature),
as well as rules imposed by the specific natural process (e.g., diffusion) play a
significant role in the choice of the appropriate metric.
Sources of physical knowledge

Yet another argument in favor of the development of modern spatiotempo-
ral geostatistics may be offered. Most people consider geostatistics to be an
applied science. As such, its intent should be to produce marketable products,
capitalizing on the stores of basic knowledge that have accumulated thus far
in a richly productive century. But no one really believes these stores of knowl-
edge consist merely of observational facts. Moreover, it is inconceivable that
cookbook recipes and techniques are capable of taking important basic knowl-
edge into account. The real meaning of "basic knowledge" in the definition
of applied science surely goes far beyond observational facts to include several
other forms of knowledge, such as physical laws, scientific theories, empirical
rules, phenomenological relationships, and factual statements (Lewis, 1983).
In most practical applications, classical geostatistics does not possess the
means which will allow it to account for important sources of scientific knowl-
edge, put together information in new ways, and find creative solutions to
changing conditions (e.g., most of the classical methods are regression-type
data-processing techniques). The quality of decisions generated by various
decision-support systems (DSS; e.g., a water resources DSS or a health man-
agement DSS) depends heavily on the proper use of high-quality physical and
medical knowledge (Foster et al, 1993; Birkin et al, 1996; van Bemmel and
Musen, 1997). As a consequence, the development of modern geostatistics re-
quires methods which are capable of accounting for various sources of scientific
knowledge and weighing all the evidence available in a rigorous and system-
atic manner. Ideally, knowledge bases (including scientific literature and expert
opinions from the specified domain) should be systematically organized so that
they are accessible electronically and interpretable by computer. New meth-
ods should make possible the interdisciplinary cross-fertilization of concepts,
models, and research tools. Modern geostatistics should also recognize the
growing sophistication of research tools, and particularly their ability to reveal
the modes of action of natural phenomena, environmental pollutants, toxi-
cokinetic and biological processes, epidemiologic patterns, etc. Furthermore,
novel methods should include explanation as part of the scientific process and
incorporate global prediction features.
The non-Procrustean spirit

The above arguments should be viewed in the context of constructive criticism
aimed at the improvement and continuing vitality of geostatistics. There is,
however, a possibility that suggestions will be misinterpreted or even rejected
by the Establishment. In fact, the tendency of the Establishment to reject