Page 183 - Modern Spatiotemporal Geostatistics
P. 183
164 Modern Spatiotemporal Geostatistics — Chapter 8
Doing Progressive Guesswork
Unlike Arthur Golden's novel, Memoirs of a Geisha, which describes a world
in which appearances are paramount (Golden, 1997), in the geostatistical world
one seeks to go beyond appearances, adding scientific substance to the maps
obtained. This can happen if not only the physical knowledge bases are in-
tegrated into space/time analysis, but the uncertainties associated with these
bases are adequately assessed and visualized, as well.
The preceding measures of mapping accuracy (or, if you prefer, measures
of mapping uncertainty) are most valuable in the vast majority of applications
in geostatistical practice in which the phenomenon being studied cannot be
isolated from a host of confounding influences that introduce a chance com-
ponent of considerable size into the quantitative analysis. In many situations,
while the amount of uncertainty estimated solely on the basis of the hard data
may initially be substantial, there is usually a significant amount of physical
knowledge currently being ignored that could improve considerably the mapping
accuracy of the phenomenon.
BME analysis establishes a progressive process, during which the maps
produced using the new knowledge show a substantial improvement in accu-
racy over the maps previously obtained. As the process tends towards the
"ultimate" map, the application-specific goals of map making allow a specified
level of uncertainty relative to the actual (but unknown) map. Assessing this
uncertainty is the important task of the quantitative measures discussed in this
chapter.