L. Decola, SIMULATING AND MAPPING SPATIAL COMPLEXITY USING MULTISCALE TECHNIQUES, International journal of geographical information systems, 8(5), 1994, pp. 411-427
A central problem in spatial analysis is the mapping of data for compl
ex spatial fields using relatively simple data structures, such as tho
se of a conventional GIS. This complexity can be measured using such i
ndices as multi-scale variance, which reflects spatial autocorrelation
, and multi-fractal dimension, which characterizes the values of field
s. These indices are computed for three spatial processes: Gaussian no
ise, a simple mathematical function, and data for a random walk. Fract
al analysis is then used to produce a vegetation map of the central re
gion of California based on a satellite image. This analysis suggests
that real world data lie on a continuum between the simple and the ran
dom, and that a major GIS challenge is the scientific representation a
nd understanding of rapidly changing multi-scale fields.