USE OF STATISTICAL TESTS TO DESCRIBE THE BASIC DISTRIBUTION PATTERN OF IRON-OXIDE NODULES IN SOIL THIN SECTIONS

Soil formation is often associated with the spatial reorganization of soil components, particularly Fe oxides. However, the methods availabl e to describe the spatial distribution of Fe oxides generally lack mea surable parameters that allow comparison between different types of so ils. In this work, we demonstrate the use of statistical methods (e.g. , chi(2) goodness-of-fit test, mean, variance, and cluster coefficient ) to describe the basic distribution pattern of Fe oxide nodules in so il thin sections. Using a computer, we mapped the spatial distribution (i.e., x and y coordinates) of Fe oxide nodules in six soil thin sect ions from the B horizons of three sandy seas from Alberta, Canada. We calculated the statistical tests at a range of grid sizes (2-10 mm) wi th a Fortran program to (i) examine the effect of grid size on the sta tistical description of the distribution and (ii) determine the ''appr opriate scale of observation'' for each thin section, We defined this stale as the grid size at which the mean number of observation is at l east five, which was found to vary in size from 5 to 9 mm in our sampl es. We recommend the determination of the appropriate stale of observa tion before any statistical description of the Fe nodule distribution. In all but one of our six samples, the statistical distribution of no dules was clustered; the remaining one was uniform in distribution. Th e methods described in this study provide statistical methods whereby results from different studies can be compared.