Abstract
A procedure for sampling and measuring meander geometry is presented. The sampling procedure, developed using a mathematically defined meander ensemble with known statistical properties, minimizes operator bias and sampling noise inherent in the digital data collection process. The suite of twenty-eight morphometric variables describes local, intermdediate-sca 1e, and ensemble features of meander geometry. Many of these variables are introduced for the first time in this paper. Sample length (expressed in terms of the number of meander loops), as well as the data collection process, influences statistical results to a greater extent than has been previously acknowledged. Samples that include fewer than 50 whole-loops (a whole-loop is the channel path between alternate inflection points) are shown to yield unreliable parameter estimates. A data base of thirty-three rivers from the United States representing a wide range of meander forms is developed to examine the variability of the morphometric variables.
Frequency of meandering of each sample river is examined using spectral analysis of the curvature series. The results of the spectral analysis suggests that there are four general classes of meandering based on the number of dominant peaks in the curvature spectrum. The classes represent unimodal, bimodal, trimodal, and f1 at/multimoda1 meander patterns. Dimensionality of meander geometry is V described using principal components analysis. The first five principal components (dimensions) account for 90.80f the total variance. These components are. in order of prominence, related to half-loop path length, half-loop sinuosity, loop asymmetry, large-scale wandering of the channel, and the spread of the dominant spectral peak.
The sampling procedure and descriptive variables can be used to study fluvial process-response mechanisms using correlations between environmental controls and channel form. Theoretical models of meandering can be tested by comparing statistics for simulated patterns to those derived from the data base of natural rivers.
Because width-normalized variables dominate the first five principal components, it appears that other scaling indices should be developed. Self-scaling using a more characteristic measure of the channel pattern, such as average half-loop path length or dominant wavelength, may provide a more reliable index upon which the sample rivers may be normalized. Tests of sample stationarity must also be developed and applied to meander geometry data.
Note: Abstract extracted from PDF file via OCR.
