HR: 0800h
AN: H51E-0803 [Abstracts]
TI: Understanding the structure of step-pool mountain streams through wavelet analysis
AU: * Burge, L M
EM: lburge@okanagan.bc.ca
AF: Department of Geography and Earth and Environmental Science, Okanagan College, 1000
KLO Road, Kelowna, BC V1Y 4X8, Canada
AU: Corbett, N
EM: ncorbett@okanagan.bc.ca
AF: Department of Mathematics, Okanagan College, 1000 KLO Road, Kelowna, BC V1Y 4X8,
Canada
AB:
The periodicity of structures, like step-pool spacing, in the long profiles of mountain streams, has been debated
in the literature. Some authors claim to have established the existence of periodicity in step-pool sequences
through the use of the Fourier transform. Other authors claim that statistical models, based on renewal
processes, adequately explain the placement of the steps. In the latter case, the implication is that step
placement is random rather than periodic.
In fact, the Fourier transform is best suited to the analysis of stationary signals: signals for which the spectral
(frequency or wavelength) content does not evolve through time or space. In the context of long profiles, this
implies that one or more dominant wavelengths persist down the entire reach. Given the complexity of river
systems, it is not unreasonable to assume that spacing of steps varies along the reach. As such, straightforward
Fourier analysis may provide an incomplete or misleading picture of the spectral content of long profiles.
On the other hand, the wavelet transform has been successfully applied to analysis of non-stationary signals.
Unlike the Fourier transform, which supplies spectral information for the entire signal record, the wavelet
transform produces a time-scale or space-scale (i.e. space-wavelength) map that preserves the local spectral
content of the signal.
Consequently, for long profiles, the wavelet transform has the capacity to detect periodic structures, which persist
along the entire reach, as well as "transient" structures that occur at isolated locations along the reach. In this
paper, we investigate the applicability of the continuous wavelet transform for the analysis of long profiles
associated with mountain streams. We begin by demonstrating how the wavelet transform differentiates
prototypical signals that cannot easily be differentiated with the Fourier transform. We then show how the
transform can be applied to reveal the space-wavelength structure of a number of mountain stream long profiles.
DE: 1815 Erosion
DE: 1825 Geomorphology: fluvial (1625)
DE: 1856 River channels (0483, 0744)
DE: 1861 Sedimentation (4863)
SC: Hydrology [H]
MN: 2007 Fall Meeting