Stream ̄ow characterization and feature detection using a discrete wavelet transform

An exploration of the wavelet transform as applied to daily river discharge records demonstrates its strong potential for quantifying stream ̄ow variability. Both periodic and non-periodic features are detected equally, and their locations in time preserved. Wavelet scalograms often reveal structures that are obscure in raw discharge data. Integration of transform magnitude vectors over time yields wavelet spectra that re ̄ect the characteristic time-scales of a river's ̄ow, which in turn are controlled by the hydroclimatic regime. For example, snowmelt rivers in Colorado possess maximum wavelet spectral energy at time-scales on the order of 4 months owing to sustained high summer ̄ows; Hawaiian streams display high energies at time-scales of a few days, re ̄ecting the domination of brief rainstorm events. Wavelet spectral analyses of daily discharge records for 91 rivers in the US and on tropical islands indicate that this is a simple and robust way to characterize stream ̄ow variability. Wavelet spectral shape is controlled by the distribution of event time-scales, which in turn re ̄ects the timing, variability and often the mechanism of water delivery to the river. Five hydroclimatic regions, listed here in order of decreasing seasonality and increasing pulsatory nature, are described from the wavelet spectral analysis: (a) western snowmelt, (b) north-eastern snowmelt, (c) mid-central humid, (d) southwestern arid and (e) `rainstorm island'. Spectral shape is qualitatively diagnostic for three of these regions. While more work is needed to establish the use of wavelets for hydrograph analysis, our results suggest that river ̄ows may be e€ectively classi®ed into distinct hydroclimatic categories using this approach. # 1998 John