1/f noise analyses of urbanization effects on streamflow characteristics

Abstract

Nonstationary hydrologic behaviour resulting from rapid industrialization and urbanization, combined with climate change effects, likely produces greater challenges in water resources and flood risk managements. Our country-wide analyses for South Korea, based on spectral analysis technique, revealed how streamflow characteristics have shifted towards a less hydrologic memory state, which indicates a weaker temporal autocorrelation in the time series. Specifically, we analysed 1/f(alpha) noise of streamflow in 78 unit watersheds in five major river basins in South Korea to investigate the effects of urbanization on stream hydrologic responses over a 30-year period. The average slope of runoff spectra, alpha(0), was 0.94 +/- 0.20, indicating that runoffs are characterized by pink noise. The distribution of alpha(0) showed a convergence towards <0.5 with increasing urbanization, indicating a clear effect of memory loss due to expanded impervious surface areas in watersheds. Among the watersheds examined, 59 showed bi-fractal scaling regimes, with scale break points located around 17.5 days. Analysis of the three spectral slopes, alpha(0) (average), alpha(L) (in low-frequency domain), and alpha(H) (in high-frequency domain), revealed a threshold of urbanization ratio (UR) of similar to 5% from which all the three slopes decrease, and additional thresholds of UR around 6-7% are found from which all the three slopes increase as UR increases. While hydrologic responses of watersheds are the result of complex and compound interplay among many factors such as climate and topography, increasing urbanization seems to dominantly control the hydrologic properties resulting in homogenization of spectral slopes among various watersheds. Copyright (C) 2015 John Wiley & Sons, Ltd.