Russian nesting dolls effect – Using wavelet analysis to reveal non‐stationary and nested stationary signals in water yield from catchments on a northern forested landscape

Abstract

AbstractDetermining catchment responses to climate signals gives insight into the potential effects of climate change. This study tested the hypothesis that a 28‐year time series of water yields from four headwater catchments in the Turkey Lakes Watershed (TLW), Ontario contains signals of non‐stationary climate change and naturally occurring stationary climate oscillations and that the effects of these signals are greater in catchments with lower rates of change in water loading and lower water storage capacity (small wetlands). Non‐stationary trends explained 0%, 18%, 44%, and 52% of the variance in the water yields of the four catchments. Wavelet analysis using Morlet wavelets identified stationary responses at multiple temporal scales, increasing the amount of variance explained to 56%, 63%, 76%, and 81% when combining stationary and non‐stationary models. The catchment with low water loading and low water storage was most sensitive to non‐stationary and stationary signals, suggesting that these catchments act as sentinels to detect climatic signals. Wavelet coherence analysis revealed correlations between global climate oscillation indices and water yield. The Atlantic Multidecadal Oscillation (AMO) index was strongly correlated with both temperature and precipitation (R2 = 0.46, P < 0.001 and R2 of 0.34, P < 0.001, respectively). Temperature in the TLW increased by 0.067 °C per year from 1981 to 2008, but approximately 0.037 °C of this increase can be explained by the AMO index. While it is likely that anthropogenic climate change impacts water yields, it is important to account for multiple nested climate oscillations to avoid exaggerating its effects. Copyright © 2012 John Wiley & Sons, Ltd.