Impact of decadal and century-scale oscillations on hydroclimate trend analyses

Studies of hydro-meteorological time series have identified decadal and inter-decadal oscillations with quasi-cyclic components as part of long-term natural variations in the data. Concerns over impacts of global warming and climate change have led to many studies of sustainable development and adaptation strategies, involving historic trend estimations in order to forecast future trends. This paper demonstrates the impacts of natural oscillations of quasi-cyclic components on the Mann-Kendal and Thiel-Sen tests, the most common methods used for analyzing data trends. Tests on synthetic and real instrumental hydroclimate records suggest that the Mann-Kendall and Thiel-Sen tests are sensitive to oscillations of quasi-cyclic components. Data record length relative to the periodicity of cycles, magnitude, and phase of the longest quasi-cyclic component are the three most important factors affecting these tests. If the length of the record is greater than three cycle lengths, the impact on the M-K and T-S tests should be minimal. Given the predominance of 45-60 year climate cycles observed in instrumental records, trend analyses of time series records <60 years should be done with caution. These results provide insight for appropriate temporal trend analyses of hydroclimate and associate time series. Crown Copyright (C) 2008 Published by Elsevier B.V. All rights reserved.