Looking for Evidence of Climatic Change in Streamflow Time Seriesby K. H. Hamed,
A. R. Rao,
Abstract: The principle of stationarity of a streamflow time series has been very commonly assumed in the analysis of hydrologic time series. Some common types of non-stationarity in time series, such as seasonal cycles or linear trends, are readily identified and removed. Other types of non-stationarity cannot be easily detected. The identification of nonstationary behavior of time series can proceed in the time domain and in the frequency domain. Methods in the time domain range from simple visual inspection to parametric and non-parametric testing. Recently, spectral methods that possess better properties regarding resolution and power leakage, such as the Maximum Entropy Method (MEM) and the Multi Taper Method (MTM), have been introduced. These methods give more reliable results for short time series than conventional methods. An interesting approach which combines time and frequency domains is Evolutionary Spectral Analysis (ESA). In the present study, six of the longest mean annual streamflow series in the United States are investigated for evidence of change in climatic conditions, and for signs of nonstationary behavior in general. Different Statistical tests as well as MTM and ESA are used. The results indicate the series to be nonstationary.
Subject Headings: Time series analysis | Climate change | Streamflow | Entropy methods | Parameters (statistics) | Hydrology | Linear functions | Inspection | Seasonal variations | North America | United States
Services: Buy this book/Buy this article
Return to search