Wavelet-Aided Analysis to Estimate Seasonal Variability and Dominant Periodicities in Temperature, Precipitation, and Streamflow in the Midwestern United States

Abstract

The study explores the conjunction of Discrete Wavelet Transform along with trend and shift detection techniques to analyze variability in seasonal temperature, precipitation, and streamflow across the Midwestern United States. The analysis was performed using three dyadic scales that corresponded to periodicities of two, four, and eight years, referred to as D1, D2, and D3, respectively. The study utilized Mann-Kendall test to analyze trends having variations accounting for serial correlation. Pettit’s test was used to detect shift changes in the hydrologic variables. The results of shift changes also were tested for coincidence with El Nino Southern Oscillation (ENSO) in relation to Pacific Decadal Oscillation (PDO). The temperature and precipitation over 106 climate divisions as well as streamflow over 88 stations were evaluated over the period of 1960–2013. Results indicated an increasing temperature trend, with D2 and D3 being the most effective periodic components in detecting trends in winter, spring, and summer; D1 and D3 were most effective in detecting trends in temperature in fall. Likewise, precipitation and streamflow showed dominance of the D3 component in detecting trends. More shifts than trends were detected in all the hydrologic variables indicating abrupt changes in climate in the region. Temporally, shifts were observed from 1975 to 1995, and spatially shift years varied across the Midwest. Most shift changes coincided with PDO and ENSO phases. The results will aid water managers to better prepare for the future emphasizing the need to make planning and operation more flexible to improve the efficiency of water use.