Evaluating Streamflow Changes in Continental U.S. Using Wavelet Transformation

Abstract:

This study focused on investigating the long term seasonal trends (gradual change and shifts) in 600 streamflow stations across the continental United States with each station having a continuous streamflow data of at least 30 years. The non-parametric Mann-Kendall test, with appropriate modifications to account for persistence in data, was used to identify the trends whereas the non-parametric Pettitt test was used to identify the shifts. Discrete Wavelet Transformation (DWT) was further applied on a subset of the selected stations (237/600 were selected for DWT) to evaluate the most significant periodicities or recurrence intervals present in the change patterns. The results showed a clear increase in streamflow in the northeast and upper-central regions whereas southeast and northwest regions underwent decrease. The central regions had assorted results while number of stations with decreasing trends was observed to increase from east to west. The shifts were found to be more spatially distributed across the whole study area and followed similar patterns as the trends. The seasons also showed certain patterns in all time-scales under DWT. The presence of persistence was also observed to increase with the increasing time-scales. The results may assist water managers to efficiently plan and manage the water resources under changing climatic conditions across continental United States.