Human and climate variables explain spatio-temporal patterns of streamflow variation across the United States

Understanding how river flow regimes vary across spatial and temporal scales is essential to predicting impacts of human and climate-induced flow alteration on riverine ecosystems. Spectral methods are useful to that end, as they can identify dominant frequencies, amplitudes, and phases in the environment—as well as changes in them. Here we calculated wavelet power at different timescales ranging from 6 hours to 10 years using 209 USGS long-term gages located below dams across the coterminous United States (CONUS). Then, we examined the relationship between streamflow and climate conditions using bivariate wavelet analyses to identify periods of anomalous hydroclimatic conditions. Finally, we explored similarities in time-frequency patterns using hierarchical clustering. This allowed evaluating how dam properties, geographical location, and climate factors explained observed flow variation in the time-frequency domain. Our analysis revealed that spatio-temporal variation in flows are explained by a combination of climate forcing and dam properties (e.g., type, size, and daily peaking vs. baseload operation). These results advance the notion that both climate fluctuations and human activities have been altering flow regimes nationwide—hence their effects need to be assessed simultaneously. An accurate understanding of the relative contributions of each driver, as well as of the potential interactions among them (e.g., droughts worsening the effects of dam operations on downstream flows), may help identify river-reservoir systems that should be prioritized for flow regime restoration.