The Color of Environmental Noise Across the United States’ Rivers: Spatial Patterns and Effects of Global Change

Abstract:

The degree of temporal autocorrelation in the environment, also known as ‘environmental noise color’, has far-reaching implications for conservation and management of natural resources. In some rivers, aseasonal variation in flows (i.e. ‘unexpected’ floods or droughts) may challenge the persistence of riverine organisms. However, our understanding of flow noise color patterns, and their drivers, remains poor. Here we analyzed the spectrum of frequencies in noise, using long-term mean daily discharge records from 7,504 USGS streamflow gauges with at least 15 years of mean daily discharge data. These gages encompass streams and rivers across the conterminous United States, encompassing a wide range of geographic and hydroclimatic contexts. We removed seasonal variability, and calculated spectral densities for each time series. We then described: (i) large-scale spatial patterns in flow noise color, (ii) the influence of several static and time-varying covariates (including hydroclimate, land use change, and regulation by dams), and (iii) potential consequences of changes in noise color for population persistence, via simulations. We found high spatial variation in daily flow noise color, which varies from white to black. Random forest models allowed for exploring the drivers of historical patterns and changes in flow noise color, and for predicting flow noise color within and across river networks. We found marked spatial patterns in flow noise color, largely driven by catchment drainage area, temperature, and wetland coverage. Flow regulation also showed a relatively strong influence, underscoring the importance of human-driven impacts to the flow regime. This research shows that spectral analyses can improve our understanding of the nature and drivers of flow regime change. Because streamflow is a ‘master variable’ controlling river ecosystem structure and functioning, changes in the contributions of deterministic and random variability in streamflow need to be further considered in hydrologic alteration assessments.