Influential factors and spatiotemporal patterns of environmental sound levels in the contiguous United States

Environmental sound levels represent the cumulative contributions of many types - and possibly an uncountable number - of sound sources. This recommends a statistical approach to modeling. Using 1.5 million hours of acoustical data from hundreds of sites, regression models were built to predict sound levels across the contiguous United States. These models discern often nonlinear and interacting relationships between measured sound levels and nonacoustic environmental summaries extracted from nationwide geospatial data layers. Tens of potential explanatory factors were examined including climate, topography, human activity, and time of day and year. Mapping sound levels at landscape scales and diagnostic tools, like partial dependence functions, can reveal the effects of influential factors on measured sound levels. These results illustrate the foundations of many spatiotemporal patterns, provide tools for understanding current acoustical conditions and demonstrate the potential consequences of shifts in environmental conditions.