Optimized Geospatial Tool for Ambient Soundscapes

This proposal addresses the need to develop a robust model for biological, geophysical, and anthropogenic ambient sounds in different environments. The spatial and temporal resolution of the sound level mapping tool will improve mission planning for both rural and urban environments. During Phase-I, a geospatial data-based ambient sound model will be developed and exercised for at least three benchmark CONUS locations, one of which contains significant anthropogenic noise. As part of this effort, geospatial database requirements will be established and new geospatial/acoustic datasets incorporated into existing databases. The soundscape modeling framework will incorporate multiple unsupervised machine-learning approaches, with the novel goal of treating geospatial modeling inputs and different acoustic outputs as part of the same optimization problem. To downselect to Phase-II modeling approaches, the modeling tradespace will be studied for multi-metric predictive capability, efficiency, and robustness. The Phase-I effort will also develop strategies to eventually extend the ambient soundscape modeling to OCONUS locations, including identifying which geospatial data are available and exploring means for acquiring additional geospatial/acoustic data. Eventual transition of the modeling tool to the military and commercial sectors (e.g. the real estate industry) will be aided by Phase-I technology transition planning and business case analyses.