Studying Environmental Sounds using Transfer Entropy: Application to an Urban Park in the City of Milan, Italy

Sounds in large cities are usually characterized by multi-source overlaying components. This makes an environment sound quality difficult to interpret, thus requiring different investigation approaches. This is the case, for instance, of urban parks where the coexistence of anthropogenic and biophonic sound sources makes the evaluation of the local soundscape quality challenging. In this context, we applied the transfer entropy measure to extract additional information with respect to the more classical approach of eco-acoustics. The case study refers to the Parco Nord of Milan, Italy, for which classical soundscape analysis is already available. We used sound recordings (3.5h long each) taken over an area of approximately 20 hectares using 16 sensors. We calculated the classical eco-acoustic indices and studied the transfer entropy (TE) among all sites in both directions, i.e. from one site to another and viceversa. It is found that the TEs are not symmetric. The considered time series correspond to three representative eco-acoustic indices: Acoustic diversity index (ADI), acoustic complexity index (ACI) and dynamic spectral centroid (DSC), calculated for all the 16 sites with 1-minute time resolution, and obtained after a principal components analysis. A preliminary interpretation of the results is provided by selecting a representative index from one principal component and by comparing its behavior with a previous standard soundscape analysis performed by the authors for the same index and area.