We investigate the possible presence of ‘long time’ memory in the auto-correlations of biophonic activity of environment sound. The study is based on recordings taken at two sites located in the Parco Nord of Milan (Italy), characterized by a wooded land, rich in biodiversity and exposed to different sources and degrees of anthropogenic disturbances. The audio files correspond to a three-day recording campaign (1-min recording followed by 5-min pause), from (17:00) April 30 to (17:00) May 3, 2019, which have been transformed into ecoacoustic indices time series. The following eight indices have been computed: Acoustic Complexity Index (ACI), Acoustic Diversity Index (ADI), Acoustic Evenness Index (AEI), Bio-acoustic Index (BI), Acoustic Entropy Index (H), Acoustic Richness index (AR), Normalized Difference Soundscape Index (NSDI) and Dynamic Spectral Centroid (DSC). We have grouped the indices carrying similar sound information by performing a principal component analysis (PCA). This allows us to reduce the number of variables from eight to three by retaining a large (≳80%) variance of the original variables. The time series corresponding to the reduced set of new variables have been analyzed, and both seasonal and possible long term trend components have been extracted. We find that no trends are present, i.e. the resulting time series are stationary, and the auto-correlations of the three selected PCA dimensions and associated residuals (obtained after extracting the seasonal components) can be determined. The calculations reveal the presence of a “memory” of few (≲5) hours long in the environment sound, for the two sites considered, which is quantified by the Hurst exponent, H. For Site 1, we find an overall effective Hurst exponent, Hdim≃0.88, for all three dimensions, and Hres≃0.75 for the residuals. For Site 2, the exponents are slightly smaller, amounting to 0.80 and 0.60, respectively. We attempt to correlate the Hurst exponents with a quality index obtained from an aural survey, aimed at determining the sound components, such as biophonies, technophonies and geophonies, at the two sites. We conclude that the higher the Hurst exponents, the higher are the periodic-structured sounds, corresponding to stronger long-term biophonic activity. We find that Site 1 has a more structured environment sound than Site 2, also consistent with the major presence of tall trees surrounding the location of the acoustic sensor at the former.
Benocci, R., Roman, H., Bisceglie, A., Angelini, F., Brambilla, G., Zambon, G. (2022). Auto-correlations and long time memory of environment sound: The case of an Urban Park in the city of Milan (Italy). ECOLOGICAL INDICATORS, 134(January 2022) [10.1016/j.ecolind.2021.108492].
Auto-correlations and long time memory of environment sound: The case of an Urban Park in the city of Milan (Italy)
Benocci R.
;Roman H. E.;Bisceglie A.;Angelini F.;Zambon G.
2022
Abstract
We investigate the possible presence of ‘long time’ memory in the auto-correlations of biophonic activity of environment sound. The study is based on recordings taken at two sites located in the Parco Nord of Milan (Italy), characterized by a wooded land, rich in biodiversity and exposed to different sources and degrees of anthropogenic disturbances. The audio files correspond to a three-day recording campaign (1-min recording followed by 5-min pause), from (17:00) April 30 to (17:00) May 3, 2019, which have been transformed into ecoacoustic indices time series. The following eight indices have been computed: Acoustic Complexity Index (ACI), Acoustic Diversity Index (ADI), Acoustic Evenness Index (AEI), Bio-acoustic Index (BI), Acoustic Entropy Index (H), Acoustic Richness index (AR), Normalized Difference Soundscape Index (NSDI) and Dynamic Spectral Centroid (DSC). We have grouped the indices carrying similar sound information by performing a principal component analysis (PCA). This allows us to reduce the number of variables from eight to three by retaining a large (≳80%) variance of the original variables. The time series corresponding to the reduced set of new variables have been analyzed, and both seasonal and possible long term trend components have been extracted. We find that no trends are present, i.e. the resulting time series are stationary, and the auto-correlations of the three selected PCA dimensions and associated residuals (obtained after extracting the seasonal components) can be determined. The calculations reveal the presence of a “memory” of few (≲5) hours long in the environment sound, for the two sites considered, which is quantified by the Hurst exponent, H. For Site 1, we find an overall effective Hurst exponent, Hdim≃0.88, for all three dimensions, and Hres≃0.75 for the residuals. For Site 2, the exponents are slightly smaller, amounting to 0.80 and 0.60, respectively. We attempt to correlate the Hurst exponents with a quality index obtained from an aural survey, aimed at determining the sound components, such as biophonies, technophonies and geophonies, at the two sites. We conclude that the higher the Hurst exponents, the higher are the periodic-structured sounds, corresponding to stronger long-term biophonic activity. We find that Site 1 has a more structured environment sound than Site 2, also consistent with the major presence of tall trees surrounding the location of the acoustic sensor at the former.
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
