Characteristics and environmental responses of birdsong soundscapes in urban green spaces based on passive acoustic monitoring

Abstract

Birdsong soundscapes in urban green spaces can reflect the dynamic relationship between bird activity and site environments. To examine the applicability of passive acoustic monitoring for recording birdsong soundscapes in urban green spaces, this study takes Bailu Park in Suzhou Industrial Park as a case study. Fixed-point recording and synchronized environmental observation were used to collect birdsong audio, temperature, and humidity data. BirdNET, Audacity, and Python were applied to a typical recording period from 09:21 to 10:21 for identification, spectrogram interpretation, and event-level index collation. The extracted indicators included the number of birdsong events, effective singing duration, bird species, confidence score, spectral characteristics, and birdsong density. The results show that the segment contained 115 birdsong events with confidence scores of 0.60 or above, mainly from common urban birds such as the Light-vented Bulbul. The mean dominant frequency was 2417.82 Hz, the mean spectral centroid was 2539.33 Hz, and acoustic energy was concentrated at 2.4-2.5 kHz. Birdsong events were unevenly distributed across time slices, showing short-term clustering and intermittent decline. This study provides a workflow for collecting, identifying, and statistically processing birdsong soundscapes in urban green spaces. It converts perceptual descriptions into recordable and comparable event and spectrogram information, and offers supplementary evidence and references for subsequent green-space evaluation and the creation of low-disturbance environments.