hardRain: An R package for quick, automated rainfall detection in ecoacoustic datasets using a threshold-based approach

Abstract

The increasing demand for cost-efficient biodiversity data at large spatiotemporal scales has led to an increase in the collection of large ecoacoustic datasets. Whilst the ease of collection and storage of audio data has rapidly increased and costs fallen, methods for robust analysis of the data have not developed so quickly. Identification and classification of audio signals to species level is extremely desirable, but reliability can be highly affected by non-target noise, especially rainfall. Despite this demand, there are few easily applicable pre-processing methods available for rainfall detection for conservation practitioners and ecologists. Here, we use threshold values of two simple measures, Power Spectrum Density (amplitude) and Signal-to-Noise Ratio at two frequency bands, to differentiate between the presence and absence of heavy rainfall. We assess the effect of using different threshold values on Accuracy and Specificity. We apply the method to four datasets from both tropical and temperate regions, and find that it has up to 99% accuracy on tropical datasets (e.g. from the Brazilian Amazon), but performs less well in temperate environments. This is likely due to the intensity of rainfall in tropical forests and its falling on dense, broadleaf vegetation amplifying the sound. We show that by choosing between different threshold values, informed trade-offs can be made between Accuracy and Specificity, thus allowing the exclusion of large amounts of audio data containing rainfall in all locations without the loss of data not containing rain. We assess the impact of using different sample sizes of audio data to set threshold values, and find that 200 15s audio files represents an optimal trade-off between effort, accuracy and specificity in most scenarios. This methodology and accompanying R package 'hardRain' is the first automated rainfall detection tool for pre-processing large acoustic datasets without the need for any additional rain gauge data.

Singing streams: Describing freshwater soundscapes with the help of acoustic indices

“…The aim of this study was to investigate the use of acoustic indices without preprocessing acoustic data. In the future, detection and removal of rainfall sounds (Metcalf, Lees, Barlow, Marsden, & Devenish, 2020) could be considered to reduce the influence of external sounds.…”

Acoustic indices perform better when applied at ecologically meaningful time and frequency scales

“…To limit microphone self‐noise, the lowest frequency included in analysis was 300 Hz. We then calculated the mean index value per 10‐min interval of data collected for each acoustic index and each of the 20 TFBs (Figure 1a), having first screened out recording periods containing heavy rainfall ( n = 527) using the hardRain package in r Studio (Metcalf, Lees, et al., 2020, v0.1.1).…”

Automated classification of avian vocal activity using acoustic indices in regional and heterogeneous datasets

“…Acoustic indices could be used to screen audio for abiotic sounds to make this process more efficient (i.e. Metcalf et al., 2020). Finally, if combining data from more than one type of recording unit, we recommend accounting for this using additional predictor variables as systematic differences in technology such as signal‐to‐noise ratio can influence recording quality (Darras et al., 2020).…”

“…This approach has successfully been used to characterize daily diel patterns (Bradfer‐Lawrence et al., 2019; Burivalova et al., 2017; Phillips et al., 2018), seasonal phenology (Buxton et al., 2016; Phillips et al., 2018) and vocal activity patterns (Bradfer‐Lawrence et al., 2020; Oliver et al., 2018) in birds and other acoustically active species. Acoustic data can be screened to identify biotic activity from geophony or anthrophony prior to manual processing to increase efficiency (Metcalf et al., 2020; Sanchez‐Giraldo et al., 2020). Automated classification could also be used to investigate patterns in seasonal phenology at larger scales, using pre‐existing ARU datasets or by integrating multiple datasets as we have done here.…”