| 1 |
|
|
Data analysis was performed using the Making Ambient Noise Trends Accessible (MANTA,
https://bitbucket.org/CLO-BRP/manta-wiki/wiki/Home, see Miksis-Olds et al., 2021;
Martin et al., 2021a,b) standalone software (Version 9.6.12) to produce hybrid millidecade
spectra of sound levels from ocean audio recordings. To efficiently tackle large datasets,
MANTA is designed around a parallel-processing Matlab package, Raven-X (Dugan et.
al., 2014, 2016, and 2018) that uses ordinary multi-core computers to accelerate processing
speeds. MANTA calculates the sound pressure spectral density (PSD) levels in units
of 1 µPa^2/Hz using Welch's Method in Matlab. The Discrete Fourier Transform length
is equal to the sample rate, a Hann window of equal length is applied to the data
and 50% overlap is used. This results in PSD estimates of mean-square pressure amplitude
(µPa^2) with a frequency resolution of 1 Hz and temporal resolution of 1 second. The
120 PSD estimates from each 1-minute segment were averaged, and the average spectrum
for each minute was further processed to a hybrid millidecade (HMD) spectrum as dB
re 1 µPa^2/Hz, as defined in Martin et al. (2021b). Hybrid millidecades are an efficient
means of storing PSD spectra from high sample rate audio files using 1-Hz values up
to 435 Hz, then millidecade wide PSD values up to one half of the sampling rate (Martin
et al., 2021b). The MANTA outputs for each day are: (1) CSV of the 1 minute HMD results;
(2) image of the daily long-term spectral average based on the 1 minute HMD results,
(3) image of the daily spectral probability density with percentiles, and (4) NetCDF
containing products 2 and 3 in addition to a deployment-level MANTA Metadata output
file containing the associated frequency-dependent calibration data used to compute
the calibrated spectrum levels.
|