At each survey site, a fine-scale acoustic grid was conducted at a speed of 8 kn. Survey lines were adjusted for the direction of the sea state. At a few sites, the survey grid was run multiple times, either during the day and then the night, or separated by several days or weeks.
A total of thirteen fine scale acoustic surveys (FSAS) were
conducted on the AR040 cruise. Each FSAS was composed of 3 to 9 approximately
10km long transects, running across an approximately 10km2 area
centered on the lander at seven of the ADEON sites. All FSAS data from the
AR040 cruise have been processed using Echoview software, as described below.
Data for all frequencies were collected using an EK80
echosounder system. For each site (VAC, HAT, etc.), water column sound speed,
temperature, and salinity were edited to reflect the mean midwater
environmental parameters derived from conductivity, temperature, and depth
(CTD) profile data.
Transducer depth was set to 3.96 m (13 feet) below the
surface, the depth of the R/V Armstrong’s hull. A surface exclusion line was
placed at 5 to 10m depth for all files and subsequently adjusted to ensure any
backscatter from surface bubble intrusion were excluded. For all files,
bottom-detection lines were automatically generated using the lowest-frequency
data (18 kHz). These bottom lines were visually examined and edited for errors
and then applied to data from all other frequencies. A bottom offset exclusion
line was generated at one meter above the bottom line, to help ensure no
backscatter from the seafloor was included in the water column data.
The removal of ambient, background, and self-noise was a
multistep process conducted in Echoview. Once areas above the surface line and
below the bottom line were excluded, time-varied gain (TVG) noise was removed
using the data generator operand to virtually generate TVG noise stripes for
each frequency based on the SV value at 1 m. The SV value
at 1 m was determined by scrutiny of the TVG evident on passive noise files,
with adjustments based on the data files. For the AR040 data, the SV
values at 1 m were: -130 dB for 18kHz data, -138 dB for 38 kHz, -132 for 70
kHz, -125 dB for 120 kHz, and -121 dB for 200 kHz. This variable allowed much
of the TVG noise to be subtracted from the data via a linear minus and a
transient noise removal operator. Once the noise spikes were removed, a
background noise removal algorithm was applied with a maximum noise threshold
of -125 dB and a minimum SNR of 10, as described in De Robertis
& Higginbottom (2007). A final 5x5 median filter was applied to remove
any remaining background noise sources. Before removing noise for each wideband
frequency, the data were converted to base Sv using a type conversion operator,
so that the background noise removal algorithms could be applied.
In some files, the above background noise removal process was
unable to remove certain sources of “bad data,” such as engine noise. In such
cases, the noise was removed manually by defining regions of “bad data” that
were then excluded from export.
The acoustic data were binned into cells of 100 m
horizontally and 5 m vertically for final exportation to .csv files.
This dataset has a readme file.
See https://adeon.unh.edu/cruise for more information about the project