Dataset: Acoustic backscatter data collected from RVIB Nathaniel B. Palmer (NBP0103, NBP0104, NBP0202, NBP0204) in the Southern Ocean from 2001-2002 (SOGLOBEC project, Southern Ocean Krill project)

Each matlab file includes a number of matrices. The data are in sv_l, sv_2, sv_3, sv_4, and sv_5, where the suffix indicates frequency: 1=43 kHz, 2=120 kHz, 3=200 kHz, 4=420 kHz, 5=1000 kHz. The other matrices and vectors (bot, pings_noise, pings_noise_art, top) are things the PI used to keep track of where bad or otherwise unwanted data were located (e.g., all cells greater than those found in bot are echoes from below the bottom). These other matrices are not served here but can be made available.

The matrix structure for sv_# is consistent across all files. Each matrix has some number of rows corresponding to the number of acoustic observation points, or 'pings,' and then 407 columns. The first 7 columns give header information, and then the next 8-407 give the acoustic observations at increasing depth at the location of each ping. For each ping, the column structure is:
Col 1 = yearday time. For NBPOI03, this is GMT, for all other cruises it's local time.
Col 2 = latitude
Col 3 = longitude
Col 4 = frequency in kilohertz (kHz)
Col 5 = start depth (m), almost always 1m.
Col 6 = For most cruises, this is the end depth (m). For NBP0103, it gives the local time.
Col 7 = Vertical bin size (m). This is the vertical distance represented by each acoustic measurement. At 43 and 120 kHz, it's l.5m, and at the other frequencies it's 1m.

Then columns 8 through 407 give the acoustic measurements. Col 8 is the measurement made from the start depth to the start depth + bin size, Col 9 gives the next bin size, and so on. So at 43 kHz, for instance, coI8=1-2.5m, coI9=2.5-4m, etc. There is in fact a bit of squishiness associated with the start depth, so the first data column can be taken as 0-1.5 or 1-2.5.

These matrices are created by taking the data collected by the up-and down-looking transducers at each frequency, finding the position in the matrix corresponding to the position of the BIOMAPER (which is of course being towed up and down in the water column) and then filling in the matrix with the up-and down-data. The cell corresponding to the position of the BIOMAPER is set to NaN, as are a number of adjacent cells -the number varies with frequency, and corresponds to the acoustic near-field where we can't make reliable measurements. Cells corresponding to depths beyond the range of the acoustic system (300m at 43 and 120 kHz, 150m at 200kHz, 100m at 420kHz, and I think 35m at 1000kHz) are also set to NaN.

The remaining non-NaN cells give the acoustic measurements, which are unit-less volume backscattering coefficients. Here is a definition of the volume backscattering coefficient, from the manuscript submitted to DSR-II:
Volume scattering strength, or S_v (where S_v = 10log₁₀(s_v) in units of decibels relative to 1m^-1, and s_v is the volume scattering coefficient), is a measure of the intensity of emitted sound that is scattered towards the acoustic receiver per cubic meter. When the source and receiver are co-located, the direction of scattering is back towards the source, and this quantity is commonly referred to as the volume backscattering strength. Target strength is a measure of the intensity of scattering received in the direction back towards the acoustic source from an individual ensonified target (TS = 10log_l0(σ_bs) in units of decibels relative to 1m2, where σ_bs is the differential backscattering cross-section). Under the assumption made in zooplankton acoustics that scattering from individual targets in the ensonified volume sums incoherently, the volume backscattering coefficient is equal to the sum of the backscattering cross-sections of all targets present, normalized by the acoustic sample volume. Typically, acoustic data are displayed and analysed in logarithmic form (ie the volume backscattering strength), which as you can see from the above defInition, is given as 10 times the log_l0 of the volume backscattering coefficient values found in the sv_# matrices. Zeroes correspond to cells where the measurements were below the limit to detection or below the noise threshold.

These served files include the acoustic data that have now been cleaned up such that cells representing unwanted signal have been set to NaN -this includes noise, echoes from below the bottom, and the surface bubble layer. These cleaned up data are again found in matrices named sv_#. The data at 120 kHz (sv_2) have been cleaned up thoroughly. The data at the other frequencies have had the bottom and surface echoes removed, but then otherwise have only been cleaned in those portions of the water column where there were krill. There was no cleaning of the 1000 kHz data and so there is no sv_5 matrix in this file.