Processed EK60 Data Documentation
The methods outlined below are published in Wall et al. (2016). The concept for the multiple frequency single beam imagery is adapted from Jech and Michaels (2006).
All files are binned by hour based on the start time of the file and then preprocessed by first aligning pings in the time/distance domain across the frequency components. Data are then binned vertically to 1,000 data points between 0 and 750 m (i.e., Sv at 0.75 m intervals). Background noise is removed following De Robertis and Higgenbottom (2007) where the signal to noise ratio is set to 10 dB. Similar to the methods in Ryan et al. (2015), impulsive noise "spikes" are removed using a two-sided comparison method where 1 ping on either side of the current ping is removed if the Sv is 10 dB higher or lower than the adjacent pings. Each excised data point is replaced with the local mean Sv calculated using the 7 Sv values at the same range (i.e., same row of the echogram). To smooth and reduce stochastic variability in the data, a 3x3 convolution filter with the maximum weight on the center pixel and summed kernel weights equal to one was then applied.
The depth of the seafloor is estimated using Echoview's best bottom candidate algorithm. The minimum depth variable is set for each file based on the minimum extracted bottom depths from the GEBCO global 30-arc second interval bathymetric grid for that transect. The GEBCO minimum depth is multiplied by 0.3 (30% of the minimum depth) to provide the bottom detection algorithm with a conservative minimum depth value to account for varying resolutions between the GEBCO grid and the sonar data. Data within 1 m of and below the estimated bottom depth are removed.
The Simrad EK500 color scale is applied to the pre-processed data with seafloor removed. The data are scaled to -70 dB to -34 dB. The color bar is visible in the images. To date, these imagery encompass the 18 kHz EK60 data collected on the OER's Okeanos Explorer.
The MFSBI algorithm enables multiple frequency data to be illustrated in a single image by depicting the dominating frequency or frequencies. A threshold of -66 dB is applied to the pre-processed Sv data which was empirically established through inspection and serves to remove low amplitude backscatter (Jech and Michaels, 2006). The Sv echograms are then transformed to a matching array of unique values where Sv greater than -66 dB are assigned a unique, positive value based on the acoustic frequency while data below the Sv threshold are set to 0. Pixels above the threshold within the 18 kHz echogram are set to 1, pixels in the 38 kHz echogram are set to 3, 70 kHz are set to 29, 120 kHz are set to 7, and 200 kHz are set to 13. These integers were chosen as the summation of any combination of the numbers will produce a unique result. This is an important aspect as the matrices of values representing each frequency component are then summed together to create a single matrix.
Applying the color scale to the data, values above the threshold for 18 kHz were set to light grey, blue for 38 kHz, dark grey for 70 kHz, red for 120 kHz, and yellow for 200 kHz (see below). The summation of the matrices results in pixels with a unique color. Using the color wheel as the fundamental concept, a pixel in the summed matrix that consisted of both 38 kHz (blue) and 120 kHz (red) will be colored purple. Similarly, the combination of 120 kHz (red) and 200 kHz (yellow) will result in an orange pixel. The addition of light grey (18 kHz) to any combination will produce a less saturated color (e.g., a data point in the summed matrix that contains components from 18, 38, and 120 kHz will be colored light purple). The addition of dark grey (70 kHz) will produce a more saturated color.
The MFSBI color scale applied to the imagery. The dots indicate the frequencies where Sv values were above the threshold value.
Processing and plotting are completed using Matlab (Mathworks, Inc, Natick, MA, USA) and data manipulation and application of the algorithm are completed using Echoview (Myriax Pty, Ltd., Hobart, Tasmania, Australia).
Any use of trade names does not imply endorsement by NOAA.
- De Robertis, A., and Higginbottom, I. 2007. A post-processing technique to estimate the signal-to-noise ratio and remove echosounder background noise. ICES Journal of Marine Science: Journal du Conseil, 64: 1282–1291. doi:10.1093/icesjms/fsm112
- Jech, J. M., and Michaels, W. L. 2006. A multifrequency method to classify and evaluate fisheries acoustics data. Canadian Journal of Fisheries and Aquatic Sciences, 63: 2225–2235. doi:10.1139/f06-126
- Ryan, T. E., Downie, R. A., Kloser, R. J., and Keith, G. 2015. Reducing bias due to noise and attenuation in open-ocean echo integration data. ICES Journal of Marine Science: Journal du Conseil, 72(8): 2482-2493. doi:10.1093/icesjms/fsv121
- Wall, C. C., Jech, J. M., and McLean, S. J. 2016. Increasing the accessibility of acoustic data through global access and imagery. ICES Journal of Marine Science. doi:10.1093/icesjms/fsw014
Example of imagery created from a single frequency single beam data file.
The file spans over 15 hours and captures a day and night cycle. The daily up and down movement (or diel vertical migration) of large volumes of marine organisms (sound scattering layer) is also captured. During the day, the sound scattering layer is observed at depth, approximately 300-500 meters depth. After sunset, the layer ascends the water column and spreads out to feed (1:00 to 9:30am UTC). Once the sun rises and visual predators become a great risk, the sound scattering layers descends and aggregates to avoid predation (beginning at 10:00am UTC). The black line is the calculated seafloor. This file was collected by the NOAA Office of Ocean Exploration and Research vessel Okeanos Explorer off the east coast of the United States in 2012 using a Simrad 18 kHz EK60 sonar system.
Example of five frequency singlebeam data files processed with the multiple frequency singlebeam imagery (MFSBI) algorithm outlined above.
Swimbladdered fish reflect strongest at the lower frequencies [18 kHz (grey) and 38 kHz (blue)]. Conversely, zooplankton reflect strongest at the higher frequencies [120 kHz (red) and 200 kHz (yellow)], assemblages of fish and zooplankton layers become immediately obvious. The assemblages of fish (dark grays and blues) appear to follow a depth range of about 450 m to 800 m depth regardless of bottom depth. Zooplankton form two layers within the water column, one (yellow orange) at approximately 50 m and another (red) near 300 m depth. The black line is the calculated seafloor. See the MFSBI color bar. These files were collected by the NOAA National Marine Fisheries Service vessel Henry Bigelow off the east coast of the United States in 2014 using a Simrad EK60 sonar system running 18 kHz, 38 kHz, 70 kHz, 120 kHz, and 200 kHz transducers.