-
Balk, Helge
(2022).
Bredbånd verktøy i Sonar5-Pro.
-
Balk, Helge
(2022).
Processing Acoustic Camera data with Sonar5-Pro.
-
-
Rakowitz, Georg; Tuser, Michal; Riha, Milan; Balk, Helge & Kubecka, Jan
(2010).
Use of high-frequency imaging sonar to observe fish behaviour with respect to an active surface trawl.
-
Balk, Helge; Kubecka, Jan & Lindem, Torfinn
(2010).
Detection, tracking and sizing of fish in data from DIDSON multibeam sonars.
-
Kubecka, Jan; Frouzova, Jaroslava; Balk, Helge; Čech, Martin; Draštík, Vladislav & Prchalová, Marie
(2009).
REGRESSIONS FOR CONVERSION BETWEEN TARGET STRENGTH AND FISH LENGTH IN HORIZONTAL ACOUSTIC SURVEYS.
Vis sammendrag
Abstract: Both ways conversions between fish target strength (TS) and length (L) are extremely important for the accuracy of acoustic fisheries surveys. At present, no universal method is available for aspect corrections during mobile horizontal surveys. It is highly recommended to check common assumption of random fish aspect distribution by the series of fixed-location observations. For random aspect situations, the deconvolution of fish aspect in the beam can be used converting all TS to side aspect whose relationship to length is well known. For converting catch data to TS, the regression of TS of average horizontal backscattering strength (TSAS) on length can be applied successfully. The average regressions for common European species are being suggested for routine acoustic surveys in combination with direct pelagic fishing for the assessment of ecological status of lakes and reservoirs for EU Water framework directive 2000/60/EC. This approach reduces the resolution towards local fish size variations but simplifies the processing and provides much better definition of fish abundance.
-
Balk, Helge; Lindem, Torfinn & Kubecka, Jan
(2009).
New Cubic Cross filter detector for multi beam data recorded with DIDSON acoustic camera.
Vis sammendrag
Target detection in data from the DIDSON (dual-frequency identification sonar) is a challenge. Due to high frequency and many narrow beams, the system can take series of acoustic pictures of passing fish. The entire body can be seen, but are often blurred, and unclear in individual pictures. It is only when series of pictures are played like a film, one can see that the target actually is a swimming fish. Hence, a good target detector should not be based on individual frames, but on using series of pictures.
The Crossfilter detector was originally designed for single, dual and split beam systems, to detect targets under very low signal to noise ratio conditions. In order to operate on multi beam data, we have added an extra dimension to the detecting filters. This resulted in a detector that takes into account both the time, range, and width dimensions in the data during the detection process. The talk will focus on the principles and construction of the detector, and discuss experience and test results.
-
Balk, Helge & Lindem, Torfinn
(2009).
Target detection in data from splitbeam echosounders and multi beam sonar systems.
-
Drastik, Vladislav; Hohausova, Eva; Kubecka, Jan; Balk, Helge & Froutsova, Jaroslava
(2009).
Classification of fish and plant echoes in acoustic surveys.
Vis sammendrag
Echoes from submerged plants could cause serious problems during acoustic estimates of fish in lakes and reservoirs. Defined experiments with known fish and submerged plants in front of horizontally oriented split beam echosounder SIMRAD EK 60, 120 kHz were carried out in calibration pond. Three widespread species of aquatic plants with different “stem/stalk” morphology: Myriophyllum spicatum,
Polygonum amphibium and Potamgeton pectinatus were installed in front of the transducer. Single plants as well as patches of different sizes of each species were recorded. Several acoustic parameters
were used to qualify the differences of fish and plant echoes. The echo duration of plant patches exceeded any values detected with fish while single plants were able to produce well defined single echoes.
Moreover, larger fish could produce wider echoes than theoretical point scatterers. The overlap of pulse durations gives little hope for distinguishing between single stems and fish. Fish echoes had
much less variable phase readings compared to even single plants. Study of fish and plant echoes allowed distinguishing of submerged plants from fish and provides more precise fish stock assessment.
-
Balk, Helge & Lindem, Torfinn
(2008).
Analysis and visualization of data from DIDSON Acoustic camera.
-
Tuser, Michael; Balk, Helge; Froutsova, Jaroslava; Kubecka, Jan & Muska, Milan
(2008).
Suitability of DIDSON acoustic camera for quantitative fish stock assessment by mobile surveys.
Vis sammendrag
Open water of lakes and rivers could be surveyed by DIDSON (dual-frequency identification sonar) beaming sideways from a moving survey boat. Surface-oriented fish is then onsequently recorded in all DIDSON beams. The detection is considered successful if fish is seen and its size can be read reliably. For mobile horizontal surveying by DIDSON, vertical opening of the sampling volume is essential. The standard acoustic experiment with a “fish rotating carousel” was conducted with DIDSON in the Rimov reservoir (Czech Republic). Cyprinid fish (bream Abramis brama, roach Rutilus rutilus and carp Cyprinus carpio) of known size and aspect angle were deployed in known positions of high-frequency DIDSON fan of beams. Fish were stunned and measured before being mounted in a rotatable frame of the carousel. Both fixed and simulated mobile records were obtained in two ranges (approx. 6 and 9.5 meters). The following aspects of fish were sampled: 1) side, head and tail aspects in different off-axis positions achieved by yawing and then tilting of DIDSON transducer, 2) all aspects in the horizontal plane by turning fish at a constant speed, and 3) random aspects by turning fish in the horizontal plane together with tilting of DIDSON transducer. Vertical dimension of the sampling volume, where the fish presence, size and aspect could be recorded varied slightly with aspect and length of fish. This means that every target has own definition of sampling volume. New echocounting philosophy for mobile surveys by DIDSON will be presented.
-
Kubecka, Jan; Balk, Helge; Drastic, Vladislav; Frouzova, Jaroslava; Gassner, Hubert & Herzig, Alois
[Vis alle 10 forfattere av denne artikkelen]
(2008).
Large-scale acoustic surveys of an extremely shallow lake: Neusiedler See, Austria.
Vis sammendrag
Neusiedler See covers an area of 321 km2, the average and maximum depth is 1.2 m and 2 m respectively. Thermal stratification is absent at night. The lake was surveyed for 10 years by extensive mobile night surveys using 120 kHz split-beam echosounders with horizontally oriented elliptical transducer. Some surveys were accompanied by direct fish capture by 15 kW DC electroshocker, specially constructed wheeled trawl, push net for sampling larval and early juvenile fish and Nordic multimesh gillnets. The paper addresses the question of maximizing target detection in data characterized with low signal-to-noise ratio (cross filters), threshold strategies, avoidane of spherical spreading, multipathing, distinguishing macrophytes and fish and sizing of horizontally recorded fish. In calm conditions it was possible to detect fish down to young-of-the-year and the length frequency distribution was similar to the catch of direct methods combined. The fish presence detected by all methods was the highest in protected areas of the lake. CPUE of the gear show good correlation with acoustically detected densities showing that acoustic surveying can be useful even in extremely shallow habitats.
-
-
Gjelland, Karl Øystein; Knudsen, Frank & Balk, Helge
(2008).
Bias in swimming speed estimates from splitbeam acoustics should be corrected for with smoothing.
Vis sammendrag
Swimming speed estimates based directly on splitbeam acoustic recordings are likely to be biased due to instability in single echo detections. Although this is well known, experiments addressing the problem have been lacking. This study quantifies the bias and evaluates the effect of smoothing fish tracks to obtain accurate swimming speed estimates. Field recordings of fish and controlled experiments with calibrations spheres were used to derive target speed detected with stationary split-beam echo sounding. The fish tracks were used to determine an appropriate smoothing window that conserved the shape of the swimming trajectories, while removing detection jitter. Speeds estimated from smoothed and nonsmoothed trajectories were then compared to investigate the effect of smoothing. The results from controlled experiments and field data showed that swimming speed estimated from non-smoothed fish tracks are likely to be heavily biased. The bias increased with range and was inversely related to the actual swimming speed. Smoothing reduced the variance of the estimates, and removed any effect of range and speed on the coefficient of variation. We conclude that fish tracks recorded with splitbeam echo sounders should be smoothed to yield reliable and precise swimming speed estimates.
-
George, Rakowitz; Keckeis, Wolfgang; Herold, Wolfgang; Kubecka, Jan & Balk, Helge
(2008).
Evaluation of different fish size estimators in horizontal acoustic applications.
Vis sammendrag
The commonly applied mean TS from tracked fish obtained by horizontal echo sounding in shallow rivers is unreliable for fish sizing due to high within track TS variations and the way echoes are formed. In order to find a better size-estimator than the existing we tested mean-TS, 13° pan corrected mean-TS, 75th, 80th, 85th, 90th 95th TSpercentiles, the MET-MIT-statistic size estimation method and echo length with regard to real size data. Total-length estimates from the mean TS as well as from the 13° pan corrected mean TS underestimated the size of the fish. The TS-percentiles of the upper quartile make use of the side aspect detections within a fish-track, represented by the strongest signals. We found that the 95% TSpercentile was the best estimator for fish sizing.
-
Froutzova, Jaroslava; Kubecka, Jan & Balk, Helge
(2008).
Fat echoes of oblique aspects : consequences for fish single echo detection.
Vis sammendrag
Echo duration is one of basic definitions of fish single target. Many of wild fish echoes fail to satisfy theoretical requirements based on pulse duration, bandwith and signal-to-noise ratio. The relationship between the echo duration and fish aspect was studied using rotating carousel by horizontally oriented split-beam echosounder, 120 kHz. While at side aspect most fish performed like point scatterer, the echo duration appear to increase nonlinearly at more diagonal aspects especially with large fish. Head and tail of larger fish may consist of several very close peaks, which may correspond to different body structures. Regressions between fish size, aspect and pulse durations are presented for common European freshwater species. The consequences for the settings of single echo detectors
are discussed.
-
Balk, Helge; Lindem, Torfinn; Kubecka, Jan & Frank, Knudsen
(2007).
Recent advances in solving signal-to-noise problems in fisheries acoustics.
Vis sammendrag
Signal is what we want while noise is disturbance that prevents us from getting it. Especially for horizontal application, surface, bottom, temp profile and unknown fish aspect can cause breakdown in the sonar equations and serious fish stock estimation errors. Noise may blur, hide or misshape the signal so that the echoes cannot be recognised by traditional detectors. Noise may also appear as disguised target echoes and bias the estimates.
We will never be able to conquer the noise. However, new detectors and target noise separators, aspect de-convolution, multi beam and multi frequency will be discussed as ways to push the limits.
Principle, later improvements and experience with the cross filter detector will be presented.
We have measured and modelled sound propagation in lakes and rivers and will show how temperature gradients influence on the sound propagation and how this break down the assumptions behind the sonar equations and cause errors in the TS and Sv estimates.
Experiences with multi beam and multi frequency are relatively new in fresh water and we will be shared as well.
-
Balk, Helge & Lindem, Torfinn
(2007).
Analysis of data from high frequency identification sonars combined with data from conventional echo-sounders.
Vis sammendrag
High frequent identification sonars have short range and produce a waste amount of data similar to videos. Analysis of these data is commonly done by watching the recorded videos. For surveys lasting for days or weeks this method is too time consuming. By generating traditional echograms from the sonar data and by combing the sonar data with data from long range split beam echo-sounders, we can improve the systems range and decrease the analysis time.
-
Rudstam, Lars G.; Gal, Gideon; Balk, Helge; Knudsen, Frank R.; Axenrot, Thomas & Boscarino, Brent
(2007).
Multiple acoustic frequencies increase our ability to discriminate between fish, mysids and zooplankton.
Vis sammendrag
We analyzed data from five acoustic frequencies (38, 120, 200, 430 and 715 kHz) and two manufacturers (Simrad and Biosonics) collected in Cayuga Lake, one of New York’s Finger Lakes with similar species composition as Lake Ontario, Michigan and Huron. Alewife dominated in gill net catches and Mysis relicta in vertical net hauls. The frequency response was dramatically different for fish and mysids indicating that frequency differencing will be useful for separating fish from mysids in the Great Lakes. We also applied filters to remove fish echoes and calculated mysid and fish densities separately using the filtered data. Empirical measurements of mysid target strength at different frequencies was compared with a published theoretical scattering model previously applied to mysid at 420 kHz by Gal et al. Our results suggest that addition of one additional frequency (38 kHz) would help separation of fish and mysids, and that higher frequencies may be useful for smaller zooplankton.
-
Balk, Helge
(2006).
Sound waves and fish abundance.
Vis sammendrag
Our knowledge about how sound can be applied to see fish, has developed since Leonardo DaVinci discovered that sound waves could propagated through water. To day we have split beam, multi beam, multi-frequency echo sounders. Wide band echo sounders are about to be introduced. The talk will give a brief overview of the history, equipment and consideration about the future. What is possible and what is not possible?
Split beam echo sounding is the most common method today and will be a main topic in the talk. Experience, analysis methods, and experimental results from rivers, lakes and ponds in Scandinavia and Europe will be presented.
-
Balk, Helge; Lindem, Torfinn & Kubecka, Jan
(2006).
Enhenced Crossfilter detector for target detection in data from echosounders.
Vis sammendrag
Acoustic target detection is commonly carried out with single echo detectors based on echo length. These detectors test one ping at a time and look for echoes with an assumed correct echo length. Obtained echoes are filtered with a set of criteria such as phase stability and position in the beam before they are accepted. In shallow water, noise phenomena can distort echoes from fish and false fish echoes can be generated. This causes the echo length detector to produce fractionated tracks from fish surrounded by numerous noise detections.
Echo length detectors utilize only small portions of the information available in a split beam echogram. By including information from more than one ping and from the background reverberation, a more robust fish detector has been designed. This detector, called the Cross Filter Detector (CFD), has been further improved by extracting detection power from the variance in the echogram
-
Balk, Helge
(2006).
Sound waves and fish abundance.
-
Balk, Helge
(2006).
Introduction, presentation and theory for multifrequency hydroacoustics.
-
Balk, Helge
(2005).
New methods for fish and bottom detection in data from split beam echosounders.
-
Balk, Helge
(2005).
A Ph. D. Class on Marine Bioacoustics: Hydroacoustics.
-
Balk, Helge; Lindem, Torfinn & Kubecka, Jan
(2005).
New single echo detection methods for shallow water fisheries acoustics.
Vis sammendrag
Acoustic target detection is commonly carried out with parametric single echo detectors. These detectors test one ping at a time and look for echoes fulfilling a set of criteria such as echo duration and shape. In shallow water, noise phenomena can distort echoes from fish and false fish echoes can be generated. This causes the parametric detector to produce fractionated tracks from fish surrounded by numerous noise detections.
Parametric detectors utilize only small portions of the information available in a split beam echogram. By including information from more than one ping and from the background reverberation, a more robust fish detector has been designed. This detector, called the Cross Filter Detector (CFD), has now been further improved by applying the variance in the split beam echo sounders phase measurements
-
Balk, Helge; Lindem, Torfinn & Kubecka, Jan
(2005).
New single echo detection methods for shallow water fishery acoustics.
Vis sammendrag
Acoustic target detection is commonly carried out with parametric single echo detectors. These detectors test one ping at a time and look for echoes fulfilling a set of criteria such as echo duration and shape. In shallow water, noise phenomena can distort echoes from fish and false fish echoes can be generated. This causes the parametric detector to produce fractionated tracks from fish surrounded by numerous noise detections.
Parametric detectors utilize only small portions of the information available in a split beam echogram. By including information from more than one ping and from the background reverberation, a more robust fish detector has been designed. This detector, called the Cross Filter Detector (CFD), has now been further improved by applying the variance in the split beam echo sounders phase measurements.
-
Balk, Helge
(2004).
Presentation of research results. Advanced analysis methods for fisheries acoustics. Detection methods for data characterised with low signal to noise ration.
-
Balk, Helge
(2004).
Reflection and diffraction of sound waves in shallow water and how this influences the sonar equations.
-
Balk, Helge & Lindem, Torfinn
(2003).
Fish detection based on spectral differences in the echogram's range and temporal domain.
-
Balk, Helge & Lindem, Torfinn
(2003).
A new method for single target detection.
Vis sammendrag
We have experienced that fish detection in sonar-data from horizontally
aligned transducers in shallow water can be difficult. Studying
recorded material revealed that echoes from fish often took on high
ping to ping variations in variables like pulse length, intensity,
shape and phase. Common single-echo detection methods based on a set of
echo criteria tended to overlook echoes from fish. At the same time
echoes from unwanted targets and fluctuations in the background
reverberation was accepted. Hence, tracking and fish counting became
difficult.
To overcome these problems, we have developed a detection method based
on 2-dimensional low-pass filters. By adjusting the cut-off frequency
in the time and range domain, noise can be removed, ping to ping
variation in signals from fish reduced, and local background
reverberation level can be found. By subtracting the echo intensity in
the filtered echograms, traces from fish can be detected.
-
Balk, Helge & Lindem, Torfinn
(2002).
Fish detection in rivers with split-beam sonars.
Vis sammendrag
Split beam echo sounders can monitor fish in shallow rivers. Positioned horizontally and normal to the river current at a suited place, the conical sound beam can cover major parts of the river cross-section and register passing fish. However, the nearby bottom and surface, air bubbles and debris makes the analysis of the echo sounder data difficult. A summary of our experiences from the last six years with data collection and analysis is presented. Methods such as single echo detections, tracking, cross filter detection and classification are discussed.
-
Lindem, Torfinn; Balk, Helge; Vibeke, J. & Kjølerbakken, K. M.
(2002).
Measurements of acoustic sound fields in a shallow river.
Vis sammendrag
From earlier experiments, we have seen indications that the sound field
from a horizontally aligned transducer in shallow water can differ from
the expected field. If this is the case, this phenomenon can have an
impact on the measured target strength from fish and on sample volume.
Hence, it has been important to verify this by controlled experiments.
We have measured the sound field produced by a Simrad ES120-4, 120kHz,
4x10 deg. split beam transducer in a shallow river. These measurements
were made with a Reason hydrophone. Reflections from the bottom and
surface made modifications to the transmitted pulse. In some positions
they could even form separate pulses. These pulses were sufficiently
separated in time to produce phantom echoes. Results from this
experiment will be presented.
-
Lindem, Torfinn; Balk, Helge & Knudsen, Frank R.
(1999).
Sound propagation through shallow water can make unexpected modification to fish echoes recorded with a horizontally looking split beam system.
Vis sammendrag
In several experiments with a 120kHz split beam system we have noticed
that shallow water in some cases can make dramatic changes to the echo
signal. Especially angular measurements in the vertical plane can be
modified by surface and bottom boundaries. Erroneous measurements of
position and target strength can be the result if beam mapping is not
done in a proper way before recording of actual fish echoes. The paper
will demonstrate some of these problems.
-
Balk, Helge & Lindem, Torfinn
(1999).
Fish tracking in shallow water by image processing.
Vis sammendrag
Automatic fish-counting in data from Split-beam sonar is traditionally
performed by a single echo detection (SED) process followed by some
sort of neighborhood tracking. The SED process uses parameters such as
echo pulse duration and phase deviation to remove echoes with non-valid
angular measurements. The neighborhood tracking algorithm then combines
echoes close in time and position. This method works fine on split-beam
data from vertically mounted sonar in open water. However, with
horizontally mounted sonars in shallow water rivers this method tends
to fail. In order to use hydroacoustics in shallow river fish stock
assessment, improvements on the counting method has to be done with
methods from image processing.
-
Lindem, Torfinn & Balk, Helge
(1999).
Problems related to Time Varied Gain (TVG) and phase measurements with elliptic split beam transducers in shallow water, - some observations.
Vis sammendrag
For several years split beam echo sounders have been used by biologists
for counting fish in shallow water. Special elliptic transducers have
been developed by manufactures like SIMRAD and HTI for use in
horizontal applications in shallow lakes and rivers. For some time
users have reported ¿strange behaviour¿ of their systems in some
locations. During our ongoing project for new counting methods for fish
in rivers we have experienced these effects and can demonstrate how
boundary effects from surface and bottom can give unexpected
modifications to echo signals received with split beam systems.
-
Balk, Helge & Lindem, Torfinn
(1998).
Hydroacoustic fish counting in rivers and shallow waters, with focus on problems related to tracking in horizontal scanning sonars.
-
Järnegren, Johanna; Balk, Helge; Uglem, Ingebrigt & Forseth, Torbjørn
(2011).
Telling av oppvandrende fisk i Mandalselva ved bruk av DIDSON. En pilotstudie.
Norsk institutt for naturforskning.
ISSN 978-82-426-2215-0.
Fulltekst i vitenarkiv
Vis sammendrag
Moderne forvaltning av norske laksebestander basert på gytebestandsmål forutsetter god og
pålitelig kunnskap om bestandsstørrelse og/eller beskatningsrater. Antall oppvandrende laks
kan estimeres enten indirekte gjennom fangststatistikk og beskatningsrater eller direkte ved
bruk av ulike fisketellesystemer.
”Dual frequency identification sonar”- systemet (DIDSON) representerer en ny type identifikasjonssonarer
som under optimale forhold kan generere undervannsbilder med tilnærmet videokvalitet.
Sammenliknet med videosystemer gir DIDSON fordeler ved at den ikke er avhengig av
eksternt lys som dagslys eller kunstig lys, samt at den også fungerer i grumset vann. Optimal
bruk av DIDSON er avhenging av observasjonslokaliteten. Hensikten med dette prosjektet var
å teste hvorvidt DIDSON kan brukes til fisketelling i Mandalselva og i andre elver av tilsvarende
størrelse.
DIDSON kan brukes til å måle antall og størrelse på oppvandrende laksefisk i elver hvor denne
typen informasjon tidligere har vært vanskelig eller umulig å fremskaffe. DIDSON er brukt for å
telle oppvandrende laks blant annet i Canada og England, men systemet er hittil ikke testet
under norske forhold. Hvorvidt DIDSON-systemet er anvendbart til rutinemessig telling og måling
av oppvandrende laks i norske elver er i hovedsakelig avhengig av to forhold; for det første
at systemet kan identifisere oppvandrende fisk på en pålitelig måte og deretter at analysene er
kostnadseffektive. Dette er igjen avhengig av optimale observasjonslokaliteter og i hvilken grad
analysene kan automatiseres.
Lokaliteten som ble vurdert som best egnet til denne studien ligger nedenfor Nødingfossen og
data ble samlet inn mellom 11. juni og 20. oktober 2009. En DIDSON-LR enhet (Long Range)
med en pan/tilt rotor ble satt ut i elven slik at avstanden fra sonarens front til nærmeste elvebredde
var ca 5 meter. Sonaren ble plassert slik at man fikk det beste bildet på skjermen
med maksimal rekkevidde.
Alle data ble analysert med programvaren Sonar5-Pro (S5), som ble modifisert for å kunne
håndtere de store datamengdene som ble samlet inn. Automatisk telling ble vurdert og mange
metoder ble testet og forkastet som ikke tilstrekkelig nøyaktige. Metodene detekterte for mange
ikke-ønskede mål og overså for mange fisk. Den eneste metoden som fungerte for de innsamlede
dataene var manuell klikktelling. Årsaken til dette var mye støy forårsaket av ekko fra
bunn, overflate, luftbobler, turbulens og ikke minst den lange avstanden over elven. Med manuell
analyse ble ett døgns opptak opparbeidet i løpet av ca 60 min effektiv analysetid. Forholden
var ikke gode nok til å måle størrelse på fisk.
Totalt ble 4471 fisk talt mellom 11. juni og 20. oktober. Kompensert for tapt opptakstid, areal
med manglende dekning samt fisk som vandret forbi sonaren flere ganger estimerte vi en minimum
oppvandring på 6451 fisk. Ut fra registrerte fangster gir dette et estimat for maksimal
beskatning på 35 %.
Konklusjonen er at DIDSON kan brukes for rutinemessig telling. Automatisk telling av oppvandrende
fisk forutsetter at lokaliteten tilpasses slik at tilstrekkelig gode data kan samles inn.
-
Balk, Helge
(2001).
Development of hydroacoustic methods for fish detection in shallow water.
Unipub forlag.
Vis sammendrag
Management of salmon stocks from various regions is of great importance. Optimal management can only be achieved with knowledge of the size and the changes in the stocks. Hence, monitoring methods are important. In marine fisheries, split-beam echo sounders are frequently applied to monitor fish. In recent years this hydroacoustic fish detection method has been applied in many rivers. Aligning one or a few transducers horizontally provides a simple way to monitor major parts of a river's cross section.
However, the horizontal shallow water application introduces new problems related to the acquisition, analysis, and interpretation of the sonar data. Fieldwork has been carried out in lakes and rivers, and a systematic analysis of these problems has been made. From the experience and the studies of the recorded material, new methods have been developed. Signal processing based on image analysis and classification theory is introduced, and promising results have been achieved.