ACOUSTICAL MONITORING OF OPEN MEDITERRANEAN SEA FISH FARMS: PROBLEMS AND STRATEGIES

ACOUSTICAL MONITORING OF OPEN MEDITERRANEAN SEA FISH FARMS: PROBLEMS AND STRATEGIES PACS no. 43.30.+m,43.80.+p,43.60.+d Espinosa, Victor 1 ; Soliveres, Ester 1 ; Estruch, Vicente D. 2 ; Redondo, Javier 1 ; Ardid, Miquel 1 ; Alba, Jesús 1 ; Escuder, Eva 1 ; Bou, Manuel 1 1 Grupo de Dispositivos y Sistemas Acústicos y Ópticos, DISAO. Departamento de Física Aplicada; Escuela Politécnica Superior de Gandía; Universidad Politécnica de Valencia Carretera Nazaret-Oliva S/N, Grao de Gandia 46730 (Valencia) España Teléfono (96) 284.93.14 - (96) 284.93.00 Fax : (96) 284.93.09 2 IMPA-UPV, Departamento de Matemática Aplicada, Escuela Politécnica Superior de Gandía; Universidad Politécnica de Valencia. e-mail: vespinos@fis.upv.es, essogon@epsg.upv.es, vdestruc@mat.upv.es, fredondo@fis.upv.es, mardid@fis.upv.es, evescude@fis.upv.es, jesalba@fis.upv.es ABSTRACT We analyse the application in aquaculture of acoustical techniques to monitor the biomass in open sea Mediterranean cages. We describe the problems when applying sonar techniques to the determination of density, behaviour or growth rate of gilthead sea bream in open sea cages in the Mediterranean. We propose different strategies and research lines from first principles, addressed to obtain a simple setup based on no expensive single beam transducers, capable to be installed in production conditions. INTRODUCTION Marine aquaculture is a strong industry and a certain reality in the Mediterranean Sea. For 2005 the production reached the 84,017 tones of European sea bass (Dicentrarchus labrax) and 93,355 tones of the gilt head sea bream (Sparus aurata), produced mainly in Greeceland, Turkey, Italy and Spain [1]. In spite of the sufficient technology to fulfil the production needs, it is necessary to optimize different production factors, not only to improve its economical profitability but also to minimize the possible ecological impacts. Among these factors we must emphasize the feeding strategy, the growing and population monitoring. The daily feeding is estimated in terms of the present biomass (usually expressed in fodder kg/100 kg of fish mass) and it is a function of different factors like the average size of the fishes, the season, the water temperature, etc. Therefore size (mass) and number of fish estimation reveal as a crucial need, not only for the adequate management of the production but also for the determination of the time to bring the product to the market. The traditional method to control the population has been the periodic manual sampling, by fishing a certain number of specimens, which results to be costly in terms of animal stress and workforce costs. Different non-invasive techniques have been assayed to estimate both fish number and size distribution: video monitoring and digital image processing, electromagnetic pass-through frames, acoustical echo sounders, etc. The effectiveness of every method is limited by different factors and one of the most conditioning facts is the necessity of monitoring a large number of cages almost continuously what impulses the achievement of a permanent, and therefore affordable system, suitable to stand long periods of time in a hard environment. We consider that the acoustical techniques provide such capabilities. Nevertheless still much work must be done to achieve all these objectives, since the technology of scientific or commercial

echosounders has been oriented mostly to pelagic surveys, and both the transducers and common algorithms can not be applied immediately to the aquaculture farms control [2]. In the following sections we will expose the problems we have detected in our experiences in open sea cages along the Mediterranean coast of Valencia (Spain), the strategies of study that we suggest, and the design of the experimental setup we have implemented to develop our research. PROBLEMS Our first approach to the biomass estimation in sea cage followed the scheme of Fig.1, Left, where a split beam 120kHz transducer was placed floating on a platform in the surface using a EY500 Simrad echosunder. Our results with this echosounder were inconclusive [2]. We applied the common theory [3] for density estimation through the volume backscattering and the average Target Strength (TS) of the breams, but it soon evidenced that the problem to solve was not simple: - Several single traces of fishes were detected, but mainly in the distance closer to the transducer, where the far field condition was hardly achieved. Even within this condition, the TS values given by the scientific echosounders must be treated with caution for the case of near range data [4]. Max. cage diameter 30 m Near field 7º Far Field Beam max. diameter 0.6m Fig.1 Left: First experimental setup with a scientific echosounder EY500, the beam width was 7 º. Right: Typical echogram of the EY500, notice the sea bottom echo in red. - The echosounder applied only the usual Time Varying Gain (TVG) correction without having the possibility of correcting the strong beam extinction inside the cage due to the high fish density [5]. - The insonified volume was small and the extrapolation to rest of the cage volume difficult because of the behaviour of the school and the movement of the net with the sea currents. - The stability of the transducer was affected by rapid oscillations due to small wind waves typical of the Mediterranean, and even more, it was practically thrown out of the water when the breams were fed during the summer and their attitude is especially anxious. - The TS results evidenced the difficulty to assign values to fish size classes, as stated in [6], and there was a total lack of studies and models about the acoustical properties of the Mediterranean species of interest in aquaculture. We reconsidered our investigations and realised about the necessity of facing the problem from the very beginning, with a double goal: first, to study theoretically and experimentally the 2

acoustical properties of Mediterranean species like the gilthead sea bream and the problem of size and density determination in a dense bank, and second, to define the characteristics of the system to perform the monitoring of the cages in production conditions. STRATEGIES We made the definition of the experimental equipment for our fundamental studies starting with the considerations about the position of transducers and we conclude that in production conditions the transducer must be permanently placed in each cage, its position must permit the works of fish extraction and feeding, using an only transponder and acquisition/processing system. The position of the fishes in the cages oscillates from the bottom-middle to the very top in the surface, so the far field condition is easily achieved many of the times with the transducer placed in the bottom of the net, permitting to monitor during the feeding, when the bank concentrates at the surface. The need of a high number of transducers in a farm with dozens of cages, and the severe environmental conditions which could cause their periodic replacement, lead us to consider the evaluation of commercial low cost transducers in the range between 50 and 200 khz, like the WS 90-36 of Lowrance, working at 192 khz and a -3 db angle of 20º with a limit of working power of 75 w. Airmar dual frequency (50-205 khz) transducers with higher limit powers were also evaluated. During this evaluation we used a basic configuration based on a function generator, an RF power amplifier ENI 240L, an 100 MHz 4 Channel Digital Oscilloscope Tektronix TDS 3014 and a laptop. The four channels allowed us to monitor three identical transducers; two in the bottom (emitter and receiver) and another one receiving at the surface (see. Fig 2). Fig.2 Left: Second experimental setup with function generator, digital oscilloscope and RF power amplifier with different low-cost commercial transducers. Right above: Correspondent suggested inverted setup with permanent no expensive single beam transducers placed at the bottom with a beam width of 20º. Right below: typical time series record of the receiving transducer at the bottom of a sea cage with gilthead sea breams; the first pulse correspond to the side lobe emission and gives the time reference, the second highest one about 13 ms is the surface echo. This setup permitted us to evaluate the necessary pulse amplitude to cross the bank in different situations, and the performance of the different transducers, we conclude that for feeding conditions a 600w rms power and a dual operation 50/200 khz provide us with the necessary amplitude and higher penetration vs. spatial resolution. We also realized that the effect of waves 3

was less important with the inverted scheme for the fish echoes below the surface than for the case of the floating transducer. This basic echosounder was capable to monitor the fish bank movements inside the cage, important to evaluate its interest in the feeding. We have also determined the sampling needs for acquisition of the echoes series to specify a card-based portable system to perform analysis and representation in a laptop. We found very suitable the use of a National Instruments portable PXI-1031DC with NI PXI-ExpressCard8360 to connect through USB to a laptop, an 100MS/s arbitrary function generator NI PXI-5412, and NI PXI-5102, 2 channel, 20 Ms/s digitizer card both synchronized through the same built-in clock reference. The bus connection between the PXI and the laptop permits to acquire long time series and represent the processed data like echograms through the Labview programming environment. The algorithms are therefore provided by the user and can take into account all the desired phenomena like, e.g., the beam extinction compensation. Together with the ENI 240L amplifier this user-purpose configurable system will allow us to continue our in situ investigations, but it will also provide us with the capability of using special programmable functions to develop a laboratory research in tanks to acoustically characterise the species of interest. Fig.3 Left: Laptop controlled NI-PXI generation and acquisition system with RF power amplifier with different commercial transducers and hydrophone. Right: works in the floating sea cages With this aim, we have already implemented the methodology described in [7] and references there in, based in the study of the changes of the properties of a reverberant cavity in the presence of scatterers, and improved it introducing the use of broadband signals like time stretched pulses (TSP) and pseudo-random maximum length sequences (MLS). This technique precises of the use of omnidirectional transducers with a planar response in the band of interest and allows obtaining the total TS from sets of impulses responses of the reverberant cavity while the fishes are swimming inside. The analysis of the total TS as a function of frequency provides a tool for shape and size characterization of the fishes. We have prepared our 3 installations with 9 identical 2x2x1 m tanks to measure different size classes of breams, from 25 g to 250 g. 4

CONCLUSIONS We have initiated a research line in a complex problem and developed a complete experimental system to perform fundamental investigations addressed to apply acoustical techniques to the control of open sea farms in de Mediterranean. We have identified the main theoretical and technical challenges and proposed a particular configuration for the production needs. Our work in laboratory tanks is now involved in the acoustical characterisation of the gilt head sea bream (Sparus auratus, Linnaeus 1758). ACKNOWLEDGEMENTS This work has been supported by the Generalitat Valenciana in the frame of the Project GV06/097. REFERENCIAS [1] APROMAR, La acuicultura marina de peces en España, 2006, www.apromar.es, 2006. [2] V. Espinosa, J. Ramis and J. Alba, Evaluación de la sonda ultrasónica EY-500 de Simrad para el control de explotaciones de dorada Sparus auratus Linnaeus, 1758, Bol. Inst. Esp. Oceanogr. 18 (1-4), 15-19, 2002 [3] MacLennan D.V. y E.J. Simmonds, Fisheries acoustics. Chapman & Hall, London, 1991. [4] Furusawa M., Hamada M. and Aoyama C., Near range errors in sound scattering measurements of fish. Fisheries Science, Vol. 65 (1), 109-116, 1999. [5] Kenneth G. Foote, Correcting acoustic measurements of scatterer density for extinction, J. Acoust. Soc. Am. Vol. 88, No. 33, September, 1990. [6] Methodology for Target Strength Measurements, ICES cooperative Research Report No. 235, Ed. E. Ona, 1999 [7] S.G. Conti, P. Roux, Ch. Fauvel, B.D. Maurer, D.A. Demer Acoustical monitoring of fish density, behavior, and growth rate in a tank, Aquaculture, 251, 314-323, 2006. 5