Objective
The overall goal of the project is enhance fisheries management through improved fish stock surveys by enabling the identification of fish species using acoustic methods. The project will apply modern multifrequency fisheries acoustic
1h techniques, in addition to the single frequency methods available previously, to establish methods for acoustic species identification of echo traces that are applicable to routine survey procedures.
The project will: 1) construct an echogram library; 2) apply single-frequency extraction / identification methods on multifrequency data; 3) develop multifrequency algorithms for the differentiation of fish from plankton and for the identification of fish (with and without swim bladders); 4) combine these techniques in a generic algorithm for plankton and fish stock identification.
The project will specifically target the identification of fish that are prevalent in the partnerships survey programme. These encompass both those fish which are targeted for stock assessment purposes (the primary species) and other secondary species, which although not required from acoustic surveys for assessment purposes, nonetheless need to be considered in order to differentiate from primarys. The primary species are anchovy, herring, horse mackerel, mackerel, sandeel and sardine; secondary species include blue whiting, bogue, capelin, cod, chub mackerel, haddock, Norway pout, sprat and whiting.
The work will concentrate on: (1) Construction of an echogram library of ground-truthed multifrequency echograms of fish species from historical survey data going back at least five years to be augmented during the course of the project from survey data. This library, containing raw acoustic data, biological information and environmental parameters will form the core data for subsequent analysis. (2) Analysis of echo trace data for classification criteria and basic environmental parameters to be used for identification. The echogram library will be used to establish the range of shape and environmental features that are typical for each species to produce a probability of identification (Pid). (3) Development of multifrequency algorithms for: the differentiation of fish from plankton the identification of fish without swim bladders; and the identification of fish with swimbladders. Multifrequency acoustic processing techniques incorporating theoretical plankton and fish scattering models where appropriate, will be combined with statistical discrimination methods to produce algorithms for each of the primary species. 4) Development of a combined algorithm for generic fish group, species and plankton identification. The multifrequency techniques will be combined with object classification techniques to provide an identification algorithm for 1° and 2° species.
Fields of science
Call for proposal
Data not availableFunding Scheme
CSC - Cost-sharing contractsCoordinator
BERGEN
Norway