The SIMFAMI database: a library of ground truthed acoustic survey data.

Acoustic surveys are being used increasingly to estimate the abundance of fish resources. One of the major uncertainties associated with the estimates is the identification of echotraces to species. Significant progress has been made recently in developing algorithms which use echotrace and multifrequency backscattering characteristics to identify fish schools. These algorithms are developed and tested on echotraces which have been identified by trawling, a process akin to groundtruthing. This paper provides a description of an ORACLE online database which holds a digital echogram library of ground-truthed echograms for a number of pelagic species. The database is based on the SIRENO (Seguimiento Integrado de los REcursos Naturales Oceánicos) software, developed by the Instituto Español de Oceanografía. It was developed for an EC project (SIMFAMI) aiming to develop species identification algorithms. The database stores data from acoustic surveys as: echogram images at different frequencies; polygon coordinates identifying the area in the echogram from which the trawl sample was taken; catch data (position, depth , species composition, etc.); and environmental data. The raw acoustic data is not held in the database, but is accessed on request from the host institute by reference to the filename. All acoustic data files are saved in a standard format (HAC files) to enable the exchange of data across multiple software platforms. The database can be filtered to obtain subsets of data verifying certain requirements. The database serves as a model for similar such species identification studies and provides a centralised searchable store of ground-truthed echotraces of the major commercial pelagic species of northern European waters.