MODEL-BASED COVARIANCE MEAN-VARIANCE CLASSIFICATION TECHNIQUES - ALGORITHM DEVELOPMENT AND APPLICATION TO THE ACOUSTIC CLASSIFICATION OF ZOOPLANKTON

For inversion problems in which the theoretical relationship between o bserved data and model parameters is well characterized, a promising a pproach to the classification problem is the application of techniques that capitalize on the predictive power of class-specific models. The oretical models have been developed for three zooplankton scattering c lasses (hard elastic-shelled, e.g., pteropods; fluid-like, e,g,, eupha usiids; and gas-bearing, e.g., siphonophores), providing a sound basis for model-based classification approaches. The Covariance Mean Varian ce Classification (CMVC) techniques classify broad-band echoes from in dividual zooplankton based on comparisons of observed echo spectra to model space realizations. Three different CMVC algorithms were develop ed: the Integrated Score Classifier, the Pairwise Score Classifier, an d the Bayesian Probability Classifier; these classifiers assign observ ations to a class based on similarities in covariance, mean, and varia nce while accounting for model space ambiguity and validity. The CMVC techniques were applied to broad-band (similar to 350-750 kHz) echoes acquired from 24 different zooplankton to invert for scatterer class a nd properties. All three classification algorithms had a high rate of success with high-quality high SNR data. Accurate acoustic classificat ion of zooplankton species has the potential to significantly improve estimates of zooplankton biomass made from ocean acoustic backscatter measurements.