In a naïve representation, as the split-beam passes by a single scatterer, the measured alongship angle will

suffer a monotonous variation from positive to negative values, while the athwartship angle detected EX 527 order will show a more uniform value. In the case of a shellfish patch, the multiple scatterings will cause the angles (determined from the phase differences detected) to spread around the actual positions, but the time evolution of the angles will be retained. Although their backscattered intensity is superimposed in the same way on the rest of the bottom backscatters, making them indistinguishable in the energy echogram, their angular information will compete with the interface returns and sediment volume backscatter, drawing a complex picture. The split-beam angular information was processed to provide a textural characterisation AZD0530 price of the echogram. First-order statistics do not offer information about variations in the angular echograms that would denote the presence of razor shells. Thus, a second-order statistical procedure, aimed at detecting correlations between neighbouring acoustic samples, should be applied in the form of a textural analysis

(Haralick et al., 1973 and Zaragozá et al., 2010). The most used second-order statistic is the co-occurrence matrix, whose cell pij contains the fraction of pairs of the neighbouring signal samples (echo bins) having quantised levels i and j respectively in a preset window and after signal quantisation in N levels ( Haralick et al. 1973). The neighbouring samples of a bin can be defined in two natural ways: along the pings (being neighbours, the previous and the next bin in the same ping) or along depths (being neighbours, the bins of consecutive pings corresponding to the same depth below the detected sea bottom). We will refer to the first neighbour definition as Type 1

(or along pings) and the second one as Type 2 (or across pings). The CYTH4 resulting co-occurrence matrix will be symmetric as if i is followed by j, then both (i, j) and (j, i) bin pairs are counted. Based on the co-occurrence matrices, Haralick et al. (1973) introduced the so-called textural features. Thirteen Haralick textural features (denoted as H1 to H13) have been calculated for both the alongship and athwartship angles. Another textural feature (lacunarity, Lac), describing the relationship between the co-occurrence standard deviation and the mean value, was also calculated. These variables are mathematically defined in the Appendix. We have restricted the textural analysis to those bins contained between the bottom surface and the equivalent to 30 cm of sediment depth. This depth corresponds to the main insonified region of the echogram and also to the corer sample depth range.