An integrated method for adaptive discrimination of bio-sonar clicks of Irrawaddy dolphins (Orcaella brevirostris) and various types of boat noise pulses for environmental impact assessment on dolphin habitat

Abstract

A new automatic discrimination technique of various types of boat noise pulses and the dolphin clicks was developed for the evaluation of the boat traffic/noise effect on the dolphins' migration behavior. To make the automatic discrimination method work for general use, considering the ICI (Inter-Click Interval) and CF (Centroid Frequency) characteristics of the boat noise pulses and the dolphin clicks, functions for the “clustering” and “cluster model” based analysis were added to the “click train” based analysis method that was previously proposed by the authors [1]. Analyzing detailed CF characteristics of the boat noise pulses and the dolphin's click trains, we created three cluster models for boat noises (type A, type B, type C) and one for the dolphin clicks (type D). To evaluate the reliability of the proposed method, the final process results of each cluster model was compared with the original recorded data. The clustering results of cluster type A and type C showed the effectiveness of the proposed method. Comparison of the results of the dolphin-occurrence (DO) between the proposed method and the previous proposed method, demonstrated that the proposed algorithm for clustering type D was worked well. The proposed method for clustering type D could be applicable for capturing the overall trends of the dolphin-occurrence frequency in a long-term span as a possible alternative of the previous proposed method.