Tarefa 5: Desenvolvimentos dos métodos para a estimativa de distâncias e localizações das fontes acústicas usando os dados OBS e EAR
Development of methods for estimation range and source location using OBS and EAR data
Provided that vocalization rates are known or estimated then data from fixed acoustic sensors can be used to obtain population density using the distance sampling methodology. This procedure requires knowledge on the range (or slant range for shallow diving species like fin whales) to the sound source and on the probability of detection (addressed in tasks 3 and 6). If, in addition to the slant range a location of the whale is known along the time, then PAM can be used for behaviour investigations.
In this task we will develop and test current and new methods for ranging and location of fin whales using passive acoustic data recorded by hydrophones (EARs) and OBS answering to several challenges: i) applying the methods in an automatic procedure to be used with large datasets; ii) applying the methods in areas where the sea bottom is covered by soft sediments using signal enhancement techniques; iii) applying the single station method developed for OBS to ranges beyond the critical distance; iv) applying the Lloyd’s mirror interference pattern to infer the sound source depth. The procedures developed in this task will be tested, calibrated and validated using the datasets collected in tasks 1 and 2 by EARs and OBS analysed in Task 3.
We propose to adapt the automatic detection and location algorithms developed by Grigoli et al. (2013) for earthquakes to the fin whale vocalizations. Instead of P and S arrivals we will consider the primary and multipath arrivals. The propagation model required will be computed by the ray tracing method of Hovem (2013) that accounts for an irregular bathymetry and uses 1D model for the sound speed in the ocean. This is a method that requires that sound sources are recorded by multiple sensors.
For single sensor ranging we will expand the method of Matias and Harris (2015) to the use of multipaths. Preliminary work on the dataset used by Harris et al. (2013) shows that the first multipath can be recovered after some signal enhancing techniques are applied. Beyond the critical distance the sound source azimuth can still be computed from the OBS data with ±180º ambiguity. This ambiguity can be solved if part of the whale track falls inside the critical range. For EAR data, the same methods will allow for slant range estimation but not 2D location.
Signal enhancement techniques that are promising are envelope computation and stacking or coherency filtering. Investigation of the spectral properties fin whale calls show a clear interference pattern that is attributed to the Lloyd’s mirror effect. OBS data that includes the recording of the vertical ground velocity and pressure have the advantage that, due to the opposing reflecting coefficient at the sea surface, interference patterns are in opposite phase thus allowing extra constrains for the inference of the vocalization depth. Without this extra constrain, EAR data processed in a similar way will also allow for depth estimates.