C1 Humpback Whale Research

Background

Humpback Whales within the southwestern Indian Ocean undertake annual migrations in summer from the Antarctic/Southern Ocean feeding grounds into winter breeding grounds in the tropical and sub-tropical coastal waters of Mozambique, classified as the C1 Breeding Stocks.

Baleen whales are well known for complex vocal behaviors that are seasonally and geographically stratified. Humpback Whales ( Megaptera Novaeangliae) are considered to possess one of the most complex acoustic reportire among Baleen whales.

The C1 Breeding Stock found in Bazaruto Archipelago has never been acoustically researched before, and no acoustically-based conservation measures have been taken before to ensure their welfare in the Archipelago. Deep Voice embarked on a research expedition of data collection regarding two research aspects:

Acoustic Data Collection-Underwater recordings of the local humpback whale song.
Acoustic Data Collection of Social Calls- the less common form of humpback communication compared to the whale song in the breeding ground.

The above Data collected is the baseline for a development process of algorithms that can impact environmentally.

The Method

Data Collection:

The first research expedition was conducted at the Bazaruto Center for Scientific Studies, located at Bazaruto Archipelago in Mozambique. The group spent 21 days at the BCSS, with 13 days of field recordings. More than 5 hours of recordings and photos of dozens of individuals were collected. The skills required for the data acquisition phase are underwater video photography of Humpback Whales behavior and acoustic data acquisition using hydrophones.

Data Analysis:

The Data is manually tagged to be used as the baseline for algorithmical development, which contains the following phases:

Denoising Algorithm- The data recorded is SNR challenging for analysis. A preceding denoising step should be taken to ensure that the data can be automatically detected.

Automated Detection Algorithm- Using the denoised annotated data to build automatic detector for the C1 group.

Statistical Acoustic Features Extraction- By understanding the physical parameters of the local whales communication, there is a potential for environmental regulations recommendations that can assist the whales.

Density Estimation Assessment-The information of the population distribution in space and time is the key for regulations recommendations that can prevent ship collisions and entanglements.

The Results

a. The research and engineering group developed a unique acoustic toolbox for denoising published in the WMMC 19 to aid the entire bio-acoustical community worldwide, as an open-source code

b. The team is currently working on implementing an automated detection algorithm for the data with high detection versus false alarm rate.

FUTURE GOALS

Collecting more data of the C1 breeding group, especially the mother- calf communication in order to improve the existing algorithmic development.
Publishing a paper that contain the algorithmic implementations developed by the group.
Implement the research findings to tangible regulations recommendations, ensuring the whales welfare in the area.
Expanding the group’s scientific span to more core bio-acoustical problems regarding cetaceans communication.