Ongoing Projects (Funded)

Understanding the ever-changing oceans, biota and atmosphere is one of the greatest global challenges. The future of measuring and forecasting long-term trends in coastal and deep-water ecosystems and climate lies in sustained long-term measurements from ocean observing systems. In this project, we will develop a novel statistical QA approach for big data from marine observatories and design a novel methodology to determine when a sensor requires calibration. Our method will eliminate the need to set hard thresholds, and instead will be able to “learn” the trends of the data - both from past measurements and from multiple sensors observing similar phenomena. Specifically, our QA will be based on comparing measured data to its local prediction. Similarly, instead of setting a hard calibration period, we will approve sensors' calibration by comparing the long-term behavior of different sensors to external sensors using autonomous underwater vehicles.

In this project we propose to develop a technological method to monitor the propagation and breaking of Internal Waves (IWs) along the continental slopes of the Israeli Mediterranean and the northern Gulf of Eilat. IWs are usually generated by the barotropic tide and are characterized by a displacement of an isodensity surface whose amplitude is strongly dependent on the density difference between the water masses that it separates. IWs are relatively easily observed in pycnoclines and density discontinuities when followed over time. Previous observations have shown that IWs can reach amplitudes on the order of 100m at a depth of 250m, and can propagate at 1 m/s. As a result of interaction of IWs with a sloping sea bottom they can break and cause vertical mixing, which has been considered to be an important mechanism for heat and nutrient transport in the oceans. Thus, in the stratified water column, IW breaking could transport nutrients through the thermocline from the nutrient replete deep waters into the nutrient poor euphotic zone supporting primary production there.

In this project, we aim to predict ocean current in small areas using only a few drifting devices. Our approach will combine water current measurements from floaters and a physical water current model. Assuming the water current is spatially and timely correlated, we will evaluate the connection between the floaters' time-varying location and their drifting velocity to also predict water current velocity in other locations. We will consider two statistical approaches: a structural model to calculate the velocity field parameters that connects the geographical location with drifting velocity, and a virtual regression approach to predict the water current using a machine learning classifier. The two methods will be evaluated in multiple sea and oceans experiments. The developed statistical model will be integrated with a small-scale ocean current model to improve prediction.

In this project, we aim to develop a proof-of-concept system for detection of submerged marine animals that can be operated from the same surveying vessel during its operation. Our system will employ a train of active high frequency wideband acoustic emissions at power levels, which are safe for marine animals, to form a time-distance matrix of recorder reflections. Using novel machine-learning pattern recognition techniques and a tailor-made clustering algorithm, we will identify mobile targets and differentiate them from clutter and static reflectors. Independent verification from Doppler shift measurements and size estimation will greatly reduce false alarms. The system will be made cost-efficient and will allow realtime detection. Demonstrations in multiple sea experiments using real animals for ground truth information will verify the practicality of the system.

Project CETI (Cetacean Translation Initiative) is an interdisciplinary research initiative that sets the goal of developing and applying industrial-grade machine learning technology to collect a large-scale longitudinal bioacoustic and behavioral dataset that is amenable to automatic processing and analysis by modern machine learning pipelines. Physeter macrocephalus (sperm whales) are the model study subject in this endeavour due to their highly-developed neuroanatomical features, cognitive abilities, social structures, and complex communication -- among the most sophisticated in the animal kingdom. This paper details the scientific and technological roadmap toward this goal. We outline the key elements required for the collection and processing of massive bioacoustic data of sperm whales, detecting their basic communication units, higher-level structures, long-distance dependencies, and language-like features such as recursion and displacement. This approach builds upon recent advances at the intersection of robotics, machine learning, natural language processing, marine biology, and linguistics.

We propose an adaptive approach to pre-set the modulation scheme for LR-UWAC. This is a channel classification approach which, based on environmental information and on prior training on various channel types, predicts the best modulation scheme for the expected channel. Our classification procedure is trained to identify the channel's important features. Thus, compared to a direct decision approach, it becomes less sensitive to possible mismatches of environmental information. Our numerical simulation and sea experiment show that our approach successfully identifies the best modulation scheme based on the environmental information - even when the information is biased or only partially available.

The main goal of our project is to quantify the impacts of shipping URN as a widespread and long-term disturbance to the behavioral of Bottlenose dolphin (Tursiops truncatus) and Common Dolphin (Delphinus delphis).