Search

We analyzed velocity and hydrographic data from 23 moorings in the northeast Chukchi Sea from 2011 to 2014. In most years the eastern side of Hanna Shoal was strongly stratified year-round, while weakly stratified regions prevailed on the shelf south and west of the Shoal. Stratification differences cause differential vertical mixing rates, which in conjunction with advection of different bottom water properties resulted in seasonally-varying along-isobath density gradients. In agreement with numerical models, we find that bottom waters flow anticyclonically around the Shoal. Whereas most of the shelf responded barotropically to wind-forcing, there was a strong baroclinic component to the flow field northeast of Hanna Shoal, resulting in no net vertically-integrated transport on average. In contrast there is a net eastward transport from west of the Shoal, which implies convergence north of the Shoal. Convergence and along-isobath density gradients may foster cross-shelf exchange north of Hanna Shoal. Modal analyses indicate that the shelf south of the Shoal and Barrow Canyon responded coherently to local and remote winds, whereas the wind-current response around Hanna Shoal was less coherent. Barotropic topographic waves, of ~3-day period, were generated episodically northeast of the Shoal and propagate clockwise around Hanna Shoal, but are blocked from entering Barrow Canyon and are possibly scattered by the horizontally sheared flow and converging isobaths on the western side of the Shoal. Analysis of water properties on the western side of Hanna Shoal suggests that these include contributions from the western and southern portions of the Chukchi Sea.

Tracking the abundance of underwater species is crucial for understanding the effects of climate change on marine ecosystems. Biologists typically monitor underwater sites with echosounders and visualize data as 2D images (echograms); they interpret these data manually or semi-automatically, which is time-consuming and prone to inconsistencies. This paper proposes a deep learning framework for the automatic detection of schools of herring from echograms. Experiments demonstrated that our approach outperforms a traditional machine learning algorithm using hand-crafted features. Our framework could easily be expanded to detect more species of interest to sustainable fisheries., Comment: Accepted to NeurIPS 2019 workshop on Tackling Climate Change with Machine Learning, Vancouver, Canada