AtlantiS will develop a set of interconnected digital capabilities building a set of open access reusable tools and approaches to support interoperability and integration with tools such as digital twins.

Data analytics: to facilitate large-scale data analysis and code sharing AtlantiS will establish a shared analysis environment through the JupyterHub system (supporting multiple languages, including Python, R, Matlab). We will modernise the approach to providing data in “analysis-ready” forms through the use of object store technology (e.g. using NERC’s JASMIN facility). Building on this infrastructure, we will develop and deploy two key pieces of open-source analysis software: (i) COAsT library to enable intercomparison of diverse ocean data sources, and (ii) AirSeaFluxCode, which calculates turbulent air-sea fluxes. AtlantiS will deploy a standardised set of APIs for accessing sensor data and greatly reducing duplication of effort across sensor data systems.

Artificial Intelligence (AI) applications to ocean data: AI, including Machine Learning (ML), provides immense opportunities across ocean science, although its full potential is still being developed. AtlantiS will focus on three main topics:

  1. Extraction of biologically-useful information from seabed imagery at PAP, through ML classification and automated detection, to maximize the use of images captured with AUVs and to develop approaches for seabed classification.
  2. ML-based early warning system for rapid North Atlantic biophysical changes as a proof of concept for how to prepare the multiple sustained observations and model data for ML applications.
  3. Assessing the implementation of automated AI-based systems for analysis of data from undersea cables, hybrid dynamical-ML modelling and for identifying water masses.

Data products: the reuse of data within and across communities is greatly facilitated by generating high-level data products that are easier to use than lower-level observations. Value can be added through regularizing the data (e.g. casting onto a regular grid), adding quality flags and improving documentation. This requires the creation of repeatable digital processing chains that can be run regularly. Our primary use case is the production of marine surface data (AirSeaFluxData).

Interactive and exploratory visualisation tools: AtlantiS will develop tools that put data at the fingertips of scientists, through interactive, intuitive, web-based interfaces. These exploratory tools will help scientists understand the characteristics of a range of datasets, assess their quality, look for interesting features and make rapid visual comparisons between data. The vision is to establish an easy-to-use, online system that enables users to browse ocean data arising from in situ observations, autonomous systems and model outputs, and interoperable with existing ocean data systems such as Copernicus Marine Services (CMEMS) to bring in third-party data. The data available through the tools will be as “live” and up to date as possible, requiring the development of near real-time processing steps. The primary use case will be rapid intercomparison of different data sources, supporting the generation of hypotheses and the detection of and response to events (e.g. marine heatwaves).