RO-Crate Bot

2025-11-10T14:55:33.954Z

RO-Crate Bot

2025-11-10T14:55:33.954Z

RO-Crate Bot

2025-11-10T14:55:33.954Z

RO-Crate Bot

2025-11-10T14:55:33.954Z

RO-Crate Bot

2025-11-10T14:55:33.954Z

RO-Crate Bot

2025-11-10T14:55:33.954Z

Fish Track mini-project

RO-Crate Bot

2025-11-10T14:55:33.954Z

RO-Crate Bot

2025-11-10T14:55:33.954Z

# Description **Duration**: 6 months **Start date**: September 2024. **Partners**: Jean-Marc Delouis (CNRS/IFREMER), Tina Odaka (IFREMER), Anne Fouilloux (Simula), Quentin Mazouni (Simula) ## Context Biologing seeks to determine the most likely fish trajectory using tagging data that records only temperature and pressure. Understanding the routes of fish is crucial for various ecological reasons to enhance the protection of the species under study. While biologing has demonstrated effectiveness [de Pontual et al, 2023], there remains scope for further improvement. Specifically, in calculating the likeliest path, the pangeo-fish algorithm models the fish's movement capability using a normal distribution, indicating that there is no favoured direction. We suggest incorporating external data to refine our understanding of the fish's most likely movement, including bathymetric data, identification of the fish through behavioural biology insights (such as depth data at high frequency) or detailed sea surface temperature measurements, among others. ## Detailed work - Month 1 - Construct the input data sets: During the first month, efforts will focus on assembling an appropriate data set for research. The initial task involves analysing the available biological dataset from the existing 450 sets and applying unsupervised classification techniques (e.g., Scattering Transform) to categorise fish behaviours. In parallel, satellite data (Sea Surface Temperature) and ocean modelling systems such as Copernicus Marine must be identified and collected. - Months 2 to 4 - Estimate a new probable trajectory for fish: This phase involves the employing cutting-edge algorithms that use scale coupling statistics methods. This approach aims to achieve highly resolved (both spatially and temporally) data to predict the direction of fish movement. The feasibility of validating this novel method is supported by multiple observations of some fish within a few tracks. Thus, disregarding this existing data, we can assess the deviation of the fish paths predicted by these algorithms from observed locations. - Months 5 to 6 - Apply the algorithm to the 450 tags and evaluate the performances: in the final two months, apply the developed algorithm to the 450 tags and assess its efficacy. If the results are satisfactory, a scientific publication can be expected at the end of this period. ## Computing & storage infrastructure Cloud resources (storage & computing) are available for the entire duration of the project. These include 1) GFTS Jupyterhub (http://gfts.minrk.net) to conduct computation tasks like applying algorithms and constructing datasets. All the data, inclusive of tagging and satellite data, will be securely stored and managed via S3-compatible object storage to facilitate access and reuse during the project. Access to GFTS JupyterHub will be granted at the start of the project. 2) Destination Earth Services (https://platform.destine.eu/) to access environmental data like Ocean data (Sea Surface Temperature, Sea Water Potential Temperature on model levels, Surface Temperature and bathymetric data) from the Climate Adaptation Digital twin; Access to the Destination Earth Services require user registration at https://platform.destine.eu/ and is subject to their availability during the project (services are still under development and not available yet).

RO-Crate Bot

2025-11-10T14:55:33.954Z

RO-Crate Bot

2025-11-10T14:55:33.954Z