POSTER SESSION 1
15:30–17:30, Monday, September 14
Poster Session | 1 | 2 | 3 | 4 | Schedule at a Glance
ABSTRACT 1395 | POSTER 197
MULTIDIMENSIONAL DATA CUBES AND MACHINE LEARNING FOR MONITORING OF AQUACULTURE SITES IN THE SPANISH MEDITERRANEAN
This study demonstrates the integration of multidimensional satellite-derived data cubes, (e.g., Sentinel-2 MSI,Sentinel-3 OLCI, SLTR, reanalysis datasets), with advanced machine learning (ML) algorithms to establish a systematic framework for monitoring marine water quality. By utilizing data cubes, the methodology facilitates the precise calculation of critical environmental parameters over long time series. Key focus is placed on determining the concentrations of the water quality parameters, which serve as primary indicators for assessing health and predicting ecological shifts in coastal environments. The development is conducted in a phased manner: i) We use ensemble models, such as Gradient Boosting, to map complex non-linear relationships and evaluate physical parameters related to water turbidity. ii) To address the temporal aspect of water quality, Long Short-Term Memory (LSTM) networks will be implemented to predict long-term trends and identify anomalous discharges. iii) Furthermore, we want to integrate Convolutional Neural Networks (CNN) into the workflow to identify the morphology of organic plumes and other distinct geometric patterns in the water column that could affect the aquaculture sites. Algorithms are initially refined and validated within specific sub-cubes in the Valencian coastal areas (Spain). The processing pipeline is built upon the European DeepESDL (ESA) infrastructure, utilizing a cloud-based environment with Python-based tools, such as xarray, for efficient data handling.
Ana Belen Ruescas, Image Processing Laboratory-University of Valencia, [email protected], https://orcid.org/0000-0002-0843-7288
Arena Martin Olivo, Image Processing Laboratory, [email protected]
Elena Martínez-Mateo, Image Processing Laboratory, [email protected]
Poster Session | 1 | 2 | 3 | 4 | Schedule at a Glance
