AquaCubeAI enables faster onboard monitoring of coastal water turbidity
AquaCubeAI-Powered Monitoring Turbidity on-board Φsat-2
Computer Vision and Pattern Recognition
Summary
Monitoring how clear or muddy coastal waters are is important for protecting the environment. Existing satellite methods take time because they send images back to Earth for processing. The authors developed AquaCubeAI, a small and efficient AI model that runs directly on a satellite to measure water turbidity faster. They trained and tested the model using simulated satellite images matched with real water quality data from European coastal areas. This approach can provide quicker alerts about changes in water quality using low-power hardware on satellites.
What this means in practice
- •For environmental monitoring teams: Deliver low-latency turbidity maps from satellites to improve coastal water quality surveillance.
- •For satellite system engineers: Implement lightweight AI models on spacecraft hardware to speed up onboard environmental data analysis.
Authors
Pietro Di Stasio, Francesca Razzano, Elisa Liparulo, Gabriele Meoni, Nicolas Longépé, Deodato Tapete, Paolo Gamba, Gilda Schirinzi, Silvia Liberata Ullo
Abstract
Timely monitoring of coastal water quality is critical for environmental protection, yet conventional satellite workflows rely on downlink and ground processing, introducing latency that can limit responsiveness to rapidly evolving turbidity events. To address this limitation, we propose AquaCubeAI, a lightweight machine-learning approach for onboard estimation of coastal water turbidity from Φsat-2 multispectral imagery. By shifting inference from the ground segment to the satellite, AquaCubeAI aims to enable lower-latency, more responsive, and more operationally useful turbidity monitoring under the strict compute and bandwidth constraints of spaceborne platforms. The model is trained on simulated Φsat-2 acquisitions spatially aligned with Copernicus Marine Service (CMEMS) High-Resolution Ocean Color (HR-OC) turbidity products over selected localized coastal sites spanning four European marine macro-regions. To provide a realistic evaluation of generalization in the presence of spatial correlation, we adopt a spatial block splitting protocol that mitigates data leakage between training and evaluation subsets. The main contributions of this work are: (i) a scalable dataset generation pipeline pairing simulated Φsat-2 multispectral patches with CMEMS HR-OC turbidity labels across selected localized European coastal sites; (ii) a compact Multi-Layer Perceptron (MLP)-based turbidity regressor trained under a leakage-aware geospatial split and tailored to embedded constraints; and (iii) a reformulation for dense spatial prediction via parameter sharing, enabling turbidity mapping and simple threshold-based anomaly masks for onboard decision logic. Embedded deployment on an Intel Myriad Vision Processing Unit (VPU) further confirms the feasibility of low-power hardware and supports low-latency inference from multispectral inputs.