2781. Hybrid AI Model for Water Quality Prediction Using Satellite and Drone Spectral Data
Invited abstract in session TD-60: DEA applications in Environment and Sustainability I, stream Data Envelopment Analysis and its applications.
Tuesday, 14:30-16:00Room: Western LT
Authors (first author is the speaker)
| 1. | hocine dai
|
| Artificial intelligence, University of BOUIRA | |
| 2. | Akli Abbas
|
| Computer Science Department, Bouira University |
Abstract
Water quality monitoring is a critical environmental challenge, requiring accurate and scalable assessment methods. Traditional in situ measurements are costly and limited in spatial coverage, while remote sensing offers a promising alternative. This study proposes a hybrid AI model that integrates satellite and drone-based hyperspectral/multispectral imagery with in situ physico-chemical water quality parameters (pH, turbidity, COD, conductivity, etc.) to enhance prediction accuracy. The methodology involves data fusion, where spectral reflectance from sensors such as Sentinel-2, Landsat-9, and PRISMA is combined with ground truth measurements from global water quality databases. We apply machine learning (Random Forest, XGBoost) and deep learning (CNN, Transformers) to predict key water quality indicators. Preliminary results demonstrate that super-resolution processing improves spectral data usability, and hybrid models outperform traditional statistical approaches. This work highlights the potential of AI-enhanced remote sensing for real-time, cost-effective water quality assessment, paving the way for automated monitoring systems at large scales.
Keywords: Water Quality Prediction, Hyperspectral Remote Sensing, Machine Learning, Deep Learning, AI, Sentinel-2, PRISMA, Super-Resolution, Data Fusion.
Keywords
- Machine Learning
- Water Management
- Artificial Intelligence
Status: accepted
Back to the list of papers