Enhancing shallow water quality monitoring efficiency with deep learning and remote sensing: A case study in Mar Menor

Abstract

Satellite remote sensing technology has proven effective in monitoring various environmental parameters, but its efficiency in assessing shallow lakes has been limited. This study applies state-of-the-art machine and deep learning algorithms supported by classical statistic methods to analyze remote sensing data to measure chlorophyll-a (Chl-a) concentration levels. Focused on a shallow coastal lagoon, Mar Menor, this work analyzes statistically daily Sentinel 3 information behaviour and compares Machine Learning and Deep Learning techniques to enhance efficiency and accuracy data of this satellite. Convolutional Neural Networks (CNNs) stand out as a robust choice, capable of delivering excellent results even in the presence of anomalous events. Our findings demonstrate that the CNN-based approach directly utilizing satellite data yields promising results in monitoring shallow lakes, offering enhanced efficiency and robustness. This research contributes to optimizing remote sensing data to and produce a continuous information flow addressed to monitoring shallow aquatic ecosystems with potential environmental management and conservation applications.