Monitoring algal blooms and coastal water quality with satellite data for the assessment of climate-related hazard indicators

Bentivogli, Riccardo (2026) Monitoring algal blooms and coastal water quality with satellite data for the assessment of climate-related hazard indicators, [Dissertation thesis], Alma Mater Studiorum Università di Bologna. Dottorato di ricerca in Scienze della terra, della vita e dell'ambiente, 38 Ciclo. DOI 10.48676/unibo/amsdottorato/13270.
Documenti full-text disponibili:
[thumbnail of bentivogli_riccardo_tesi.pdf] Documento PDF (English) - Richiede un lettore di PDF come Xpdf o Adobe Acrobat Reader
Disponibile con Licenza: Creative Commons: Attribuzione - Non Commerciale 4.0 (CC BY-NC 4.0) .
Download (6MB)

Abstract

This thesis investigates the capabilities and limitations of optical satellite remote sensing for the characterization of phytoplankton and the retrieval of key water quality parameters in optically complex coastal and transitional waters. By deliberately combining hyperspectral and multispectral observations, the work bridges process-level optical understanding with operational monitoring constraints. The first part of the research employs hyperspectral laboratory measurements to characterize the optical signatures of phytoplankton species and functional groups typical of the northern Adriatic Sea. Under controlled conditions, narrow-band absorption features and subtle reflectance structures are shown to contain diagnostically relevant information. However, the study also demonstrates that theoretical spectral separability does not directly translate into operational detectability from space, as atmospheric effects, sensor limitations, and environmental variability reshape the information content of reflectance signals. A second component examines the impact of large-scale environmental perturbations on optical variability, highlighting the contextual nature of satellite-derived observations. Case-study analysis reveals how terrestrial inputs, suspended matter, and hydrodynamic changes modulate reflectance patterns, reinforcing the need to couple satellite data with in-situ measurements and hydro-biogeochemical interpretation. Finally, the thesis evaluates multispectral retrieval frameworks, including traditional algorithms and deep learning approaches, for chlorophyll-a and turbidity estimation. Results reveal parameterdependent asymmetries: turbidity shows robust retrievability, while chlorophyll-a remains more sensitive to spectral bandwidth and environmental regime. Deep learning models improve predictive stability but cannot compensate for limited spectral information. Overall, the work demonstrates that retrieval performance is governed by the interplay between intrinsic optical expressiveness, environmental variability, and sensor information content. The integration of hyperspectral knowledge, contextual environmental data, and physically consistent machine learning represents a realistic pathway toward more robust and ecosystem-aware aquatic monitoring.

Abstract
Tipologia del documento
Tesi di dottorato
Autore
Bentivogli, Riccardo
Supervisore
Co-supervisore
Dottorato di ricerca
Ciclo
38
Coordinatore
Settore disciplinare
Settore concorsuale
Parole chiave
remote sensing, harmful algal blooms, water quality, environmental monitoring, machine learning
DOI
10.48676/unibo/amsdottorato/13270
Data di discussione
9 Luglio 2026
URI

Altri metadati

Statistica sui download

Gestione del documento: Visualizza la tesi

^