Predictive mapping of harmful algal blooms with Sentinel-2 imagery and random forest modelling

• Sentinel-2 satellite is used for reservoir monitoring. • Models based on the Random Forest algorithm are calculated. • Image algebra is used to obtain relationships between bands. • R 2 values > 0.90 are obtained in the Chlorophyll model. • The combination of TSI, BSI and BI models optimises the rapid detection of blooms. Climate change and anthropogenic nutrient enrichment are driving the proliferation of algal blooms worldwide, posing a significant threat to aquatic ecosystems, including urban reservoirs. As water security in these systems is critical, inadequate monitoring continues to contribute to water quality deterioration and the loss of vital ecosystem services. This study aims to develop early detection tools for harmful algal blooms using Sentinel-2 multispectral imagery, the Random Forest algorithm, and in situ data. Model performance was evaluated for Trophic State Index (TSI), Chlorophyll-a (Chl-a), and surface water temperature in urban reservoirs, with initial calibration and validation on the A Baxe reservoir, followed by external validation in As Forcadas (northwest Spain). Spectral indices – such as the Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI), Brightness Index (BI) – and specific bands were derived using Google Earth Engine (GEE). The developed models demonstrated high accuracy, with coefficients of determination (R 2 ) of 0.97 for Chl-a, > 0.80 for TSI, and > 0.70 for temperature. The TSI model’s relevance is underscored by its integration of various physicochemical parameters (N total , P total , Secchi Disk, Chl-a), and the BI index showed strong performance in estimating temperature despite Sentinel-2′s lack of a thermal band. This study proposes transferable methodologies for any reservoir, delivering an effective tool for the early detection of harmful algal blooms in stagnant waters, and contributing significantly to improved water resource management and the protection of aquatic ecosystems.

Authors

Publication Details

Journal
Water Security
Published
2025-12-27
DOI
https://doi.org/10.1016/j.wasec.2025.100194
Primary Topic
Aquatic Ecosystems and Phytoplankton Dynamics
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

Predictive mapping of harmful algal blooms with Sentinel-2 imagery and random forest modelling

Carolina Acuña-Alonso, Diego Barba-Barragáns, Xana Álvarez, Joaquín Martínez-Sánchez
Water Security
Aquatic Ecosystems and Phytoplankton Dynamics
article

Predictive mapping of harmful algal blooms with Sentinel-2 imagery and random forest modelling

Carolina Acuña-Alonso, Diego Barba-Barragáns, Xana Álvarez, Joaquín Martínez-Sánchez
article en

Abstract

• Sentinel-2 satellite is used for reservoir monitoring. • Models based on the Random Forest algorithm are calculated. • Image algebra is used to obtain relationships between bands. • R 2 values > 0.90 are obtained in the Chlorophyll model. • The combination of TSI, BSI and BI models optimises the rapid detection of blooms. Climate change and anthropogenic nutrient enrichment are driving the proliferation of algal blooms worldwide, posing a significant threat to aquatic ecosystems, including urban reservoirs. As water security in these systems is critical, inadequate monitoring continues to contribute to water quality deterioration and the loss of vital ecosystem services. This study aims to develop early detection tools for harmful algal blooms using Sentinel-2 multispectral imagery, the Random Forest algorithm, and in situ data. Model performance was evaluated for Trophic State Index (TSI), Chlorophyll-a (Chl-a), and surface water temperature in urban reservoirs, with initial calibration and validation on the A Baxe reservoir, followed by external validation in As Forcadas (northwest Spain). Spectral indices – such as the Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI), Brightness Index (BI) – and specific bands were derived using Google Earth Engine (GEE). The developed models demonstrated high accuracy, with coefficients of determination (R 2 ) of 0.97 for Chl-a, > 0.80 for TSI, and > 0.70 for temperature. The TSI model’s relevance is underscored by its integration of various physicochemical parameters (N total , P total , Secchi Disk, Chl-a), and the BI index showed strong performance in estimating temperature despite Sentinel-2′s lack of a thermal band. This study proposes transferable methodologies for any reservoir, delivering an effective tool for the early detection of harmful algal blooms in stagnant waters, and contributing significantly to improved water resource management and the protection of aquatic ecosystems.

Water SecurityVol. 27
Clean water and sanitation
Openalex Percentile: Top 38%
Aquatic Ecosystems and Phytoplankton Dynamics
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.