Efficient Detection of Floating Algal Blooms Using Sentinel-2 Imagery: The Introduction of the SFABI Index

Abstract

Algal blooms are among the most serious challenges affecting inland waters, disrupting ecosystems, degrading water quality, and posing risks to human activities. Developing reliable monitoring and mapping methods is crucial for mitigating their harmful impact. This study introduces the Sentinel-2 Floating Algal Bloom Index (SFABI), designed for detecting and mapping algal blooms at varying densities. Lake Burdur was selected as the study area for this research. Sentinel-2 images from three different dates were used as the primary data source. As a pre-processing step, Sentinel-2 Level-1C images were converted to bottom-of-atmosphere reflectance values by applying the iCOR atmospheric correction technique. Subsequently, statistical analysis was conducted to compare the spectral separability of the bands based on the sample pixels. The pixels were categorized into three classes: low-and high-density algal blooms, and water. Based on the results, the proposed index was developed based on the vegetation red-edge (B06 and B07) and near-infrared (B08) with the highest average M-statistic values and the visible-region (B02 and B03) and short-wave infrared (B12) with one the lowest M-statistic values. Furthermore, three thresholding techniques were utilized and evaluated to automatically create thematic maps representing water and algae from the grey-level index maps. The accuracy of each SAFABI map, classified using a specific single threshold value, was evaluated based on the F-score metric. To ensure an objective evaluation, two additional spectral indices specifically designed for detecting algal blooms, namely, the Floating Algae Index (FAI) and the Adjusted Floating Algae Index (AFAI), were also applied, and their classified maps were thoroughly analysed and compared. The results showed that the SFABI achieved an F-Score of over 97% across all three datasets, significantly surpassing the performance of other indices, which remained under 70%. Additionally, the SFABI index achieved F-Score values of about 90% in detecting low-density algal blooms. This demonstrates the effectiveness of the proposed index in identifying low-density blooms, which are often overlooked in algal bloom analyses, even when using a single threshold value. © 2025 Elsevier B.V., All rights reserved.