Multispectral Detection of Planktothrix rubescens Blooms in Volcanic Lakes Using Sentinel-2: Proof of Concept of a Newly Characterized Spectral Signature

Source abstract

Cyanobacterial harmful algal blooms threaten freshwater quality, ecosystem stability, and public health, creating a need for spatially extensive monitoring tools that complement in situ observations. This study evaluates whether Sentinel-2 multispectral imagery can be used to identify a recurrent spectral pattern associated with Planktothrix rubescens blooms in two volcanic lakes in Central Italy (Albano and Vico). Sentinel-2 Level-1C data were processed with C2RCC to retrieve water-leaving reflectance and compared with microscopy-based observations across 13 acquisition dates (eight bloom and five no-bloom events). At the event level, bloom and no-bloom conditions were spectrally distinguishable. Six of eight bloom spectra showed a green-band peak at 560 nm and a pronounced increase from blue (443–490 nm) to green wavelengths. The B3/B1 ratio was 3.04 under bloom conditions versus 1.12 under no-bloom conditions, while B3/B2 was 1.77 versus 0.76 (both p < 0.05). Within bloom events, the blue-to-green ratio was not significantly correlated with cell density, indicating that the pattern is more suitable for bloom detection than abundance estimation. The recurrent pattern was formalized as a simple Type A criterion and further assessed using cosine similarity. These results support Sentinel-2 as a proof-of-concept screening tool for P. rubescens-dominated blooms, pending broader validation.