Hyperspectral remote sensing monitoring of cyanobacteria blooms in a large South American reservoir: High-and medium-spatial resolution satellite algorithm simulation

Abstract

We used hyperspectral remote sensing with the aim of establishing a monitoring program for cyanobacteria in South American reservoir. We sampled at a wide temporal (2012-2016, 10 seasons) and spatial (30km) gradient, and retrieved 111 field hyperspectral signatures, chlorophyll-a, cyanobacteria densities, and total suspended solids. The hyperspectral signatures at cyanobacteria dominated situations (n=75) were used for selecting the most suitable spectral bands in 7 high and medium spatial resolution satellites (Sentinel-2, Landsat 5, 7 and8, Spot 4/5 and6/7, WorldView 2), and for the development of chlorophyll and cyanobacteria cell abundance algorithms. The best performing chlorophyll algorithm was Sentinel 2 ((λ_560- λ_660+ λ_703)/(λ_560+ λ_660+ λ_703)) (R2 0.80), followed by WorldView 2 ((λ_550- λ_660+ λ_720)/(λ_550+ λ_660+ λ_720)) (R20.78), Landsat and SPOT series(λ_550- λ_650+ λ_800)/(λ_550+ λ_650+ λ_800) (R2 0.67-0.74). When these models were run for cyanobacteria abundance, the coefficient of determination remained rather similar, but the root mean square increased. This could affect the cyanobacteria cell abundance estimation in about 20%, yet it still allowed assessing the alert level categories for risk assessment. Our results highlight the importance of the red and near infrared region in identifying cyanobacteria in hypereutrophic waters, showed coherence with field cyanobacteria abundance, and allowed assessing bloom distribution in this ecosystem.