Remote detection of cyanobacteria through phycocyanin for water supply source using three-band model

• We detect cyanobacteria through phycocyanin (PC) using remote sensing (RS) data.
• We use three- and two-band indices to build models for PC estimation.
• The three-band model (TBM) shows good temporal transferability for PC estimation.
• The TBM with RS data has promising implications for cyanobacterial blooms monitoring.

Accurate remote estimation of phycocyanin (PC) in turbid productive waters is challenging because of their optical complexity. This study aims to test the performance of three-band models (TBM) and two-band models (TWM), as well as optimal band models (OBM) for spectral estimation of PC concentration. The Eagle Creek Reservoir is one of the major drinking and recreational water systems for the Indianapolis metropolitan area and has been strongly affected by algal, particularly cyanobacterial blooms. A total of 222 effective samples collected under a diverse range of algal bloom and optical conditions in 15 field campaigns during 2006, 2007, 2008, and 2010 were used for testing different models. The TBM, using wavebands at approximately 630, 710, and 720 nm, can explain 84%–93% of the PC concentration variability when applied to various datasets that were grouped in terms of different years or aggregated together. These TBMs can be used to estimate PC concentrations with RMSE ranging from 7.31 to 29.26 μg/L. The results for the aggregated original, narrow band spectra, simulated MEdium Resolution Imaging Spectrometer (MERIS) and Hyperion spectra further indicate that TBM, TWM, and OBM can be used to estimate PC concentration with acceptable accuracy (0.60 < R2 < 0.73). The TBM can be used to estimate PC concentrations with RMSE of 22.7, 31.5, and 28.3 μg/L, respectively, for the original narrow band, simulated MERIS, and simulated Hyperion spectra. The TBM and OBM exhibited similar accuracy, but outperformed TWM for PC estimation. When calibrated with large representative dataset, the TBM presents robust performance, which can be used as an effective tool for water resources managers to assess cyanobacterial blooms in water supply source.