Initial Assessment of the Detectability of Cyanotoxin Occurrence Using PACE Satellite Observations

Microcystin is a major public health hazard during freshwater cyanobacterial blooms, yet its remote detection remains challenging because it lacks any direct optical absorption features. We demonstrate that hyperspectral observations from NASA's PACE Ocean Color Instrument (OCI) can be used to detect elevated microcystin concentrations by exploiting bloom-related optical signatures. Two years of NOAA in situ microcystin measurements from the Western Lake Erie Basin are analyzed using bootstrapped Welch's t-tests to identify OCI wavelengths sensitive to high-microcystin conditions. These bands and derived spectral-shape metrics are then used to train a nested cross-validated random forest classifier, achieving 89.5% accuracy despite limited training data. The results reproduce expected spatial and seasonal patterns of toxic blooms and highlight the dominant predictive role of spectral-shape metrics in the pigment absorption and red-edge regions.