Evaluation of Atmospheric Preprocessing Methods and Chlorophyll Algorithms for Sentinel-2 Imagery in Coastal Waters

Area of Science:

• Environmental Science
• Remote Sensing
• Ecology

Background:

• Cyanobacterial blooms are increasing in estuaries and freshwater tidal rivers.
• Efficient remote sensing methods are crucial for monitoring these blooms and focusing on cyanobacteria.

Purpose of the Study:

• To compare remote sensing processing methods for accurate chlorophyll-a mapping.
• To identify optimal algorithms for predicting algal blooms and harmful algal bloom (HABs) risk.

Main Methods:

• Compared Acolite and Polymer atmospheric processing methods with Sentinel-2 imagery (2015-2022).
• Evaluated empirical (band ratio, spectral shape indices) and machine learning algorithms for chlorophyll-a prediction.
• Utilized paired chlorophyll-a observations from Chesapeake Bay and Indian River sites.

Main Results:

• Acolite processing yielded more observation points and a wider chlorophyll-a range than Polymer.
• Polymer showed better responsiveness at lower chlorophyll-a concentrations.
• The machine learning mixture density network (MDN) and Mishra band-ratio algorithms demonstrated superior performance in predicting chlorophyll-a, trophic state, and HABs risk.

Conclusions:

• The Mishra band-ratio and MDN algorithms are effective for mapping chlorophyll-a and assessing HABs risk in estuarine environments.
• Optimized remote sensing approaches enhance the detailed monitoring of cyanobacterial blooms.