Quantification of chlorophyll-a in typical lakes across China using Sentinel-2 MSI imagery with machine learning...

Area of Science:

• Environmental Science
• Remote Sensing
• Data Science

Background:

• Lake eutrophication is a growing concern, with Chlorophyll-a (Chl-a) as a key indicator of algal biomass.
• Accurate Chl-a estimation is crucial for monitoring water quality and ecosystem health.

Purpose of the Study:

• To develop and validate machine learning algorithms for estimating Chl-a concentrations in lakes using satellite imagery.
• To compare the performance of different machine learning models (LR, SVM, Catboost) against traditional algorithms.

Main Methods:

• Collected 273 lake samples across China (2017-2019) for Chl-a analysis.
• Utilized Multispectral Imager (MSI) data and the Case 2 Regional Coast Colour (CR2CC) processor to extract water-leaving reflectance spectra (Rrs(λ)).
• Applied K-means clustering to Rrs(λ) spectra and implemented machine learning models (LR, SVM, Catboost) for Chl-a retrieval.

Main Results:

• The Support Vector Machine (SVM) model demonstrated superior performance in estimating Chl-a concentrations, with high R² values during both calibration and validation.
• Traditional Chl-a algorithms yielded poor results compared to the developed machine learning models.
• K-means clustering revealed variations in retrieval performance based on water spectral characteristics, with SVM performing best on specific clusters.

Conclusions:

• Machine learning, especially SVM, provides a robust and accurate method for quantifying lake Chl-a concentrations from MSI imagery.
• This approach enables effective large-scale monitoring of lake eutrophication, particularly for medium to low Chl-a levels.
• AI-driven remote sensing of Chl-a offers a powerful tool for understanding lake ecosystem responses to global change.