Three-dimensional augmentation for hyperspectral image data of water quality: An Integrated approach using machine...

Area of Science:

• Environmental Science
• Remote Sensing
• Data Science

Background:

• Limited field water quality data hinders accurate monitoring.
• Hyperspectral image data (HSD) offers potential but requires enhancement for 3-D analysis.
• Chlorophyll-a (Chl-a) is a key indicator of aquatic ecosystem health.

Purpose of the Study:

• To develop a comprehensive methodology for 3-D augmentation of hyperspectral image data (HSD).
• To enhance the utility of HSD for accurate Chlorophyll-a (Chl-a) estimation.
• To integrate neural networks, numerical models, and machine learning for advanced water quality monitoring.

Main Methods:

• Augmented limited water quality data using Multilayer Perceptron (MLP) neural networks.
• Generated 3-D data by integrating MLP outputs with numerical models.
• Extended HSD into 3-D Chl-a data using ten machine learning models, including Gaussian Process Regression.

Main Results:

• MLP models successfully generated high-frequency water quality data and predicted detailed variables.
• Integrated numerical and neural network models effectively produced 3-D data.
• Gaussian Process Regression achieved superior 3-D Chl-a estimation accuracy (R-square = 0.99), aligning with algal bloom dynamics.

Conclusions:

• The integrated methodology successfully extends HSD using machine learning and numerical models.
• This approach offers a robust and efficient method for water quality monitoring and estimation.
• The findings facilitate better management of aquatic ecosystems through improved Chl-a prediction.