Rapid COD Sensing in Complex Surface Water Using Physicochemical-Informed Spectral Transformer with UV-Vis-SWNIR...

Area of Science:

• Environmental Science
• Analytical Chemistry
• Data Science

Background:

• Accurate water quality monitoring is crucial for sustainable resource management.
• Measuring chemical oxygen demand (COD) is vital for assessing water health.
• Existing spectroscopic methods face challenges in complex water environments due to limited spectral understanding and generalizability.

Purpose of the Study:

• To introduce a novel deep learning model for enhanced water quality sensing.
• To improve the accuracy and generalizability of chemical oxygen demand (COD) measurements in diverse water bodies.
• To integrate physicochemical knowledge into spectral analysis for domain adaptation.

Main Methods:

• Development of the physicochemical-informed spectral Transformer (PIST) model.
• Combination of PIST with ultraviolet-visible-shortwave-near-infrared (UV-vis-SWNIR) spectroscopy.
• Validation using extensive surface water spectral data from the Yangtze River and Poyang Lake.

Main Results:

• PIST achieved a high coefficient of determination (R^2) of 0.9008 for COD sensing.
• Significant reduction in root mean squared error (RMSE) by 45.20% and 29.38% compared to SVR and CNN models.
• Demonstrated notable accuracy and generalizability in complex water environments.

Conclusions:

• The PIST model represents a significant advancement in water quality sensing using spectroscopy and deep learning.
• The integration of physicochemical information enhances spectral encoding and domain adaptation.
• PIST offers a robust and accurate solution for rapid, large-scale water quality monitoring.