Regression Models Enhance Fluorescence Spectra for Smart Surface Water Surveillance

Area of Science:

• Environmental Chemistry
• Analytical Chemistry
• Spectroscopy

Background:

• Conventional fluorescence spectroscopy methods for aquatic dissolved organic matter analysis are limited.
• Advanced analytical toolkits are needed for real-time water monitoring and wastewater tracing.

Purpose of the Study:

• To develop easily implementable regression models for advanced fluorescence spectra analysis.
• To establish a correlation map between fluorescence EEMs and DOC.
• To identify pharmaceutical wastewater in surface water.

Main Methods:

• Weighted linear regression (WLR) was used to create a fluorescence intensity (FI) and dissolved organic carbon (DOC) correlation map.
• Multivariable linear regression (MLR) was developed to detect pharmaceutical wastewater.
• 191 surface water samples from diverse aquatic environments were analyzed.

Main Results:

• Humic-like FI at 300-380/440-490 nm excitation/emission wavelengths reliably indicates aquatic DOC.
• The MLR model achieved 10-20% fitting errors in identifying pharmaceutical wastewater, even with fluorescence quenching.
• Novel regression-based toolkits were established for fluorescence spectra analysis.

Conclusions:

• Regression-based toolkits advance the application of fluorescence spectroscopy in environmental protection.
• Developed methods offer improved capabilities for smart water surveillance.
• The study provides reliable indicators for aquatic DOC and tools for wastewater tracing.