Environmental Effects on NDIR-Based CH₄ Monitoring: Characterization and Correction.

Area of Science:

• Environmental Science
• Sensor Technology
• Data Science

Background:

• Nondispersive infrared (NDIR) sensors are crucial for environmental gas monitoring (CH4, CO2) due to their sensitivity, selectivity, and low cost.
• Sensor performance is significantly impacted by environmental variables like temperature and humidity, affecting detection accuracy.
• Accurate gas detection is essential for environmental monitoring and climate change studies.

Purpose of the Study:

• To characterize the impact of temperature and humidity on NDIR sensor performance for CH4 monitoring.
• To develop and validate machine learning models for correcting environmental signal biases in NDIR sensors.
• To provide a cost-effective solution for enhancing the accuracy and reliability of environmental gas sensors.

Main Methods:

• Laboratory experiments simulating environmental conditions (10-40 °C, 10-70% RH, 0-1000 ppm CO2) for CH4 monitoring.
• Application of machine learning regression algorithms to compensate for environmental influences on sensor signals.
• Field validation of the developed models at the Ito Natural Analogue Site (INAS).

Main Results:

• Significant signal variability was observed in NDIR sensors under varying temperature and humidity conditions.
• Machine learning models effectively mitigated signal biases caused by multiple environmental factors.
• The validated approach demonstrated improved accuracy and reliability in real-world environmental monitoring.

Conclusions:

• Machine learning-based compensation is a viable and cost-effective method to improve NDIR sensor accuracy in diverse environmental conditions.
• This approach reduces system complexity and operational costs for environmental gas monitoring.
• The study offers a pathway to more reliable and precise gas detection for environmental applications.