Long-term stability from adaptive auto-calibration in mid-infrared sensing instruments

Mid-infrared laser-based sensing is widely used for the detection and quantification of trace gases for environmental monitoring, industrial process control, defense, and security applications. However, such laser spectroscopy-based field instruments require frequent calibration to correct for long-term drifts and precision. In many situations, such calibrations may be redundant and may not be effective in monitoring instrument errors or discriminating “real” environmental effects with instrument drifts. This paper shows an adaptive and auto-calibration methodology to account for drifts during field deployments and an adaptive and machine-learning-based method to validate environmental changes, e.g., plume detection, diurnal variations, and flux emissions. We show that physics and diffusion-based models can be developed to predict better gradual or rapid changes in density at given temperatures and appropriate source characterizations.