A Highly Accurate UV-Based Soil Organic Carbon Measurement System Enabled With the Measurement of Environmental Contributors and a Precise Superposition of Prediction Algorithms

This letter presents a UV-induced soil organic carbon (SOC) measurement system enhanced with an ensemble machine learning algorithm for environmental calibration. The system uses a 30-min UV exposure to extract CO<sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> via photo-oxidation and integrates temperature and moisture data to correct for environmental variability. A custom ensemble learning model composed of six algorithms processes the data to deliver highly accurate SOC predictions. Field validation of this system demonstrated a prediction accuracy of 93.95% with an R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> of 0.91, representing a 21.03% improvement over models lacking environmental calibration and underscoring the systems strong potential for real-time, in-situ carbon monitoring.