# Short-Term Machine-Learning Calibration of PID Sensors for Ambient VOC OH Reactivity

## TL;DR

This paper introduces a machine-learning method to improve the accuracy of PID sensors for measuring VOC reactivity in real-world conditions.

## Contribution

A rapid, time-aware machine-learning calibration workflow is proposed to enhance PID sensor reliability for VOC OH reactivity estimation.

## Key findings

- XGBoost models achieved strong agreement with PTR-derived VOC OH reactivity (Pearson’s r = 0.85, R2 = 0.64).
- The method improved inter-sensor consistency and supports harmonization of PID networks for reactivity monitoring.
- Time-aware validation ensured realistic generalization under temporal autocorrelation.

## Abstract

Photoionization detector (PID) sensors are widely used for ambient Volatile organic compound (VOC) monitoring because they are inexpensive, flexible, and fast. However, PID outputs are strongly influenced by environmental conditions (especially temperature and relative humidity) and exhibit substantial inter-sensor variability, limiting their quantitative reliability. Here we present a rapid machine-learning calibration workflow that maps PID signals and meteorological covariates to a photochemically relevant reference metric, PTR-derived VOC OH reactivity (ROH,PTR, s−1), calculated from online PTR-ToF-MS VOC measurements weighted by OH reaction rate constants. Four MiniPID sensors were co-located with a PTR-ToF-MS and a thermohygrometer, and data were harmonized to 10-s resolution. Multiple regression models were evaluated, with ensemble methods (RF and XGBoost) providing the best overall performance. To ensure realistic generalization under temporal autocorrelation, validation used a time-aware split: models were trained on a contiguous 24-h co-location period and evaluated on subsequent days (out-of-time). In this out-of-time evaluation, XGBoost achieved strong agreement with ROH,PTR across sensors (Pearson’s r = 0.85, R2 = 0.64, RMSE = 1.74 s−1), while substantially improving inter-sensor consistency. This short-duration calibration approach supports practical co-location-based harmonization of PID networks for high-temporal-resolution VOC reactivity monitoring in urban and industrial environments.