Rapid assessment of air pollution data quality using a composite index

This paper proposes a transparent Data Quality Index (DQI) for air-pollution datasets and demonstrates its application to Los Angeles PM_₂.₅ records from OpenAQ. The DQI aggregates five dimensions: accuracy, calibration, completeness, representational consistency, and timeliness. Each was scored on a 0-100 scale, and combined with explicit weights (30%, 20%, 20%, 15%, 15%) that can vary based on the stakeholders and context. Methods include hourly alignment (1-5 July 2025), Pearson correlation, bias and RMSE for accuracy, and ordinary least squares (slope/intercept) for calibration; completeness is computed against expected observations, representational consistency via outlier screening, and timeliness via last-update recency. A reference-grade monitor (ID 1948) and a low-cost sensor (ID 1037394) yielded different quality profiles: the reference stream scored DQI ≈99 (near-continuous coverage, stable representation, routine calibration, current delivery), whereas the sensor scored ≈58, driven by moderate accuracy, calibration bias, low completeness, extreme outliers during the 4 July episode, and zero timeliness after reporting ceased on 5 July. The contribution of this work is a field based rapid assessment methodology that preserves interpretability at the component level, while offering a concise composite signal of fitness-for-use, supporting decisions in research, regulatory, and public-health contexts.