Comparison of Policy-Relevant Air Quality Metrics Calculated With Sparse In Situ Monitoring and Contiguous Satellite-Derived Data

Abstract

Compliance with the U.S. National Ambient Air Quality Standards (NAAQS) traditionally relies on a sparse network of ground-based monitors. We compare two methods for assessing whether U.S. counties are above or below the annual fine particulate matter (PM_2.5) NAAQS of 9.0 μg/m³: the monitor-based methodology used to calculate county design values (CDVs; an EPA metric representing the maximum, in-county PM_2.5 monitor value over a 3-year annual average), and a satellite-based methodology that computes county design value equivalents (CDVEs) from Washington University's global satellite-derived PM_2.5 data product. The satellite-based approach uses the 90th percentile PM_2.5 grid value in a county and shows strong overall agreement with monitor-based CDVs across 536 monitored U.S. counties (r = 0.76; r_s = 0.74). Counties were classified as aligned positives (APs), aligned negatives (ANs), non-aligned positives (NPs), or non-aligned negatives (NNs) based on whether CDVEs align or don't align with CDVs on NAAQS status. Risk factors likely contributing to larger differences between CDVEs and CDVs within non-aligned counties include ≤2 monitors, low (<7 μg/m³) or high (>10 μg/m³) CDVs, low (<0.05%) or high (>0.1%) monitor coverage, large county size, wildfires, mountains, deserts, and low urbanization. In non-aligned counties, differences grow with the number of risk factors, with seven-risk-factor counties showing 5x higher median differences than those with one. High error-risk counties cluster in the Western U.S.; low error-risk counties are in the Midwest and East. These findings highlight trade-offs between sparse and contiguous data for regulatory assessments and support integration of satellite-derived data into policy frameworks.