sweetapricity/bridgedeck-aqs
Bridgedeck — EPA AQS Air Pollution Annual Summaries (2000–2023) Part of the bridgedeck dataset collection for US bridge deck deterioration research. ⚠️ Pipeline dependency — This dataset contains pre-downloaded EPA AQS annual concentration files. It is shared to avoid re-downloading from EPA's servers. See sweetapricity/bridgedeck-nbi for the main bridge dataset. Overview Annual monitor-level concentration data for PM2.5 and NO2 from the EPA Air Quality System… See the full description on the dataset page: https://huggingface.co/datasets/sweetapricity/bridgedeck-aqs.
Bridgedeck — EPA AQS Air Pollution Annual Summaries (2000–2023)
Part of the bridgedeck dataset collection for US bridge deck deterioration research.
⚠️ Pipeline dependency — This dataset contains pre-downloaded EPA AQS annual concentration files. It is shared to avoid re-downloading from EPA's servers. See `sweetapricity/bridgedeck-nbi` for the main bridge dataset.
Overview
Annual monitor-level concentration data for PM2.5 and NO2 from the EPA Air Quality System (AQS), 2000–2023. Used to derive county-level air pollution exposure features for bridge deck deterioration modeling.
Air pollution accelerates concrete deterioration through:
- PM2.5 — fine particulate matter promotes carbonation and surface degradation
- NO2 — nitrogen dioxide contributes to acid deposition and rebar corrosion
Source
Source: EPA AirData Pre-Generated Files Files: annual_conc_by_monitor_{YEAR}.zip (24 files, 2000–2023) License: Public domain (US government data)
Contents
How It's Used
The bridgedeck ETL script src/bridgedeck/etl/aqs_clean.py:
- Reads each annual zip file
- Filters to PM2.5 (parameter code 88101) and NO2 (parameter code 42602)
- Aggregates multiple monitors per county to county-level annual means
- Joins to bridges on
(county_fips, year)— same pattern as NOAA
Output: data/interim/aqs/aqs_features.parquet — one row per (county_fips, year).
Features Derived
Note: Not all counties have monitors. Counties without monitors receive NaN. Coverage is higher in urban areas. ~50-60% of bridge-county-years have PM2.5 data, ~30-40% have NO2 data.