ACT-AP Mobile Monitoring

1       About This Document

This directory contains air pollution concentration data from the Adult Changes in Thought – Traffic-Related Air Pollution (ACT-TRAP) mobile monitoring study (Blanco et al., 2022).

2       Data Source

The ACT-TRAP mobile monitoring study was an air pollution monitoring campaign that was conducted between March 2019 and March 2020 in the greater Seattle area. Briefly, the campaign collected 2-minute pollutant concentrations at 309 stop locations throughout the greater Seattle area. Pollutants were simultaneously measured with high temporal resolution (measurements every 1-60 sec). These included particle number concentration (PNC, an indicator of ultrafine particulates or UFP) from four different instruments, black carbon (BC), nitrogen dioxide (NO2), carbon dioxide (CO2), and fine particulate matter (PM2.5). Each stop location was visited approximately 29 times during all seasons and days of the week between the hours of approximately 5 AM and 11 PM.

3       The Data

We summarized high-resolution instrument data into median (and mean) stop concentrations, and these were used to calculate annual average concentrations for each of the 309 mobile monitoring sites. Using universal kriging-partial least squares (UK-PLS) regression models with hundreds of geographic covariate predictors, we generated out-of-sample TRAP predictions for monitoring locations, cohort locations (for epidemiologic purposes), census block centroids, and a grid (for visual purposes).

The relevant data files are:

1.     Stop-level data (stop_data.csv)

2.     Annual average estimates and predictions for the mobile monitoring sites (N=309) (monitoring_location_estimates_and_predictions.csv)

3.     Annual average model predictions for:

a.     a grid in the region (grid_predictions.csv)

b.     2010 Census block centroids for Washington state (census_block_predictions.csv)

4.     The geographic covariates available for modeling for the:  

a.     monitoring sites (dr0311_mobile_covars.csv)

b.     grid (dr0311_grid_covars.csv)

c.     2010 census block covariates (block10_intpts_wa.csv)

5.     UK-PLS model performances for annual average TRAP from the mobile monitoring campaign (model_performances.csv).  

6.     Spatial files for

a.     A polygon encompassing all of the monitoring sites. There are three versions:

                                               i.     The first includes all land and water areas (monitoring_area.rda) and is the simplest.

                                             ii.     Same as above but excludes major water areas (monitoring_land.rda). This is useful for mapping, for example.

                                            iii.     Same as above but excludes all water areas (monitoring_land_zero_water.rda). This is the file used to make predictions to ensure that no predictions are made on bodies of water.

On-road data were also collected while the vehicle was in motion. Please inquire if you are interested in these data.

4       Data Dictionaries

Data dictionaries for all the campaign data can be found in data_dictionaries.docx.

Geographic covariates come from the MESA Air geodatabase. Details on these covariates can be found in MESAAirDOOP_20190501.pdf and here: https://deohs.washington.edu/sites/default/files/MESAAirDOOP_Rev12.pdf.

5       Reference

Blanco MN, Gassett A, Gould T, Doubleday A, Slager DL, Austin E, Seto E, Larson TV, Marshall JD, Sheppard L. Characterization of Annual Average Traffic-Related Air Pollution Concentrations in the Greater Seattle Area from a Year-Long Mobile Monitoring Campaign. Environ Sci Technol. 2022 Aug 16;56(16):11460-11472. doi: 10.1021/acs.est.2c01077. Epub 2022 Aug 2. PMID: 35917479; PMCID: PMC9396693.