EDGAR Emissions

This article demonstrates a compact workflow for EDGAR emissions surfaces supported by amadeus.

This vignette runs its live workflow when rendered locally. The heavy download, processing, extraction, and plotting chunks are skipped automatically on CI, CRAN checks, and pkgdown builds; set AMADEUS_RUN_VIGNETTES=true to force live execution in those environments.

Each workflow uses two small example surfaces: example_points_sf, a saved subset of AQS monitor locations from tests/testdata/aqs/aqs-location-sample.rds, for point extraction; and the packaged Durham County Uber H3 resolution-8 hexagons at system.file("extdata", "data_files", "durham_h3_res8.rds", package = "amadeus") for polygon extraction.

Available EDGAR variables and information

The species argument in download_data(dataset_name = "edgar", ...) currently supports the following EDGAR air-pollutant emissions variables. For the emissions downloads used in this vignette, values are reported in tonnes.

amadeus variable name	short description
BC	Black carbon (carbonaceous particulate fraction)
CO	Carbon monoxide
NH3	Ammonia
NMVOC	Non-methane volatile organic compounds
NOx	Nitrogen oxides
OC	Organic carbon (carbonaceous particulate fraction)
PM10	Particulate matter <= 10 um aerodynamic diameter
PM2.5	Particulate matter <= 2.5 um aerodynamic diameter
SO2	Sulfur dioxide

download_edgar() also exposes several other useful switches that determine which EDGAR products are available:

version = "8.1" covers the standard EDGAR air-pollutant products shown in this vignette.
version = "8.1_voc" switches to VOC speciation products instead of the standard pollutant grids.
temp_res supports "yearly", "monthly", and "timeseries" for version = "8.1". VOC speciation downloads do not use temp_res.
output supports "emi" for emissions and "flx" for flux products.
format is typically "nc" or "txt", but monthly outputs and flux outputs are only available in NetCDF form.
year_range can be a single year or a range. Yearly EDGAR products span 1970-2022, while monthly and VOC-speciation products span 2000-2022.

The sector arguments depend on which EDGAR product family you request:

input	supported values	when used
sector_yearly	AGS, AWB, CHE, ENE, IND, MNM, NMM, PRU_SOL, RCO, REF_TRF, SWD_INC, SWD_LDF, TNR_Aviation_CDS, TNR_Aviation_CRS, TNR_Aviation_LTO, TNR_Aviation_SPS, TNR_Other, TNR_Ship, TRO, WWT	Yearly standard EDGAR products (`version = "8.1"`)
sector_monthly	AGRICULTURE, BUILDINGS, FUEL_EXPLOITATION, IND_COMBUSTION, IND_PROCESSES, POWER_INDUSTRY, TRANSPORT, WASTE	Monthly standard EDGAR products (`version = "8.1"`)
sector_voc	AGRICULTURE, BUILDINGS, FUEL_EXPLOITATION, IND_COMBUSTION, IND_PROCESSES, POWER_INDUSTRY, TRANSPORT, WASTE	VOC speciation products (`version = "8.1_voc"`)
voc	Integers 1-25 for VOC speciation groups	VOC speciation products (`version = "8.1_voc"`)

EDGAR groups these variables broadly as ozone precursors (CO, NOx, NMVOC), acidifying gases (NH3, NOx, SO2), and primary particulates (PM10, PM2.5, BC, OC).

Download representative requests

directory_to_save <- file.path(tempdir(), "edgar_workflow")
download_data(
  dataset_name = "edgar",
  species = c("CO", "NOx"),
  temp_res = "yearly",
  sector_yearly = "ENE",
  year_range = 2021,
  directory_to_save = directory_to_save,
  acknowledgement = TRUE,
  unzip = TRUE,
  remove_zip = FALSE
)

Process one workflow-ready data product

Amadeus process functions returns a terra::SpatRaster so we can apply the terra extent option on the bounding box of the points to reduce processing time.

processed_data <- process_covariates(
  covariate = "edgar",
  path = list.files(
    directory_to_save,
    pattern = "\\.(tif|tiff|nc4?|grd|img)$",
    recursive = TRUE,
    full.names = TRUE
  ),
  extent = sf::st_bbox(example_points_sf)
)

Calculate covariates at points

We start with a radius of 0, which extracts the raster gric cell value at each point with no smoothing or averaging over an area.

point_values <- calculate_covariates(
  covariate = "edgar",
  from = processed_data,
  locs = example_points_sf,
  locs_id = "site_id",
  radius = 0,
  geom = "sf"
)

print(point_values)

At point-level extractions, we often see more NA because data has to overlap exactly. Here, we smooth out the covariate calculation by using a buffer radius.

point_values_1km <- calculate_covariates(
  covariate = "edgar",
  from = processed_data,
  locs = example_points_sf,
  locs_id = "site_id",
  radius = 1000, # distance units are in meters, see crs(processed_data)
  geom = "sf"
)

print(point_values_1km)

point_values_10km <- calculate_covariates(
  covariate = "edgar",
  from = processed_data,
  locs = example_points_sf,
  locs_id = "site_id",
  radius = 10000,
  geom = "sf"
)

print(point_values_10km)

Calculate covariates across Durham County H3 hexagons

polygon_values <- calculate_covariates(
  covariate = "edgar",
  from = processed_data,
  locs = durham_hex,
  locs_id = "h3_id",
  radius = 0,
  geom = "sf"
)

Visualize the point outputs

plot_points(point_values, paste0("EDGAR Emissions", ": point extraction"))