Code and data for ES&T: An Unrecorded Reference State Moves Cities Across WHO Guideline Levels in Both Directions

Code and data deposit for the manuscript "An Unrecorded Reference State Moves Cities Across WHO Guideline Levels in Both Directions". A concentration in micrograms per cubic metre divides a mass by a volume, and the volume a given mass of air occupies depends on its temperature and pressure. The reference state is the temperature and pressure at which that volume is expressed. The WHO Ambient Air Quality Database records the concentration but not the reference state, so the same air can appear at two different numbers, and a city can sit on either side of a WHO guideline level depending on a convention that is written down nowhere in the record. This deposit reproduces every number in the paper. It contains four folders. code/ : every script, including the 32 analysis scripts named and checksummed in Supporting Information Table S24, the scripts that build the SI tables, and the script that assembled this deposit. derived/ : every intermediate and result file the scripts write, including the place-level tables, the validation tables and the results in JSON. inputs/ : small inputs that can be redistributed. figures/ : Figures 1 and 2 and the table of contents graphic, as the scripts produce them. Four files at the top level describe the rest. MANIFEST.csv lists every script, input and output with its size and SHA-256, and says whether the file travels in this deposit. CODE_CHECKSUMS_from_SI.csv reproduces SI Table S24 verbatim. SOURCE_DOCUMENTS_from_SI.csv reproduces SI Table S18 verbatim, giving for each jurisdiction the document read, what it says and how it was checked. DATA_DICTIONARY.md explains the fields. The raw inputs are not redistributed, because their publishers licence them separately. They are named, with their source and checksum, in the manifest and in the Supporting Information, so anyone can obtain the same files and verify them before rerunning. Requires Python 3.10 or later with pandas, numpy, matplotlib, geopandas, rasterio, openpyxl and python-docx.