One monitor cannot referee a satellite: code and derived data for monitor requirements in 21 national networks

Analysis code, derived data and display-item scripts for One monitor cannot referee a satellite: monitor requirements in 21 national networks. Twenty-one national government reference networks, 7,353 stations in all, are thinned by Monte Carlo subsampling to between one and fifty stations, and the width of the resulting interval on each country's product-error statistic is tracked as it collapses. Two products are held separate throughout: a satellite machine-learning estimate calibrated to ground monitors (ACAG) and a reanalysis assimilating aerosol optical depth but no surface PM2.5 (MERRA-2). The deposit contains the station inventory, the resampling ladders, the band-stratified curves, the burden propagation, the null simulation, and the scripts that regenerate every figure and table in the manuscript and Supplementary Information. It also contains the verification suites: verify_display_items.py (134 checks against the built manuscript), verify_si.py (35 against the Supplementary Information), verify_cl.py (12 against the cover letter), verify_poison.py (10 retired values that must not reappear) and claim_inventory.py (every numeric token in the prose, traced or listed as unaccounted). Two of these read documents that are not deposited before publication and say so rather than reporting a pass. Keep Sufficiency_Pilot/ and Figures/ side by side. Every script resolves its inputs relative to its own location, so the deposit will not run flattened. MANIFEST.txt carries the md5 and byte size of every file. Third-party products (ACAG, MERRA-2, the WHO Ambient Air Quality Database and World Bank indicators) are cited in the manuscript and are not redistributed. See LICENSE.