Central Asian aerosol size and vertical regimes track the seasons rather than monthly stability anomalies: data, outputs and code

Processed research data, analysis outputs and code for the study “Central Asian aerosol size and vertical regimes track the seasons rather than monthly stability anomalies”. This is the data and code archive; the manuscript and Supplementary Information themselves are available through the journal. Version v11 (revised manuscript, September 2026). Supersedes v10. Adds daily (synoptic) and extreme-month tests of the stability-fine-fraction null (Dushanbe AERONET SDA Level 2.0 daily averages in Data_aeronet_daily.zip; analysis script with the plan fixed in its docstring; outputs; SI Table S31), a table of HYSPLIT trajectories shortened at the top of the data domain, and consistency fixes. Other input data archives are unchanged. Version v10 (revised manuscript, September 2026). Supersedes v9. Adds a MISR test of the central result over the western dust basins (MIL3MAEN V004 domain subset in Data_misr.zip, analysis script with the plan fixed in its docstring, outputs; SI Table S30); corrects the normalisation of the HYSPLIT residence maps (Fig. 5a-c); and adds a full-duration trajectory sensitivity to the HYSPLIT class contrasts. Other input data archives are unchanged. Version v9 (revised manuscript, September 2026). Supersedes v8. Adds an independent-lidar check of the CALIOP vertical contrast with the CATS-ISS Level 2 5 km layer product (Mode 7.2, V3-01, March 2015 to October 2017): the analysis script (plan fixed in its docstring before the data were analysed), monthly and summary outputs, granule log and the granule manifest (Data_cats.zip). The CATS HDF5 granules themselves (about 66 GB) are public at the NASA ASDC and are not archived here. Other input data archives are unchanged from v6. Version v8 (revised manuscript, September 2026). Supersedes v7. The HYSPLIT back-trajectory ensemble had been run on an earlier ranking of event months; the ten month-class slots that differed from the reported top-10 event months (810 trajectories) were rerun with the same HYSPLIT build (reproducing existing trajectories exactly), the superseded runs were removed, and the attribution tables, figures and a month-resampled class-contrast table were regenerated. Input data archives are unchanged from v6. Version v7 (revised manuscript, September 2026). Supersedes v6. IGRA2 925 and 700 hPa levels are now matched on the exact standard pressures (previously pressures were rounded to whole hectopascals, so a few non-standard levels could replace the standard level); all stability-dependent outputs were regenerated, which changes two monthly index values and reclassifies December 2021. The fourteen-station wild-cluster bootstrap now enumerates all 214 sign vectors, the extended-network rerun records the reclassified and replaced months, and the PM2.5 total-improvement bootstrap table is produced by the pipeline. Input data archives are unchanged from v6. Version v6 (revised manuscript, September 2026). Accompanies the revision of the manuscript. Adds terrain-resolved per-cell MERRA-2 three-dimensional dust metrics, every IGRA2 station in the study domain reporting after 2010 and IGRA2 sounding-derived mixed-layer heights for the index stations, corrected statistical code (adjacent-month AR(1) effective sample size, year-block bootstrap intervals, exact wild-cluster bootstrap), and the new revision outputs (Supplementary Tables S19–S28). The superseded MERRA-2 layer-profile table built from regionally averaged fields is not included. Contents (structured ZIP archives that unpack to the repository tree): Data_*.zip observational and model source products (IGRA radiosondes, AERONET, ERA5, MERRA-2 column and 3-D dust, CALIOP, CERES, CAMS, MODIS, OpenAQ, GEOS-Chem tagged-tracer output, health/population baselines); outputs.zip regenerated tables, figures and HYSPLIT/GEOS-Chem run artifacts; scripts.zip the full analysis and figure code. README.md gives step-by-step reproduction instructions and requirements.txt the Python dependencies. To reproduce: unzip Data_*.zip, outputs.zip and scripts.zip into one directory (reconstructing the Data/, outputs/, scripts/ tree), install Python 3.10+ with pip install -r requirements.txt, and run the ordered commands in README.md. The manuscript tables and figures regenerate from the archived Data/ into outputs/. Deliberately excluded (large, re-downloadable third-party resources, not research data): the software environments under Data/runtime/ (libRadtran, SNICAR/BioSNICAR, HYSPLIT binaries, compiled toolchains) and the ~48.5 GB NOAA GDAS1 meteorology that drives the HYSPLIT ensemble. These are only needed to re-run the raw trajectory and radiative-transfer computations from scratch; the intermediate HYSPLIT and GEOS-Chem outputs needed to regenerate the figures are included in outputs/. Rebuild the environments locally and download the public GDAS1 archive from NOAA ARL (READY/ARL FTP) into Data/hysplit/met/gdas1/. Source code repository: github.com/ProgrmerJack/Stratospheric-science-Analysis.