Fire where it burns, smoke where people breathe: a 26-year open-data reconstruction of burned area, population-weighted smoke exposure and compound haze-disaster risk in Kalimantan, Indonesia (2001-2026) - analysis code, derived data and verification log

Analysis code, derived data products, SHA-256 manifest and verification log supporting the article 'Fire where it burns, smoke where people breathe: a 26-year open-data reconstruction of burned area, population-weighted smoke exposure and compound haze-disaster risk in Kalimantan, Indonesia (2001-2026)' (prepared for submission to the International Journal of Disaster Risk Reduction). CONTENTS: kalfire_reproducibility_code.zip (84 entries): the full analysis pipeline (src/*.py, stages s01-s19), run_all.sh, requirements.txt, stage logs and output tables. The code itself is released under the MIT licence. kalfire_derived_data.zip (17 entries): derived interim and processed products (parquet/geopackage) behind every published number, including burned-area retrievals, the province and regency exposure panels, the regency risk index and the AERONET-CAMS match table. MANIFEST.sha256.json: SHA-256 digests for 533 files (raw inputs and derived products). Raw inputs are NOT bundled; they are openly retrievable from the endpoints listed in Supplementary Table S1 of the article, and the pipeline re-downloads them. The manifest records each file's URL, byte count and SHA-256 digest. README.md: pipeline documentation and quick-start guide. s15_verification.json: 53/53 verification checks passed (estimator correctness, data integrity, independence of validation evidence from the fitted period, determinism). Seed fixed at 20240917. s16_rebuild_check.json: 8/8 stochastic-stage artefacts reproduced bit for bit on re-execution. REPRODUCTION: python -m venv .venv && source .venv/bin/activate; pip install -r requirements.txt; bash run_all.sh. A cold run downloads roughly 2 GB and takes about 2-3 hours (mostly network-bound). Every result for 2001-2024 reproduces with no credential at all; two free, instantly issued credentials (a NASA FIRMS map key and a WAQI token) only extend the record into 2025-2026 and supply the reference-station inventory. LICENCE NOTE: The code is MIT-licensed. Derived data remain under the terms of their original providers (NASA FIRMS, GFED4.1s, CAMS via Open-Meteo CC-BY 4.0, AERONET, WorldPop, GADM, NOAA PSL, World Air Quality Index, World Bank). The record itself is published under CC-BY 4.0.