Derived data and code for DPSUR-GHRD-HSI recorded-date rainfall-grid matched landslide occurrence analysis
Description
This release provides the derived data and code underlying the revised DPSUR-GHRD-HSI analysis of mapped rainfall-induced landslide occurrence under matched recorded-date and gridded rainfall exposure in northern Sichuan, China. It contains the strict rainfall-only cohort audit (1,918 records and 1,888 unique event-unit-recorded-date keys), trigger/date/position-quality audits, the mapped-record control-exclusion calendar, derived slope-unit descriptors, spatial out-of-fold assignments and scores, HSI definitions and weights, rainfall summaries, pair-level matching and match-level inference tables, area/exclusion/random-seed sensitivities, forward temporal validation, K=2-6 GHRD stability, comparator-model outputs, current Online Resource Tables S1-S9, figure-level numeric inputs, approved artwork and analysis code. R2 analytical results are frozen; this deposit synchronizes them with the R3 minor revision. Direct inventory coordinates and raw third-party spatial datasets are not redistributed. Consistent anonymous keys and encoded rainfall-cell/cluster identifiers preserve analytical joins and grouping. A portable verification script reproduces key statistics and clustered intervals from the public derived tables. Complete raw GIS extraction and exact map re-rendering require the upstream datasets and permissions documented in the README and Online Resource. The matched observational design does not establish verified failure timing or causal triggering.
Files
Steps to reproduce
1. Download all files associated with the current dataset version and extract the derived-data and code ZIP archive. 2. Read README.md and DATA_ACCESS_AND_LICENSE.md before running the verification workflow. CHECKSUMS.sha256 records the SHA-256 digest of each released payload file. 3. Use Python 3.12. Install the required verification packages from the extracted package root: pip install -r requirements-verification.txt 4. Run the portable verification workflow from the extracted package root: python code/verify_public_release.py This command verifies file integrity, cohort grain and joins, the frozen event-control matching tables, zero-overlap spatial cross-fitting, rainfall balance, out-of-fold HSI calculations, baseline results, forward-temporal point estimates, and K-stability summaries. 5. To rerun the primary and forward-temporal two-way bootstrap calculations using the archived functions and random seeds, run: python code/verify_public_release.py --bootstrap 6. Verification outputs are written to verification_run/. The workflow reads only files included in the released package and does not require access to the authors' workstation. 7. The released data support numerical verification from anonymized and frozen derived tables. Direct inventory coordinates, raw inventory identifiers, licensed source rasters and vectors, and exact geospatial layers are not redistributed. Consequently, raw GIS extraction, complete rainfall re-extraction, original fold construction from coordinates, and exact spatial figure regeneration require the source datasets and access conditions described in the manuscript and DATA_ACCESS_AND_LICENSE.md.
Institutions
- University of Chinese Academy of SciencesBeijing, Beijing