Data and code for “From public lithium disclosures to product-level environmental decisions: Evidence gaps and data requirements”
Description
This data and code repository supports the manuscript “From public lithium disclosures to product-level environmental decisions: Evidence gaps and data requirements”. It documents a structured empirical audit of 163 quantitative environmental and production-cost claims from 18 reporting entities, with 25 registered production systems. Version 3 supplies the complete ESP revision package, local release 2.1.2-esp.20261017, rather than the earlier Version 2 implementation. The archive includes frozen claim and system records, the public-source register and formal search log, de-identified coding and adjudication records, protocol amendments, analysis code, figure scripts and source data, validation tests, and a file manifest. Additions include the environmental-only sensitivity (141 claims), coder-specific candidate-record-set optimization, reporting-entity and source-document overlap analyses, and the illustrative production-system decision-use screen. Candidate record combinations are coding-dependent evidence requests, not validated optimal regulatory requirements or estimates of recovered environmental performance. Frozen scientific input records are preserved. Historical unavailable-input sentinels are retained for provenance and do not denote pending requirements for the submitted analysis. Third-party copyrighted source documents, private governance forms and personal contact details are not redistributed; the source register provides bibliographic and locator information. Author-generated data are licensed under CC BY 4.0; analysis code is licensed under MIT. Reproduction instructions are provided in README.md and REPRODUCIBILITY.md.
Files
Steps to reproduce
Unzip Supplementary_Data_and_Code.zip and enter 06_SUPPLEMENTARY_DATA_AND_CODE/. Follow README.md and REPRODUCIBILITY.md to create the documented Python environment. Run python reproduce_results.py. The recorded clean reproduction executes 25 commands and checks 76 numerical output files byte for byte against the supplied reference outputs. Run pytest -q tests_release for release validation. Generated figure files can vary in rendering metadata across environments; numerical reproducibility is checked separately. Review MANIFEST.csv, CITATION.cff, environment.yml and the code/source-data maps for version, licence, dependencies and provenance.
Institutions
- Qinghai UniversityQinghai, Xining