Data and code supporting dual-product burned-area detection and persistent agro-pastoral conversion in the Peruvian Amazon

This deposit contains curated derived data, analysis code, figure source data, documentation, and integrity manifests supporting the article “Dual-product burned-area detection at first forest loss marks persistent agro-pastoral conversion in the Peruvian Amazon.” The study uses a retrospective spatial cohort sampled on a systematic 0.01° lattice in stable forests of the Peruvian Amazon. First forest loss during 2014–2021 was identified using Hansen Global Forest Change. Baseline and post-loss land cover were derived from MapBiomas Peru Collection 3, and concurrent burned-area detection at the index loss year was assessed using MapBiomas Fire and GABAM. Persistent conversion required anthropogenic land cover in each of the three consecutive years after the indexed loss event (L+1 to L+3). The archive supports reproduction from two analysis-ready derived panels of standardized risks, spatial block-bootstrap confidence intervals, temporal restrictions, coarsened comparisons, modified-Poisson sensitivity analysis, tables, and Figures 1–4. It does not reconstruct the raster-to-panel extraction and does not redistribute original third-party rasters. The estimates are associational. Dual-product burned-area detection is a screening marker and does not establish that fire independently caused subsequent conversion. Funding: This research received no external funding. Third-party products remain subject to their original terms.