论文摘要
Wood is harvested from forest by design, so Regulation (EU) 2023/1115 asks whether a harvest left the forest standing or turned the land into something else. A canopy-loss pixel cannot tell a selective harvest from a clear-felled conversion; this per-block record separates them and feeds a shipment report.
核心方法论发现
Held-out harvest blocks that received a harvest finding: 19 of 19.
Held-out conversion blocks that received a harvest finding: 0 of 18.
Controls were pre-registered and run in two rounds, with a fresh held-out set for the second.
物理原理与反演方法论
Each harvest block is read against the public Landsat moist-forest record and the Commission 2020 forest maps, which separate a canopy disturbance that closes again from one that persists. The block record returns one of three outcomes with its dated sources. Controls were chosen and their pass lines written down before the result was read.
文献引用格式与数字对象标识符 (DOI)
@techreport{axine_wood_harvest_record_2026,
author = {Axine Labs},
title = {Harvest or conversion: a per-block record for wood placed on the EU market},
institution = {Axine Labs Technical Whitepaper Series},
number = {12},
year = {2026},
doi = {10.5281/zenodo.23191471},
url = {https://axinelabs.com/papers/wood-harvest-record}
}