Surface roughness profiles of seven Eucalyptus and Corymbia species under planing, abrasive sanding (P80–P1200), and drilling — LPF/IBAMA, Brasília, Brazil

This dataset contains machined-surface roughness profiles of seven Eucalyptus and Corymbia species (Corymbia citriodora, Eucalyptus camaldulensis, E. pilularis, E. urophylla, E. cloeziana, E. grandis, E. dunnii), measured at the Laboratório de Produtos Florestais (LPF/IBAMA), Brasília, Brazil, on a Mitutoyo Surftest SJ-301 contact profilometer (JIS B 0601:2001, converted to ISO 21920:2021). Three machining categories are covered — planing, abrasive sanding (FEPA P80–P1200), and drilling (14 mm and 25 mm HSS bits) — in longitudinal (parallel to grain) and transverse (perpendicular) directions. The curated master table holds 2,281 profiles (117 columns); a companion file provides the same profiles re-filtered at a wood-native cut-off (λc = 1 mm), 2,375 profiles, after deduplication, which is the basis of the analyses. Physical properties (basic density and dimensional shrinkage) are included for 112 specimens across the seven species. Curated photographs of the specimens and machined surfaces are included with captions. Licence: data files under CC BY 4.0; photographs under a credit line ("Photograph by Gerson H. Sternadt, made during the machining-workability study at the LPF/IBAMA") — see README. Associated companion articles (submitted to Forests, MDPI, 2026): Paper 1 — Vessel Anatomy and the Machined Surface of Eucalyptus and Corymbia: The Rvk Paradox and an Abrasive-to-Vessel Scaling Model; Paper 2 — A Fixed Transversal Characteristic Grit in Sanded Eucalyptus and Corymbia: Localized Multiscale Surface Analysis with Band-Uniformity Safeguards. Version 3 (6 August 2026). The curated master is corrected to v7 (2,281 rows × 117 columns). One row was removed, in which a single measurement had been filed under two grit labels: "ci 19 lixa 220 L.xls" of C. citriodora appeared under both the P120 and the P220 folder with every non-grit column identical, and the filename, the raw archive and the P220 folder all agree on 220. After this correction the master contains no repeated (species, filename) key. The compact raw archive was rebuilt without duplicates (2,415 to 2,375 records). All 40 repeated keys were verified byte-for-byte before removal: in every case both the primary and the roughness trace had identical digests, so they were the same measurement ingested twice rather than a filename collision. The duplication was concentrated in two cells — E. dunnii at P320 and E. urophylla at P600 — so any statistic computed by iterating the previous archive double-weighted those cells. The codebook has been rewritten and completed to 117 of 117 columns. It corrects the description of the pore index, which earlier versions gave as a ratio of Rvk values and which is in fact the pore-region share of Ra above a fixed threshold of two RMS below the mean line; it declares, for the first time, the cut-off, detrending order and moment range behind the MF-DFA columns; it records that the archived Weibull scale is algebraically equal to Ra and should not be reported as an independent descriptor; and it explains why the advanced columns are 98% filled — 51 rows of E. cloeziana top-face measurements at P320 and P400, added to the master in June 2026, never received those computations. Four generators are added, each with a validation gate against the column or artefact it replaces: MF-DFA (r = 0.928 against the archived H2), the Weibull fit (r = 0.985 and 1.000 against W_beta and W_eta), the archive deduplication (every kept record checked digest-for-digest) and the master correction (which refuses to run if the two rows differ in more than the grit label). Two new derived tables accompany them — a two-component Weibull mixture and a recomputed pore index — each computed at two declared cut-offs, the wood-native cut-off of 1 mm and the instrument cut-off recorded per file. The version-3 additions are grouped in forests_v3_package.zip because this record is at the 100-file limit of the repository and all 84 per-species archives of raw exports are primary material that stays. Two files from version 2 are not carried forward — ALL_SPECIES_master_FINAL_v4.csv and codebook_master_v4_116cols.md — because each is superseded by its v7 counterpart, and in the codebook's case because an incorrect description of a column should not sit beside a correct one. Neither was removed to gain space, and both remain permanently available at the version-2 DOI. This version supersedes v2 for all analytical use. Anyone reproducing figures published before August 2026 should cite 10.5281/zenodo.21347008.