Radar mondial UDTR

UDTR — Radar mondial des droits de la Nature

Registre documentaire des références détectées à travers les infrastructures académiques, juridiques et institutionnelles surveillées par UDTR.

Une infrastructure documentaire mondiale conçue pour atteindre la couverture la plus large, vérifiable et scientifiquement défendable des droits de la Nature.

← Retour au Radar

Late-Successional and Old-Growth Forest Across Northern New England and in Maine: CRSF Technical Reports, with the Maine Analytical Products

Identification

Clé publique
radar:r-d02a0d1171f61b0a114b0b3136a59b84
Auteurs
Weiskittel, Aaron · Woodall, Christopher W. · D'Amato, Anthony W.
Année déclarée
2026
Langue déclarée
en
Type source
Report
Première détection
2026-11-13T08:39:53.335854+00:00
Dernière observation des métadonnées
2026-11-13T08:39:53.335854+00:00
Dernière vérification HTTP
Non effectuée

Institutions déclarées par la source

University of Maine, Center for Research on Sustainable Forests · CTrees · University of Vermont

Résumé documentaire — langue originale

Two CRSF technical reports are archived here together with the Maine analytical products behind them: a four-state regional report covering Maine, New Hampshire, Vermont, and New York, and a Maine-focused companion that tests the LiDAR-based older-forest map of Hagan et al. (2026) directly. Both ask the same three questions. How much late-successional and old-growth (LSOG) forest is there, where is it, and is it changing? Both answer from the same evidence base, the USDA Forest Service Forest Inventory and Analysis (FIA) probability sample, and both treat satellite and LiDAR maps as products to be tested against that sample rather than as substitutes for it. These are technical reports. They have not been peer reviewed. What the inventory shows. Design-based FIA estimates put any-LSOG at 14.7% of forestland in Maine [13.4, 15.9], 27.8% in New Hampshire [25.0, 30.6], 29.0% in Vermont [25.8, 32.1], and 30.5% in New York [28.8, 32.2]. Maine's interval does not overlap any of the other three, while the three neighbors are not separable from one another, so what the inventory resolves is Maine at the bottom and the other three together, not a four-way rank. Restricted to the mapped footprint of the unorganized townships, about 4.10 million ha of forest, the Maine design-based any-LSOG estimate is lower at 12.6% [11.1, 14.1], because those townships are more heavily managed than the state as a whole; both the statewide and footprint values are computed on Maine EVALID 232301, measurement years 2019 to 2023, with every forest condition on every plot expanded onto the full FIA forestland denominator, so they are directly comparable. The strict four-axis definition, which requires a plot to satisfy every structural criterion at once rather than any one of them, is far more restrictive: 2.5% in Maine [1.9, 3.1], 11.3% in New Hampshire, 15.1% in Vermont, and 12.4% in New York, with New Hampshire and New York indistinguishable from one another. The Maine four-axis value is reported on the denominator where the continuity axis is actually defined, and 22% of Maine forestland sits outside the disturbance product's area of interest and is reported as undefined rather than silently passed. The gap between the permissive and the strict figure is not a disagreement between methods. It is the definition doing the work, and it is the single most important thing a reader should carry away, because a reported LSOG percentage means little until the criterion behind it is stated. Honest reporting gives both ends. A third quantity appears in the reports and must not be confused with either: the design-based LS+OG structural grade, which is 2.1% of Maine forestland [1.6, 2.6]. That grade is a direct FIA design-based estimate of the combined late-successional and old-growth structural class, while the four-axis value is the share of forestland clearing all four structural criteria simultaneously in the multi-axis funnel. The two are built differently, they answer different questions, and they are reported separately throughout. The any-LSOG and four-axis figures also do not rest on the same FIA evaluation, and each report states which evaluation carries which number. What the maps can and cannot do. AlphaEarth satellite embeddings are the strongest single input tested. A classifier trained on true-coordinate field plots discriminates the multi-axis class at an area under the curve of 0.865 under random folds and 0.843 under spatially blocked cross-validation, blocked on H3 grid cells because FIA plots are spatially autocorrelated and ordinary random folds flatter every model by roughly 0.02 to 0.05. Nothing tested improves on that, and nothing tested measurably degrades it either. Adding a second embedding moves discrimination by an amount indistinguishable from zero: the paired block-bootstrap delta against AlphaEarth alone is -0.005 with a 95% interval of -0.013 to +0.002, and 7% of resamples favor the stack, so the step from 0.843 to 0.838 should not be read as a decline. TESSERA embeddings alone are clearly the weakest input, at 0.818 under random folds and 0.772 under blocking, and that contrast is resolvable, a paired delta of -0.071 on an interval of -0.095 to -0.047 that excludes zero. The four-source fused row, published at 0.835, is flagged in both reports as not currently reproducible, because its disturbance input no longer survives on the cluster; the canopy half of that row recomputes to 0.838 [0.804, 0.870], inside the same undifferentiated band. The defensible operational choice is therefore the single strongest embedding, on the grounds that stacking buys nothing, not on the grounds that stacking costs anything. That level is good ranking, and it is not good enough for a rare-class area estimate. The practical consequence is visible in the cross-map comparison: among map products that are not calibrated to match the FIA share (the reproduced airborne-LiDAR classifier, the ORNL continental old-growth stratum, and the TreeMap FIA imputation), flagged share disagrees by a factor of 1.6 to 12.4 in the area they call older forest. A fourth product, the Potapov/GEDI canopy-height layer, is excluded from this comparison because its operating threshold is calibrated to match the statewide FIA any-LSOG share rather than independently estimated. All four maps agree on only a small fraction of the specific ground, and no field validation to date resolves which is closer to the truth. Reported extent also moves with the grain at which it is computed, falling by about a third between a 10 m and a 200 m cell on a fixed footprint, so a percentage without its grain is incomplete. Maps should therefore be published as calibrated probability surfaces carrying an uncertainty layer and used for ranking, screening, and field targeting, while area accounting stays with the design-based inventory. The Maine analytical products. Nine aggregated tables and seven scripts are deposited alongside the reports, so that the Maine numbers can be checked rather than taken on trust. They cover the four-axis scoring outputs and the multi-axis funnel, the per-axis pass rates, the coverage of the disturbance product's area of interest, the design-based per-class estimates that carry the any-LSOG headline, an ablation that attributes the movement in that headline to each legacy estimator defect one at a time, and the true-coordinate cross-map agreement analysis with H3 block-bootstrap intervals. Every table is a state-level or comparison-level aggregate, and every table is paired with the script that writes it. The scripts are deposited even where their inputs cannot be, because the input restriction is what makes the code the only auditable form of the method. Why the reports exist. They were written for a mixed audience of state agencies, landowners, conservation organizations, and the research community, and they are the methodological source of record for several results that the companion peer-reviewed outputs cite but cannot reproduce from public data. Chief among these is the spatially blocked AUC of 0.843 for AlphaEarth embeddings alone, and the true-coordinate agreement analysis, both of which were computed on restricted true-coordinate plots and are reported here as summary statistics only. Relationship to work under review. A Comment on Hagan et al. (2026) by an overlapping author group is under review at Ecosphere, and a companion methods manuscript is under review elsewhere. Those are separate documents from these reports. They share a reference list and four supporting-information figures that present the same underlying analyses in different form; they do not share body text, and no figure file is common to them. These reports are posted as CRSF working papers and carry no claim of peer review. Restricted data. FIA true plot coordinates are restricted by USDA Forest Service policy. They were held on a secured server, never left it, and appear nowhere in these reports or anywhere in this record; every FIA plot position shown in any figure is the publicly released fuzzed and swapped coordinate. Every staged file was scanned before deposit for decimal-degree coordinates inside the study-area envelope, for degree-minute-second strings, for FIA true-coordinate field names, and for restricted-file tokens; the audit result is recorded in the deposit runbook. Ecological reserve monitoring data from the Maine Natural Areas Program and The Nature Conservancy, continuous forest inventory data from Baxter State Park, and an operational LSOG comparison on a working-timberland ownership were each used under data-use agreement and are not redistributable; only aggregate or agreement-class products derived from them appear here, and no company product is displayed. New Brunswick MagPlot per-plot records and J.D. Irving ownership layers are excluded, and the proprietary J.D. Irving report produced in the same project is deliberately not deposited. Every map and table in this record is aggregated, fuzzed, or derived. Related archives. The wider data, analysis code, figures, and derived raster products supporting both reports are openly archived under concept DOI 10.5281/zenodo.20614496, and the companion manuscript package is at concept DOI 10.5281/zenodo.20836852. The archived training data and code for the map evaluated here are at 10.5281/zenodo.19696494. Cite this record for the reports themselves. Suggested citation. Weiskittel, A., C.W. Woodall, and A.W. D'Amato. 2026. Late-Successional and Old-Growth Forest Across Northern New England: What the Inventory Shows, and the Limits of LiDAR-Based Mapping. CRSF Technical Report, version 1.2.0. Center for Research on Sustainable Forests, University of Maine, Orono, Maine, USA. https://doi.org/10.5281/zenodo.21684054

Identifiants

DOI
10.5281/zenodo.21684054

Repères documentaires

Famille documentaire
Rapport — provenance institutionnelle à préciser
Types déclarés par les sources
Report
Éditeurs
Zenodo

Les mentions d’entités ne constituent pas une preuve de reconnaissance juridique. La juridiction étudiée est distincte du pays d’affiliation des auteurs.

Sources

DataCite · Document 1

Citer cette référence

Weiskittel, Aaron; Woodall, Christopher W.; D'Amato, Anthony W.. Late-Successional and Old-Growth Forest Across Northern New England and in Maine: CRSF Technical Reports, with the Maine Analytical Products. 2026. https://doi.org/10.5281/zenodo.21684054

JSON · CSL-JSON · BibTeX · RIS · CSV · JSON-LD ·

APA, Chicago, ISO 690 : formats bibliographiques fondés sur les champs disponibles. OSCOLA / Bluebook : données structurées uniquement ; aucune conformité juridique complète n’est revendiquée sans vérification des éléments nécessaires.

APA
Weiskittel, Aaron; Woodall, Christopher W.; D'Amato, Anthony W.. (2026). Late-Successional and Old-Growth Forest Across Northern New England and in Maine: CRSF Technical Reports, with the Maine Analytical Products. https://doi.org/10.5281/zenodo.21684054
Chicago
Weiskittel, Aaron; Woodall, Christopher W.; D'Amato, Anthony W.. Late-Successional and Old-Growth Forest Across Northern New England and in Maine: CRSF Technical Reports, with the Maine Analytical Products. 2026. https://doi.org/10.5281/zenodo.21684054
ISO 690
Weiskittel, Aaron; Woodall, Christopher W.; D'Amato, Anthony W.. Late-Successional and Old-Growth Forest Across Northern New England and in Maine: CRSF Technical Reports, with the Maine Analytical Products. 2026. https://doi.org/10.5281/zenodo.21684054

Warning: Undefined array key "last_link_verified" in /home/clients/e27d81e12f04d2bfd747aa264540ac56/sites/_UDTR_APP/legal_library_reference.php on line 31

UDTR — Bibliothèque juridique mondiale des droits de la Nature. Late-Successional and Old-Growth Forest Across Northern New England and in Maine: CRSF Technical Reports, with the Maine Analytical Products. Notice documentaire: https://www.declarationuniverselledesdroitsdelarbre.org/global-rights-of-nature-radar/reference/r-d02a0d1171f61b0a114b0b3136a59b84/ · Dernière vérification HTTP : non effectuée

Signaler une correction

Données de cette version figée

Dossiers et relations documentaires

Rôles des sources

Source bibliographique / publication d’origine · https://zenodo.org/doi/10.5281/zenodo.21684054

Un lien vers une autorité ne prouve pas à lui seul une affirmation juridique. Les résumés et les textes intégraux restent soumis aux conditions de leur source.

Statut documentaire

Pertinence à examiner

Bibliothèque

Cette identité n’est pas encore intégrée à la Bibliothèque juridique.

UDTR — Bibliothèque juridique mondiale des droits de la Nature