Multi-method late-successional and old-growth (LSOG) forest mapping uncertainty for Maine's unorganized townships

This dataset quantifies the uncertainty in mapping late-successional and old-growth (LSOG) forest across the approximately 4.2 million hectares of Maine's unorganized townships, and tests whether LSOG is rapidly disappearing. Three to four independent, credible mapping methods are compared on a common 100 m grid: (M1) a reproduction of the Hagan et al. (2026) airborne-LiDAR canopy random forest, rebuilt from their public Zenodo deposit; (M2) a logistic model of the FIA field-structure LSOG class on Potapov (GEDI-calibrated) canopy height; (M3) a direct canopy-height threshold; and (M4) the FIA structural class imputed to every pixel via USFS TreeMap (2016, 2020, 2022). Headline findings. Credible methods disagree by roughly 2.8 times on how much LSOG exists (7.8 to 21.9 percent any-LSOG) and on the location of about four of every five LSOG hectares (Cohen kappa near 0.21 between the airborne-LiDAR and canopy-height views). Protecting the top 5 to 20 percent of hectares by one map versus another overlaps on only 16 to 30 percent of the ground, so single-map patch-level prioritization for large expenditures is fragile. Two independent stock measurements (TreeMap imputation and FIA design-based plot panels) show LSOG stable to increasing over 2016 to 2023 (+2.4 to +3.2 percent per year), the opposite sign to the rapid-loss premise; the apparent loss reported elsewhere is a gross harvest flux, not a net stock decline. Contents. Derived 100 m GeoTIFFs (reproduced Hagan class, v5.1-GEDI probability, TreeMap class, a per-cell method-consensus layer), summary tables (area by method, pairwise agreement, concordance, prioritization fragility, temporal trend, and FIA validation), the analysis R scripts, quick-look figures, and a full methods-and-findings report (PDF). Privacy. No FIA plot coordinates are included; all products are derived rasters or aggregate summary tables. Caveats: TreeMap year differences blend real change with FIA-panel vintage and imputation-model updates, so the robust temporal signal is direction rather than precise rate; TreeMap imputation smooths rare classes, so it is best read for amount and trend, not pixel location. See the README and report for full methods, provenance, and limitations.