Continental forest disturbance vulnerability prediction rasters for CONUS (TreeMap 2016, 2020, 2022), v2.0.0 recalibrated three-stage relative density model

v2.0.0 recalibrated three-stage relative density model. This release supersedes v1.0.0 and replaces every surface in the prior version with a fully recalibrated and internally consistent set built from one relative density definition. Contents. Wall to wall 30 m prediction surfaces for the conterminous United States that map forest disturbance vulnerability as a function of pre-disturbance relative density, for three TreeMap structural vintages (2016, 2020, 2022). Each vintage carries four raster surfaces (recalibrated relative density, probability of any disturbance, unconditional expected proportional mortality, and most probable disturbance agent) plus a categorical legend and a per year summary table. All rasters are EPSG:5070 (NAD83 / Conus Albers), grid 154179 by 97279 cells, single band GeoTIFF, LZW compressed; continuous surfaces are Float32 with NaN NoData and the dominant agent surface is 8 bit unsigned integer. Recalibration correction. v1.0.0 computed relative density at the FIA subplot level, which returned only about one quarter of the true stand density because the FIA per acre expansion factor is designed to sum across all four subplots. v2.0.0 computes relative density at the FIA condition level using the additive stand density index over all live trees with diameter at or above 1 inch divided by the production maximum stand density index, the identical recipe used to build the TreeMap relative density surface. The recalibrated relative density is a true proportion of maximum density (mean near 0.63). The direction of every result is unchanged. Three-stage model. Stage 1 occurrence (Bernoulli mixed effects, odds ratio 3.04 per unit relative density), Stage 2 agent composition (multinomial logit against an animal damage baseline, insect relative risk 2.75 and disease 3.18 per unit relative density), Stage 3 severity (Beta regression, density slope -0.75 on the logit scale, odds ratio 0.47 per unit relative density). The model is fit on 151261 FIA conditions (disturbance prevalence 0.251). Dual role finding. Higher relative density raises the probability that disturbance occurs and tilts the agent mix toward biotic agents, while at the same time lowering the proportional mortality when disturbance occurs. Density is both a driver of disturbance likelihood and a buffer on disturbance severity. Associated manuscript. Weiskittel A.R., Chivhenge E., Woodall C.W., D'Amato A.W., Daigneault A. (2026). Forest relative density structures growth and disturbance tradeoffs in continental carbon dynamics. One Earth (in review). Funding. USDA NIFA McIntire-Stennis project ME0-42414; NSF EPSCoR Track 1 RII through the Maine FOREST E-RISE program; Center for Research on Sustainable Forests, University of Maine. Computation used the Ohio Supercomputer Center Cardinal cluster.