CONUS forest carbon under climate x harvest x disturbance: a yield-curve factorial (2025-2125)

Projections of conterminous United States (CONUS) forest aboveground carbon from 2025 to 2125 under a full factorial of three drivers: climate (RCP4.5, RCP8.5), harvest (five removal intensities), and natural disturbance (off, observed, climate-amplified, and fire-only and insect-only amplification). FIA-anchored yield-curve engine, run per state and summed to CONUS on the OSC Cardinal cluster. Disturbance is parameterized from a raster-calibrated product and applied via a regrowth-aware recursion validated against Monte Carlo (0.4 to 2.5 percent); the disturbance-timing variance band at CONUS scale is about 0.1 percent. Version 1.2.0 adds a cross-method calibration of the disturbance carbon effect. A pool-consistent disturbance arm scales the engine's aboveground-carbon loss rate to the CBM/libcbm live-versus-dead accounting; this gives a CONUS disturbance penalty of 612 Tg C, which reproduces an independent libcbm TreeMap live-carbon result (615 Tg C) to within 0.5 percent and leaves CONUS a +29 percent sink. The disturbance effect is therefore BRACKETED: 612 Tg C (pool-consistent, recommended central estimate, CONUS stays a sink) to 5,386 Tg C (aboveground-drag upper bound, CONUS flips to a source). The robust, method-independent result is the direction (disturbance lowers live carbon and concentrates exposure in unharvested reserves); the absolute sink-to-source flip is a model-form artifact of the aboveground-only drag and is not the central result. Added files: treemap_harmonization_table.csv (per-state YC versus CBM rates), tm_harmonize.csv (per-state scaling), and yc_pool_consistent_states.csv (per-state off, observed, pool-consistent 2125 carbon). All released values are state- or CONUS-level aggregates; no FIA true plot coordinates are included.