Composite Site Productivity Index (CSPI) analytical chain v2.0.0 for the conterminous United States

Companion analytical chain deposit for the Composite Site Productivity Index (CSPI) data release at Zenodo concept DOI 10.5281/zenodo.20515034. Version 2.0.0 adds the external remote-sensing validation and the condensation of the manuscript to a single Forest Ecology and Management length paper. New in v2.0.0: External repeat and structural remote-sensing validation. At up to 38,978 unique FIA plots, the biomass growth increment (BGI) tracks observed MODIS GPP at pooled Pearson r = +0.74 (partial r = +0.68 controlling for site index), while the unified-target site index (ESI) is negative (r = -0.41). The pooled value is shown to be partly a between-region aggregation effect (within-region r = +0.46 West, +0.11 East), reproducing the multi-dimensional structure in an independent sensor system. The 1 km MIAMI climatology NPP currently in the composite does not track observed MODIS NPP (r = -0.03), motivating a MOD17 swap. Sentinel-2 and the iBGI generalization. BGI correlates with all Sentinel-2 indices (CIRE +0.76, NDRE +0.74, NDVI +0.73, S2REP +0.70); ESI is negative across all. Adding the Sentinel-2 red-edge position to BGI raises adjusted R-squared for observed NPP from 0.37 to 0.44, generalizing the iBGI result of Lamb et al. (2020, Remote Sensing 12(12):2056) to the conterminous United States, with the shared-optical-variance caveat documented. Canopy height. Absolute LANDFIRE 2023 canopy height is only about 16 percent predictable from the productivity measures jointly, confirming that the productivity signal resides in the height increment rather than in the absolute height. Single-paper condensation. The manuscript was condensed from roughly 15,700 to about 8,900 main-text words by moving the secondary-story results to Supplementary Text S1 to S13, with cross-references renumbered and the abstract trimmed to 325 words. A full pre-submission assessment (corpus consistency, quantitative scaling self-audit, citation integrity, anti-AI-tell, simulated review) is included. Retained from v1.0.0: the v0.10g manuscript drafts, intermediate result tables, pipeline scripts, and outreach package. Source repository at holoros/cspi-conus. This is a pre-submission working draft archived for the analytical chain; the peer-reviewed version of record will be the journal publication.