Prospective Health under Declared Specifications

Scientific abstract and context

This study develops an auditable, model-conditional implementation of prospective Health using longitudinal woody-stem observations from the Harvard Forest ForestGEO plot and controlled synthetic measurement designs. Its stochastic candidate-history law preserves original weight through viable and failed outcomes and distinguishes specification-execution validity from ecological measurement validity.

Held-out observations reveal poor individual-diameter interval coverage, and later-only stem records fail to identify biological threshold entry uniquely. An independent 2021–2024 adult-mortality evaluation reports 685 observed hemlock deaths against 280.3 predicted by the earlier fitted model, identifying severe component-level predictive miscalibration.

Near an imposed adequacy boundary, observation-only synthetic classifications remain uncertain despite strong accuracy on favorable, well-separated worlds. A query-specific counterexample reveals juvenile-threshold information erased by cohort RMS compression. The results identify the evidence needed for a reliable forest Health measurement; an operational forest Health diagnosis is not established.

Explore the five-paper series →

Citation

James, Z. (2026). Prospective Health under Declared Specifications: An Auditable Empirical-Synthetic Forest Measurement Study. Zenodo. https://doi.org/10.5281/zenodo.23268987.

Publication DOI: 10.5281/zenodo.23268987. All-versions DOI: 10.5281/zenodo.23268986. Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International.

← All ILC publications