Prospective Final Execution Registration for Censored-Likelihood Inference in Networked Environmental Transport, Version 1.2

Résumé fourni par la source

This record prospectively fixes the production benchmark and final bounded Trinity analysis plan for censored-likelihood inference on river networks. Data collection, related Trinity analyses, staged software development, and a result-blind computational pilot preceded this record. At publication, no 51-cell production benchmark and no final Trinity model fit under the registered plan had been executed. The registration locks 51 synthetic scenario cells, 200 replicates per cell, 200 multiple imputations, four masked comparison pathways, exact seed and worklist rules, initialization settings, runtime and process-memory limits, reporting-limit inputs, recovery thresholds, failure handling, and an empty deviation log. The Trinity component records an outcome-informed Campaign 2 expansion and the H2 hydrologic scope: full-network concentration and site-total outputs, with loads only at directly gaged site-campaigns and no ungaged interval-load reconstruction. The uploaded archive is deliberately lean. It excludes unpublished article drafts, raw laboratory reports, the full analytical dataset, software binaries, private method mappings, private pilot receipts, figures, logs, caches, and intermediate build artifacts. It contains no production benchmark results or scientific performance conclusions. Registered archive SHA-256:84c897626ba17ffd673017eada34f480e20b5ffa99f909d981ce8103fb175753

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.