A logical and reproducible geological-analogy model for diagnosing evaluability and prioritizing geochemical investigation of rare earth elements in data-incomplete territories Application to Mato Grosso do Sul, Brazil

Carlos Busón, Sandra Gabas

Territories with uneven geoscientific coverage pose a central problem for mineral investigation because missing information may be mistaken for negative evidence, while the accumulation of proxies may produce high classifications even when necessary geological requirements remain unknown. This study presents PAG ETR as a logical, categorical, multiscale, and reproducible framework for diagnosing evaluability and organizing geochemical investigation hypotheses for rare earth elements, accompanied by a deliberately partial spatial proof of concept in Mato Grosso do Sul, Brazil. The method organizes eleven mineralizing models, M1 to M11, and distinguishes TRUE, FALSE, UNKNOWN, CONFLICT, and NA, retaining SEM_DADO and INDETERMINADO as states of unknown information rather than absence. Progression among N0, non-evaluable, N1, open model, N2, justified targeted geochemical investigation, N3, compatible local evidence, and N4, local contradiction, is governed by non-compensatory rules and by the functions REQ, SUP, and CONF. The 250, 500, and 1,000 km² grids are evaluated independently. The conceptual library contains eleven models, but the spatial proof of concept materialized only M2 and M4 and classified 49 of 1,554 cells as N1 at 250 km², 37 of 793 at 500 km², and 24 of 412 at 1,000 km². No cell reached N2 or N3 in the spatial snapshot. This result is not interpreted as low potential but as a limit on the evaluability of the materialized evidence. A sensitivity scenario for M4 showed that 38, 28, and 17 N1 cells remain when point phosphorus records are excluded and only polygonal phosphorite units are considered. Computational reproduction recovered the same sets of hex_id and 15 executable tests returned PASS. These tests demonstrate logical consistency and determinism of the snapshot, not predictive validity. PAG ETR does not predict the location of deposits, resources, or reserves. In this version, the product should be interpreted as a diagnosis of evaluability and knowledge gaps rather than as a regional prospectivity map. Its contribution is to make explicit when a hypothesis can be opened, tested, supported, or contradicted, and when the available information remains insufficient. Field validation, independent predictive validation, and quantitative comparison with alternative methods remain future steps.