Embedding Relational Philosophy in Socio-Hydrology: An ABM-LLM Prototype of Hydrosocial Territories and Compensation Dynamics

From the paper

Abstract

Abstract Socio-hydrology has advanced our understanding of human–water interactions but often adopts technocratic, reductionist frameworks that overlook power, place, and community values. Financial Compensation for the Use of Water Resources (FCUWR) in the Amazon exemplifies this gap: it internalizes socio-environmental costs of Hydropower Plants (HPP) through narrow hydrological and economic metrics and neglects deeper territorial dynamics. We present a prototype that operationalises the relational philosophy of hydrosocial territories by coupling a Mesa Agent-Based Model (ABM) with a Large Language Model (LLM). Rules derive from a systematic review of 47 papers (55 HPP cases). Operators, regulators, and community agents negotiate water, compliance, and FCUWR under varying scarcity. At each step, the LLM produces a one-line narrative and policy recommendation that adaptively updates a FCUWR multiplier. Fourteen Latin-Hypercube scenarios verify four hypotheses: conflict escalates under scarcity (14/14 scenarios), intensifies with power imbalances (11/14 scenarios), is mitigated by adaptive FCUWR payouts (11/14 scenarios), and conflict-variance grows with community plurality (14/14 scenarios). This proof of concept shows how narrative-driven ABM-LLM embeds social complexity – power, identity, territory – missing from conventional socio-hydrological tools.