Ponente
Descripción
This study introduces a novel sociohydrological modeling framework that couples an agent-based model (ABM) with a groundwater flow model to simulate groundwater dynamics and human decision-making in the Patiño Aquifer system of Paraguay. Supplying drinking water to more than 1.5 million people in Greater Asunción, the aquifer also serves as the primary water source for several industries, placing this critical resource at increasing risk of overexploitation in the near future. Water provision across the region is fragmented among three provider types: the public agency (ESSAP), community-managed juntas de saneamiento, and private aguaterías. Large industrial users are treated as direct water extractors independent of the distribution network.
The ABM, implemented in Python using NetworkX and FloPy, represents the three provider typologies, large industrial consumers (bottling companies and other facilities), and domestic consumers. Agent behavior is governed by bounded rationality, with decision trees parameterized using a combination of regulatory instruments, electricity tariffs, stakeholder interviews, and household income data. Pumping costs are computed dynamically as a function of water table depth, well screen exposure, and electricity tariffs. Consumer financial stress modulates demand through a five-tier affordability scale. Model coupling uses an epoch-based strategy, with quarters defined as 91.25 day periods, to reconcile discrete agent state transitions with the numerical stability requirements of the groundwater flow model. The model is currently in calibration, benchmarked against 29 targets drawn from hydrogeological, socioeconomic, and governance sources.
Preliminary results (2020-2021 proof-of-concept). The coupled model was executed for eight quarterly epochs. The resulting water budget shows pronounced seasonality, with recharge ranging from 13.9 Mm³ in dry season to 112.6 Mm³ in wet season. Total extraction remained stable at 31-34 Mm³/quarter throughout both years. Despite seasonal recharge recovery, cumulative aquifer storage declined by 110.4 Mm³ over the simulation period, with dry-season drawdown consistently exceeding wet-season replenishment. Head-change maps show 66% of active cells declining by Q4/2021, with a mean change of −3.5 m concentrated in the urbanized central zone. Well criticality increased: stressed wells (depth-to-water 60–80 m) went from 31 to 70 and critical wells (>80 m) from 27 to 35.
At the provider level, 380 Juntas de saneamiento serve approximately 688,000 people with a mean per-capita supply of 284 L p⁻¹ d⁻¹, well above the 150 L p⁻¹ d⁻¹ regulatory minimum. Private Aguaterías (254 providers, 571,000 people) maintain supply above 279 L p⁻¹ d⁻¹ but incur elevated pumping costs in Q1/2020 under La Niña drought conditions before stabilizing by Q4/2021. No provider reaches formal financial distress across either simulated year, though intermittent debt accumulation is recorded among a small subset of private operators in dry quarters, indicating emerging financial vulnerability under sustained water-table decline.
These preliminary results demonstrate the model’s capacity to reproduce emergent patterns (drawdown hotspots, extraction-cost feedback, seasonal demand) and highlight the relevance of improved calibration of storage parameters and pumping-cost functions. The framework is designed to conduct scenario analysis of tariff reform, enforcement policies, and climate-driven recharge changes in one of South America’s most pressured peri-urban aquifers.