Increasing behavioral richness and managing structural uncertainty in social-ecological system agent-based models
Year: 2024
Type: article
Abstract: Responding to the challenges of societal transformation in the face of climate change, efforts to integrate behaviorally rich models of adaptation decision-making into large-scale macroeconomic and Earth system models are growing and agent-based models (ABMs) are an effective tool for doing so. However, behavioral richness in ABMs has been limited to implementations of single decision models for all agents in a simulated population. The main goals of this study were to: 1) implement the ‘building-block processes’ (BBPs) approach for decision model heterogeneity; 2) demonstrate the application of sensitivity and uncertainty analyses to quantify the scope of structural uncertainty produced by alternative decision models under variable price and climate conditions; and 3) apply the Observing System Simulation Experiment (OSSE) approach to validate such a behaviorally rich BBPs model at the level of individual agent decisions. Using an ABM of agricultural producers’ decision-making, we demonstrated that uncertainty in crop and farm management decisions introduced by heterogeneous decision models was equal to and in some instances greater than that due to variable price or precipitation conditions. Unrealistically rapid or stagnant behavioral dynamics were evident in model versions implementing single decision models for all agents. Moreover, interactions among agents with diverse decision models in the same population produced consistently more accurate outcomes and realistic behavioral dynamics. The BBPs framework and accompanying sensitivity and uncertainty analyses demonstrated here offer a path forward for increasing behavioral richness in ABMs, which is key to understanding processes of adaptation central to societal responses to climate change.
DOI: 10.18174/sesmo.18749
Authors: Nicholas Magliocca, Ruchie Pathak, Ashleigh Price, Hashir Tanveer, Mukesh Kumar, Hamid Moradkhani
Institution: University of Alabama
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
SDG: Climate action
Open Access status: diamond
OpenAlex ID: W4405366844