An analysis of spatial and temporal uncertainty propagation in agent-based models

Area of Science:

• Computational modeling and simulation
• Geospatial analysis
• Uncertainty quantification

Background:

• Complex systems simulations generate spatial uncertainties impacting outcome reliability.
• Current uncertainty analyses often fail to capture the temporal dynamics of spatial uncertainties.
• Effective visualization of spatio-temporal uncertainty propagation is needed for model assessment.

Purpose of the Study:

• To develop and present a novel Spatio-Temporal Uncertainty Analysis (ST-UA) approach.
• To generate an uncertainty propagation index for visualizing temporal uncertainty dynamics.
• To address limitations in current uncertainty analyses for spatio-temporal models.

Main Methods:

• Proposed a Spatio-Temporal UA (ST-UA) framework.
• Utilized the total effects sensitivity measure (Sobol index) for uncertainty assessment.
• Applied the ST-UA to the Tobacco Town Agent-Based Model (ABM) simulating smoking behaviors.

Main Results:

• Demonstrated the temporal propagation of uncertainty in cigarette purchases due to variations in wages and smoking rates.
• Generated an uncertainty propagation index to visualize uncertainty dynamics over time.
• Showcased the ST-UA's ability to identify areas of high uncertainty at different model states.

Conclusions:

• The ST-UA effectively communicates the reliability of spatial model outcomes.
• The approach guides modelers to regions with significant uncertainty across temporal steps.
• ST-UA is transferable to diverse application domains with spatio-temporal uncertainties.