Applying Time-Dependent Variance-Based Global Sensitivity Analysis to Represent the Dynamics of an Agent-Based Model of Land Use Change.
Arika Ligmann-Zielinska,Libo Sun
DOI: https://doi.org/10.1080/13658816.2010.490533
2010-01-01
International Journal of Geographical Information Science
Abstract:The growing body of knowledge on modeling land use systems points to epistemic uncertainty as one of the challenging obstacles in development and application of agent-based models (ABMs). To decrease outcome uncertainty, sensitivity analysis (SA) is performed as part of model verification and validation. Oftentimes, however, it is inadequately addressed, partly because of the lack of tools and techniques that focus on an explicit evaluation of ABM dynamics. The nonlinear processes, inherent in such models, necessitate longitudinal SA with time path investigation of input-output relationships of endogenous variables. In response to the outlined deficiencies, this study investigates the potential of time-dependent global sensitivity analysis (time-GSA) in examining the dynamics of outcome uncertainty of a simple ABM of land use change. Specifically, we apply first and total order sensitivity indices to decompose variance of output landscape fragmentation, apportioned to model inputs for multiple time steps and multiple realizations of the ABM. We focus the analysis on selected complex systems characteristics including preference uncertainty, path dependence, access to information, and magnitude of interactions and feedbacks. We conclude that the factor sensitivity measures vary significantly during model execution. Consequently, a static snapshot of ABM sensitivity, taken at the end of the simulation, is inadequate when deciding on factor prioritization and reduction. Assuming that ABM dynamics is a result of factor interaction, we observe a distinct time lag of nonlinearity, which unfolds after the formation of the seeds of development. Therefore, we argue for further application of time-GSA in ABM as one of the visual quantitative techniques contributing to evaluation of ABM nonlinearity.