Demonstrating correspondence between decision-support models and dynamics of real-world environmental systems

Abstract

There are increasing calls to audit decision-support models used for environmental policy to ensure that they correspond with the reality facing policy makers. Modelers can establish correspondence by providing empirical evidence of real-world behavior that their models skillfully simulate. Since real-world behavior—especially in environmental systems—is often complex, credibly modeling underlying dynamics is essential. We present a pre-modeling diagnostic framework based on Nonlinear Time Series (NLTS) methods for reconstructing real-world environmental dynamics from observed data. The framework is illustrated with a case study of saltwater intrusion into coastal wetlands in Everglades National Park, Florida, USA. We propose that environmental modelers test for systematic dynamic behavior in observed data before resorting to conventional stochastic exploratory approaches unable to detect this valuable information. Reconstructed data dynamics can be used, along with other expert information, as a rigorous benchmark to guide specification and testing of environmental decision-support models corresponding with real-world behavior.