A Peep into the Uncertainty-Complexity-Relevance Modeling Trilemma through Global Sensitivity and Uncertainty Analysis

Abstract:

Model Model complexity resulting from the need to integrate environmental system components cannot be understated. In particular, additional emphasis is urgently needed on rational approaches to guide decision making through uncertainties surrounding the integrated system across decision-relevant scales. However, in spite of the difficulties that the consideration of modeling uncertainty represent for the decision process, it should not be avoided or the value and science behind the models will be undermined. These two issues; i.e., the need for coupled models that can answer the pertinent questions and the need for models that do so with sufficient certainty, are the key indicators of a model’s relevance. Model relevance is inextricably linked with model complexity. Although model complexity has advanced greatly in recent years there has been little work to rigorously characterize the threshold of relevance in integrated and complex models. Formally assessing the relevance of the model in the face of increasing complexity would be valuable because there is growing unease among developers and users of complex models about the cumulative effects of various sources of uncertainty on model outputs. In particular, this issue has prompted doubt over whether the considerable effort going into further elaborating complex models will in fact yield the expected payback. New approaches have been proposed recently to evaluate the uncertainty-complexity-relevance modeling trilemma (Muller, Muñoz-Carpena and Kiker, 2011) by incorporating state-of-the-art global sensitivity and uncertainty analysis (GSA/UA) in every step of the model development so as to quantify not only the uncertainty introduced by the addition of new environmental components, but the effect that these new components have over existing components (interactions, non-linear responses). Outputs from the analysis can also be used to quantify system resilience (stability, alternative states, thresholds or tipping points) in the face of environmental and anthropogenic change (Perz, Muñoz-Carpena, Kiker and Holt, 2013), and through MonteCarlo mapping potential management activities over the most important factors or processes to influence the system towards behavioral (desirable) outcomes (Chu-Agor, Muñoz-Carpena et al., 2012).