2.1 The Model System and Hierarchical Framework of Uncertainty Quantification
In this research, we have built a complex model system to predict the behavior of a hydrological system (e.g., discharge and surface runoff) considering climate change uncertainty. In this model system, the hydrological models were integrated with GGESs and GCMs to capture the impacts of climate change on the hydrological system. The model system thus includes four uncertainty sources: multiple alternative GGESs and plausible GCMs, two hydrological models: variable infiltration capacity (VIC) model and Xinanjiang model, and variant model parameters. Based on the work of Dai et al. (2015, 2017a), we have developed a new hierarchical framework for the uncertainty sources in this research, as shown in Figure 1. This hierarchical framework arranged the four uncertainty factors following the deterministic relationships among parameters, models, and scenarios: parameters depend on models and models depend on scenarios.
This hierarchical framework is flexible and can be simplified without the hydrological model uncertainty layer. This simplification is based on the fact that only the VIC model can provide surface runoff distribution for the model domain (the Xinanjiang model is a lumped watershed hydrological model and provides only a single prediction value for the entire domain); therefore, we can ignore model uncertainty for the surface runoff output.