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.