2.2 GCMs and GGESs
GGESs provide plausible descriptions of how the world might evolve
during the 21st century with respect to a range of variables, such as
technological change, socioeconomic change, energy and land use, and
emissions of greenhouse gases and air pollutants (Vuuren et al., 2011).
GCMs driven by a time series of GGESs are considered as the most
essential and feasible tools for projections of future climate change
and have been widely applied to evaluate the hydrological impacts of
climate change (Kay et al., 2009; Habets et al., 2013; Xu et al., 2013;
Alfieri et al., 2015; Wu et al., 2014, 2015). In this study, the GCM
simulations were obtained from the Coupled Model Intercomparison Project
Phase 5 (CMIP5) database. The CMIP5 climate change projections are
driven by a time series of emissions and concentrations from the
representative concentration pathways (RCPs), consistent with a high
energy–intensive scenario RCP8.5, a midrange mitigation emission
scenario RCP4.5, and a low energy–intensive scenario RCP2.6 (IPCC,
2013). The high–resolution (0.25°×0.25°) downscaling multi–model
ensemble averages of 13 CMIP5 GCMs for the baseline 1970–2000 and
future period 2020–2050 with 3 different GGESs (i.e., RCP2.6, RCP4.5
and RCP8.5) were used at the time of this analysis (as shown in Table
1). For the baseline 1970–2000 and future period 2020–2050 under each
GGES, a total of 50 simulation samples of daily temperature and
precipitation were generated based on the following statistical
downscaling process (Wu et al., 2014): (1) The GCM outputs (i.e.,
temperature and precipitation) were interpolated to the sites of the
study basin using the bilinear interpolation method; (2) On the basis of
the observed monthly temperature and precipitation at multiple sites,
generalized additive model was fitted to generate local–scale simulated
sequences of the baseline and future climate change scenarios for each
GCM; (3) Local–scale simulated sequences from the selected GCMs were
weighted averaged using Bayesian model averaging method; (4) The monthly
weighted averaged climate simulations were temporally disaggregated into
daily weather forcings based on the stochastic weather generation
method. Each GGES consists of 50 simulated samples of daily temperature
and precipitation; (5) The 50 simulated samples of daily temperature and
precipitation were finally interpolated to a high–resolution grid
(0.25°× 0.25°) of the study basin using the bilinear interpolation
method.
2.3 Hydrological Models
and Uncertain Parameters
The model uncertainty of this research consists of two different
hydrological models: the variable infiltration capacity (VIC) model and
Xinanjiang model. These two widely used but totally distinct models have
some similar outputs and are suitable for comparison.
2.3.1 VIC Model
The VIC model is a semidistributed, grid–based, hydrological model that
was developed by the University of Washington and Princeton University.
The model can simulate the physical exchange of water and energy among
the atmosphere, soil and vegetation in a surface vegetation–atmospheric
transfer scheme (Liang et al., 1994; Lohmann et al., 1998). In this
study, version 4.1.2d of the VIC model (available at
www.hydro.washington.edu/Lettenmaier/Models/VIC/index.shtml) was run to
simulate the water balance over 50 grid points (0.25°× 0.25°) of the
study catchment. The Dag Lohmann model (Nijssen et al., 1997) was used
for transporting the grid cell surface runoff and baseflow simulated by
the VIC model within each grid cell to the outlet of that grid cell and
then into the river system.
There are seven uncertain hydrological parameters in the VIC model: the
infiltration curve shape parameter B ; the soil depth of layers 1
(d 1), 2 (d 2), and 3
(d 3); and the three base flow–related parametersDm , DS , andWS . In this study, we chose the parametersd 2 and B , since they are commonly
considered as the most sensitive parameters (Demaria et al., 2007).B represents the relative area ratio of the average water content
of the grid to the maximum water content of the grid. Larger Bvalues indicate greater inhomogeneity of the spatial distribution of
moisture content and more surface runoff. Meanwhile, changes ind 2 can significantly increase the evaporation
loss and decrease the seasonal peak discharge. The ranges of parameters
were chosen based on the minimum and maximum parameter values in the
Global Land Data Assimilation Systems (GLDAS) data set for China. In
this research, 20 samples were generated following Latin hypercube
sampling (LHS) method for these uncertain parameters. We assumed that
uncertain parameters follow uniform or normal distributions (Table 2).