close

Вход

Забыли?

вход по аккаунту

?

j.jhydrol.2017.10.023

код для вставкиСкачать
Journal of Hydrology 555 (2017) 298–313
Contents lists available at ScienceDirect
Journal of Hydrology
journal homepage: www.elsevier.com/locate/jhydrol
Research papers
Projecting the potential evapotranspiration by coupling different
formulations and input data reliabilities: The possible uncertainty source
for climate change impacts on hydrological regime
Weiguang Wang a,b,⇑, Changni Li a,b, Wanqiu Xing a,c, Jianyu Fu a,b
a
b
c
State Key Laboratory of Hydrology-Water Resources and Hydraulic Engineering, Hohai University, Nanjing 210098, China
College of Hydrology and Water Resources, Hohai University, Nanjing 210098, China
School of Earth Sciences and Engineering, Hohai University, Nanjing 211100, China
a r t i c l e
i n f o
Article history:
Received 13 September 2017
Received in revised form 8 October 2017
Accepted 9 October 2017
Available online 14 October 2017
This manuscript was handled by G. Syme,
Editor-in-Chief
Keywords:
Potential evapotranspiration
Statistical downscaling model
abcd model
Runoff changes
Uncertainty
a b s t r a c t
Representing atmospheric evaporating capability for a hypothetical reference surface, potential evapotranspiration (PET) determines the upper limit of actual evapotranspiration and is an important input
to hydrological models. Due that present climate models do not give direct estimates of PET when simulating the hydrological response to future climate change, the PET must be estimated first and is subject
to the uncertainty on account of many existing formulae and different input data reliabilities. Using four
different PET estimation approaches, i.e., the more physically Penman (PN) equation with less reliable
input variables, more empirical radiation-based Priestley-Taylor (PT) equation with relatively dependable
downscaled data, the most simply temperature-based Hamon (HM) equation with the most reliable
downscaled variable, and downscaling PET directly by the statistical downscaling model, this paper
investigated the differences of runoff projection caused by the alternative PET methods by a well calibrated abcd monthly hydrological model. Three catchments, i.e., the Luanhe River Basin, the Source
Region of the Yellow River and the Ganjiang River Basin, representing a large climatic diversity were chosen as examples to illustrate this issue. The results indicated that although similar monthly patterns of
PET over the period 2021–2050 for each catchment were provided by the four methods, the magnitudes
of PET were still slightly different, especially for spring and summer months in the Luanhe River Basin
and the Source Region of the Yellow River with relatively dry climate feature. The apparent discrepancy
in magnitude of change in future runoff and even the diverse change direction for summer months in the
Luanhe River Basin and spring months in the Source Region of the Yellow River indicated that the PET
method related uncertainty occurred, especially in the Luanhe River Basin and the Source Region of
the Yellow River with smaller aridity index. Moreover, the possible reason of discrepancies in uncertainty
between three catchments was quantitatively discussed by the contribution analysis based on climatic
elasticity method. This study can provide beneficial reference to comprehensively understand the
impacts of climate change on hydrological regime and thus improve the regional strategy for future water
resource management.
Ó 2017 Elsevier B.V. All rights reserved.
1. Introduction
Evapotranspiration not only is an essential element of energy
budget in the earth-atmosphere system, but also plays an important role in water resources (Wang et al., 2012b). Among different
terms to describe the evapotranspiration, potential evapotranspiration (PET) was first introduced by Thornthwaite (1948) and for⇑ Corresponding author at: State Key Laboratory of Hydrology-Water Resources
and Hydraulic Engineering, Hohai University, Nanjing 210098, China.
E-mail address: wangweiguang006@126.com (W. Wang).
https://doi.org/10.1016/j.jhydrol.2017.10.023
0022-1694/Ó 2017 Elsevier B.V. All rights reserved.
mally defined by Penman (1956) as ‘‘the amount of water
transpired in a given time by a short green crop, completely shading the ground, of uniform height and with adequate water status
in the soil profile.’’ As the indicator of evaporative power of atmosphere, PET determines the maximum possible water consumption
from the land surface, and thus is the most excellent indicator for
the changing behavior of climatic and hydrological regime. Due
that PET is the important input for hydrological modelling, reliable
estimation of PET constitutes the basis of evaluating climatic effect
on hydrological processes, especially for future PET projection in
W. Wang et al. / Journal of Hydrology 555 (2017) 298–313
the background that climate change become more pronounced
(Bates et al., 2008].
More than 50 different methods with various complexities
existed for the estimation of PET (Lu et al., 2005). Generally, in
terms of the required inputs of meteorological variables, these
methods can be roughly classified into three categories, i.e.,
temperature-based methods, radiation-based methods, and
aerodynamic-and radiation-based methods. Among them, the Penman (PN) method, the aerodynamic-and radiation-based one, is
always considered to the most reliable method for all climatic conditions due to its physically based characteristic, and is thus recommended as the single standard method for determining the
PET by the Food and Agricultural Organization of the United
Nations (FAO) (Xu et al., 2006). In many regions, however, the
use of PN method is always prevented by the insufficient input
data. The application of temperature-based methods and
radiation-based methods requiring less meteorological data were
thus compelled. For example, with only requiring air temperature
as input, temperature-based methods were widely used in hydrological models (Bai et al., 2016), such as the early version of the Soil
and Water Assessment Tool model (SWAT, Arnold et al., 1998) and
the Hydro-Informatic Modeling System model (HIMS, Liu et al.,
2008). Some works on intercomparisons of PET method suggested
that less data-intensive methods can also give reliable approximation of PET in certain climatological condition if the simplified
methods were sufficiently calibrated (e.g., Federer et al., 1996;
Vorosmarty et al., 1998; Lu et al., 2005).
However, for the estimation of PET in global change study, the
abilities of less data-intensive methods, particularly the
temperature-based methods, to describe PET temporal variability
have recently been questioned in the context of climate change.
For example, compared with the more physically based one, the
temperature-based version of the Palmer Drought Severity Index
(PDSI) overestimated the recent trend of the global drought
(Sheffield et al., 2012). Generally, physically based methods, e.g.,
PN method, are considered more competent in historical climate
change assessment than temperature based ones (Roderick et al.,
2009) due that changes in other atmospheric variables (e.g., wind
speed and relative humidity) other than temperature are proved
to have dominant effect on overall change in PET (e.g., Xu et al.,
2006; McVicar et al., 2012; Wang et al., 2012b). However, as for
future projection study, the issues become more complicated.
Although PN method is more reliable compared with
temperature-based or radiation-based one, more confidences are
always found in downscaling GCMs-derived temperature and radiation data than that of relative humidity and wind speed data
(Randall et al., 2007; Wang et al., 2015), which are the indispensable input data of the PN method. This thus leaves us in a dilemma
in practice with respect to future PET projection: should we use
more reliable methods (e.g., PN method) with uncertain data quality, or more empirical methods (e.g., temperature-based methods)
with more reliable input data (Kingston et al., 2009). Recently,
Wang et al. (2015) investigated the performance of different project approaches for future reference evapotranspiration (RET), a
more narrowly defined term of PET with clearer vegetation type
definition, by combination between RET estimating method and
input data reliabilities and found uncertainties still lied in estimating how much the RET changed.
Apart from the most excellent indicator for the activity of climate change, PET is still the important input data to hydrological
models of water balance study, especially under changing climate
conditions (Hobbins et al., 2001; Xu and Singh, 2005). However,
evidence from many studies suggested that the PET estimation is
not critical for the performance of hydrological model in runoff
simulation (Bai et al., 2016). For example, using 27 different PET
estimation methods, Oudin et al. (2005) compared the perfor-
299
mance of four conceptual rainfall-runoff models for 308 catchments and found simplistic (e.g., temperature-based methods)
performed similarly (even better sometimes) compared with complex PET estimation methods. Similarly, Kannan et al. (2007) concluded that the temperature-based Hargreaves method appears
to be at least as good as the more complex Penman-Montieth
method in SWAT distributed hydrological model run for a small
catchment in Southeastern regions of the United Kingdom. More
recently, Bai et al. (2016) investigated the sensitivities of monthly
hydrological models to different PET across 37 catchments in China
under different climatic conditions and found different PET inputs
can produce similar runoff in both non-humid and humid regions.
However, for the studies on the impact climate change on water
availability though hydrological models, the issue may not be that
simple, especially in the context of more pronounced climatic
effect in the future (Bates et al., 2008; Wang et al., 2013a, 2015;
Yang et al., 2015). The choice of PET method for the hydrological
modelling should be restricted by more factors. On the one hand,
PET changes are proved to more sensitive to changes in relative
humidity and wind speed than air temperature, which is particularly true in China (Xing et al., 2017). On the other hand, data availability may have important influence for climate change impact
assessments since less confidence is proved in GCM-derived vapor
pressure, cloud cover, wind speed and net radiation compared with
temperature (Randall et al., 2007; Haddeland et al., 2011). The
choice of PET method used in the hydrological model may thus
be a specific source of uncertainty in future projection of runoff.
However, systematic investigation on the impact of applying different PET estimation approaches in hydrological model on prediction of future runoff is scare. Moreover, the influences of PET on
AET and hydrological modeling are considered to be different in
energy-limited region and water-limited regions (Donohue et al.,
2007; Roderick et al., 2009; Wang et al., 2012b, 2016a). The diverse
PET estimation approaches may thus give rise to different uncertainty of runoff projections between various climatic regions.
Therefore, to address these research gap, this paper further
extends our previous study of Wang et al. (2015) by comprehensively investigating PET methods dependence for future runoff projections for three catchments in China representing a large
geographic and climatic diversity. Four different PET projection
approaches include more physically based Penman (PN) equation
with relatively uncertain downscaled data, more empirical
radiation-based Priestley-Taylor (PT) equation with more reliable
downscaled data, the simplest and temperature-based Hamon
(HM) equation with the most reliable downscaled temperature
data, and statistical downscaling method with directly selecting
PET as predictand. The abcd model was used to achieve the uncertainty analysis.
2. Study areas and data descriptions
This study was conducted for the three catchments, i.e., the
Luanhe River Basin, the Source Region of the Yellow River and
the Ganjiang River Basin, representing a large climatic diversity.
The locations and the aridity index from geo-spatial datasets
(UNEP, 1997) of the three catchments are shown in Fig. 1a. The
Luanhe River Basin, located in the northeastern part of the Haihe
River Basin with a drainage area of 44,900 km2, is characterized
by the temperate continental monsoon climate type (Fig. 1b). The
average temperature is between 0.3 and 11 °C, gradually decreasing from the lower basin to the upper basin. With strong interannual and intra-annual variability, precipitation of the Luanhe
River Basin has the multi-year average value of 560 mm, which
mostly occurs in summer, especially in July and August. Located
in 95.5–103.5°E and 32–36.5°N, the Source Region of the Yellow
300
W. Wang et al. / Journal of Hydrology 555 (2017) 298–313
Fig. 1. Locations and the aridity index from geo-spatial datasets of the three catchments (a), meteorological stations, hydrological station, elevation, the mean annual
precipitation (P), potential evapotranspiration (PET) and runoff (R) in the Luanhe River Basin (b), the Source Region of the Yellow River (c) and the Ganjiang River Basin (d).
River has a drainage area of 121,972 km2 with the Tangnaihai
hydrological station as the control outlet of the basin (Fig. 1c).
Characterized by a wet and warm summer and a cold and dry winter, the region belongs to the Tibetan Plateau climate system with
annual average temperature and precipitation of 0.5 °C and 495
mm, respectively. The Source Region of the Yellow River has a great
elevation varying from 2670 m in the east to 6253 m in the west,
and most of the regions are covered by alpine meadows and
steppes. With a controlling area of 81,158 km2, the Ganjiang River
Basin is the seventh largest branch of the Yangtze River (Fig. 1d). As
one of the typical rainstorm regions in China, the Ganjiang River
Basin, belongs to the subtropical moist monsoon climate zone with
moderate climate and sufficient rainfall. The multi-year mean
annual precipitation of the Ganjiang River Basin are 1640 mm,
with over 70% of the annual precipitation occurring from March
to June (Yang et al., 2003; Li et al., 2017).
Daily meteorological data from 1961 to 2000 of 28 observation
stations (see Fig. 1b–d) including precipitation, temperature, wind
speed, relative humidity and sunshine duration were provided by
the National Climatic Centre (NCC) of the China Meteorological
Administration (CMA). Monthly runoff observations of the three
catchments (1961–2000) were collected from the Hydrological
Bureau. In addition, twenty-six large-scale atmospheric variables
during the period of 1961–2000 obtained from the NCEP (National
Centers for Environmental prediction) reanalysis dataset, such as
mean sea level pressure, total precipitation, mean temperature at
2 m, specific near surface humidity, wind speed at 1000 hPa, etc.,
were used to calibrate and validate the statistical downscaling
W. Wang et al. / Journal of Hydrology 555 (2017) 298–313
model. The future climate projections were obtained from the second generation of Canadian Earth System Model (CanESM2), participating in the CMIP5 experiment (Taylor et al., 2012).
CanESM2 was selected since it can fully meet the data requirement
to force the PET estimation model and hydrological model used in
this study. More importantly, CanESM2 was identified as one of the
five most suitable climate models for China based on specific evaluation on 20 climate models (Chen and Frauenfeld, 2014) and
ranked in top two in simulating temperature in high altitude area
(Su et al., 2013). In practice, CanESM2 were broadly used in climate
impact studies with favorable feasibility in China (e.g., Wen et al.,
2013; Das and Umamahesh, 2016; Zhang et al., 2016). The past
observed atmospheric variables from NCEP reanalysis dataset processed at the same spatial resolution with CanESM2 (2.8125° 2.8125°), which is required by the statistical downscaling model.
Daily climate model variables under a stabilization scenario
(RCP4.5) including long-term and global emissions of greenhouse
gases, short-lived species and land-use-land-cover (Clarke et al.,
2007; Moss et al., 2010) over the period of 2006–2100 were used
to generate future climatic variables series. The dataset of NCEP
reanalysis and CanESM2 were downloaded from the internet site
(http://www.cccsn.ec.gc.ca).
3. Methodology
In order to investigate the uncertainty of future runoff response
to climate change due to different future PET estimation
approaches, the projected future PETs from physically-based formulation, radiation-based formulation, temperature-based formulation and direct downscaling output from GCM were used as
inputs to drive the calibrated hydrological model. The differences
of projected future runoff were then investigated. The schematic
illustration of the computation processes is shown in Fig. 2. The
detailed descriptions of downscaling method, PET estimation
methods and hydrological model are given below.
301
3.1. Statistical downscaling model (SDSM)
As a hybrid of the stochastic weather generator and regression
methods, the statistical downscaling model (SDSM) developed by
Wilby et al. (2002) was used to project future variables including
precipitation, mean daily air temperature (Tmean), daily maximum
air temperature (Tmax), daily minimum air temperature (Tmin),
net radiation (Rn), relative humidity (RHmean) and wind speed
(U). The major work of the SDSM is to establish quantitative relationship between large-scale circulation patterns and local-scale
observed variables. Due to low-cost and rapid assessments, statistical downscaling methodologies represents the more promising
option than dynamical downscaling approaches (Wilby and
Dawson, 2007; Tatsumi et al., 2015), and have thus been widely
applied to assessments on meteorological, hydrological and environmental responses to climatic change (e.g., Pervez and
Henebry, 2014; Mahmood and Babel, 2014; Gulacha and
Mulungu, 2016). The main procedure for downscaling large-scale
climatic variables consists of the following steps. First, determining
appropriate atmospheric variables for predictors by the screening
operation in SDSM version 4.2. Second, constructing the multiple
regression model between the predictors (the screened largescale climatic variables from NCEP reanalysis) and predictands
(atmospheric variables in meteorological stations) during the period of 1961–1990. Third, calibrating and validating the regression
model in the sub-periods of 1961–1990 and 1991–2000, respectively. Finally, the future climate data provided by CanESM2 under
RCP4.5 scenario over the period of 2006–2100 were inputted into
the multiple regression models to generate the future daily climatic series of each meteorological station. Besides, four statistics,
i.e., the correlation coefficient (r), the ratio of simulated to
observed standard deviation (RS), the coefficient of efficiency
(Ens) and the model biases (Bias) were selected to assess the performance of SDSM. The formulas of the four statistics are shown in
Table 1. Generally, the closer the values of r, RS and Ens to 1 and
the value of Bias to 0, the better the model performed.
Fig. 2. Schematic illustration of this study.
302
W. Wang et al. / Journal of Hydrology 555 (2017) 298–313
Table 1
Formulas of the four assessment criteria (Qs,i, simulated value; Qo,i, observed value;
s , mean simulated value; Q
o , mean observed value).
Q
Assessment criteria
Formula
r
Pn
s ÞðQ Q
oÞ
ðQ s;i Q
o;i
i¼1
qP
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
r ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
P
n
i¼1
sÞ
ðQ s;i Q
n
2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
P
RS
i¼1
oÞ
ðQ o;i Q
2
s Þ2
ðQ s;i Q
i¼1
1
n
n
1
n
n
RS ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
P
Ens
Ens
Bias
i¼1
oÞ
ðQ o;i Q
2
Pn
ðQ o;i Q s;i Þ2
¼ 1 Pi¼1
n
2
Bias ¼ 1n
n
P
i¼1
ðQ o;i Q o Þ
ðQ s;i Q o;i Þ
i¼1
3.2. Potential evapotranspiration estimation methods
Three typical equations reflecting various data requirements
and usually used within hydrological models, i.e., Penman (PN)
(Penman, 1948), Priestley-Taylor (PT) (Priestley and Taylor, 1972)
and Hamon (HM) (Hamon, 1961), were employed to estimate
PET in this study. Among them, Penman (PN) is the most reliable
physically-based one with explicitly incorporating both physiological and aerodynamic parameters (Xu et al., 2006). However, it
requires multiple meteorological variables, including Tmean, Tmax,
Tmin, RHmean, U and sunshine duration (n), and not all the variables
can be ensured with great accuracy and reliability, especially in
future conditions. Priestley-Taylor (PT) is radiation-based method
that requires Tmean and Rn as inputs. The aerodynamic term in PT
equation was replaced by a coefficient (a). Considering that the
temperature observations are obtained more easily than other
meteorological variables, a number of alternative PET estimation
methods only requiring temperature had been developed.
Wherein, Hamon (HM) is a popular one and was employed in this
study, which only requires Tmean as an input. The detailed information of the three methods is shown in Table 2.
Previous study revealed that not all the future variables
obtained by SDSM have high confidence (Wang et al., 2014). The
built SDSM in constructing temperature variables (Tmean, Tmax,
Tmin) and Rn always perform better than in constructing other
meteorological variables (e.g., U, RHmean) (Wang et al., 2015). In
order to investigate the impact of different methods and data in
different reliability on future PET, the following four approaches
were selected: (1) PN (physically-based method) with six downscaled variables (including Tmean, Tmax, Tmin, Rn, U and RHmean)
based on SDSM (for the brevity, we call it SD-PN method after
here), (2) PT (radiation-based method) with relatively reliable
downscaled inputs (Rn, Tmean) based on SDSM (SD-PT method),
(3) HM (temperature-based method) with the most reliable downscaled variable (Tmean) based on SDSM (SD-HM method), (4) downscaling with directly selecting PET calculated by PN equation as
predictand (PN-SD method).
3.3. The abcd model
The abcd model is a widely used monthly hydrological model
worldwide. It has been successfully applied in many basins in
China with different climatic features (e.g., Bai et al., 2016; Du
et al., 2016; Wang and Zhou, 2016) and was thus used in this study
to simulate the hydrological processes of three catchments with a
large climatic diversity. Taking precipitation and PET as inputs
(Alley, 1984; Vandewiele et al., 1992; Sankarasubramanian and
Vogel, 2002), the abcd model proposed by Thomas, 1981, is a conceptual monthly water balance model with four parameters (a, b, c,
and d). The catchment-averaged monthly precipitation and PET
were obtained by Thiessen polygon method (Thiessen, 1911) based
on the stations observations. The structure of abcd model consists
of two parts: soil moisture storage in the upper layer and groundwater storage in the lower layer. In general, the actual evapotranspiration is a function of precipitation and soil water storage in
abcd model. There are two state variables, i.e., the water availability (W) and the evapotranspiration opportunity (Y) Thomas (1981).
The water availability is derived by
W i ¼ Si1 þ Pi
ð1Þ
where W i is the water availability at the end of the i th month;
Si1 is the soil moisture storage at the beginning of the i th month;
Pi is the monthly precipitation. While the other state variable (Y) is
determined by
Y i ¼ Ei þ Si
ð2Þ
where Y i and Ei are the monthly evapotranspiration opportunity
and actual evapotranspiration, respectively. Besides, a nonlinear
relationship between Y i and W i was assumed in the abcd model,
as follows:
Wi þ b
Yi ¼
2a
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
Wi þ b
W ib
2a
a
ð3Þ
the parameter a (0 < a 6 1) represents the propensity of runoff
to occur before the soil is fully saturated (Thomas et al., 1983), b is
the upper limit on the sum of Ei and Si . Thomas (1981) assumed
that after a rainfall the loss rate of soil moisture by actual evapotranspiration was proportionate to PET for the month,
PET i
Si ¼ Y i exp b
ð4Þ
where PET i is the monthly potential evapotranspiration. Combining Eqs. (2) and (4), actual evapotranspiration can be estimated
as
PET i
Ei ¼ Y i 1 exp :
b
ð5Þ
At the part of groundwater storage, runoff is divided into direct
and indirect parts, as follows:
Table 2
Methods used to estimate potential evapotranspiration. Tmean, mean daily air temperature (°C), which is defined as the mean of the Tmax and Tmin; Tmax, daily maximum air
temperature (°C); Tmin, daily minimum air temperature (°C); RHmean, mean relative humidity; U2, wind speed at 2 m (m/s); n, sunshine duration (h); Rn, net radiation (MJ/
m2day), which is calculated by the measured meteorological data (e.g. Tmax, Tmin, RHmean, n et al.); D, slope of vapor pressure curve (kPa/°C); c, psychrometric constant (kPa/°C);
k, latent heat of vaporization (MJ/kg); es, saturation vapor pressure (kPa); e*(Tmean), saturation vapor pressure at the mean temperature; N, daylight hours (h); the above
unobservable variables are calculated based on the Allen et al. (1998). Besides, a and k are the empirical coefficients, which are calibrated by the PN equation in this study.
Method
Formula
Data needed
Reference
Penman (PN)
2 Þð1RH mean Þes
PET ¼ DRn þ6:43cð1þ0:536U
ðDþcÞk
Tmean, Tmax, Tmin, RHmean, U2 and n
Penman (1948)
Priestley-Taylor (PT)
PET ¼ a ðDDþRcnÞk
Tmean and Rn
Priestley and Taylor (1972)
Hamon (HM)
N e ðT mean Þ
PET ¼ k 12
T mean þ273:2
Tmean
Hamon (1961)
W. Wang et al. / Journal of Hydrology 555 (2017) 298–313
Rsi ¼ ð1 cÞðW i Y i Þ
ð6Þ
Gi1 þ cðW i Y i Þ
1þd
ð7Þ
Gi ¼
Rg i ¼ dGi
ð8Þ
Ri ¼ Rsi þ Rg i
ð9Þ
where Rsi , Rg i and Ri are the monthly direct runoff, groundwater
discharge and total runoff, respectively. Gi and Gi1 are the groundwater storages at the end and beginning of the i th month, respectively. Parameter c is a groundwater recharge coefficient which
controls the water input to the aquifers. Parameter d is the groundwater runoff recession constant.
In this study, the 40 years observed data was divided into two
periods. The data in the first sub-period (1961–1990) and second
sub-period (1991–2000) were used for model calibration and validation, respectively. The widely used genetic algorithm (Wang,
1991) was selected to optimize the parameter values of the abcd
model. Meanwhile, the coefficient of efficiency (Ens) and correlation coefficient (r) (see Table 1) were used to assess the performance of the abcd model.
4. Results
303
(Tmean, Tmax, Tmin, Rn, RHmean and U) at each meteorological station.
The statistic assessment of calibration and validation results for
Duolun station are summarized in Table 4, from which it can be
seen that simulated meteorological variables were generally consistent with observations. It is especially true for Tmean, Tmax, Tmin
and Rn. For these variables, r, RS and Ens between simulated and
observed ones exceeded 0.90 in calibration and validation. During
the calibration period, the values of Bias for Tmean, Tmax, Tmin and Rn
were 0.00, while during the validation period, the related values
were 0.22 °C/day, 0.15 °C/day, 0.31 °C/day and 0.51 MJ/m2day, respectively. However, the simulations of U and RHmean were
less satisfactory than that of Tmean, Tmax, Tmin and Rn with the r and
Ens values of about 0.81. Meanwhile, the comparisons of daily
meteorological variables simulated by SDSM and observed data
during the validation period (1991–2000) are displayed in Fig. 4,
showing similar conclusions with the above analysis. Concurrently,
Fig. 5 compares the observed and downscaled mean monthly precipitation in the validation period. The simulated precipitation was
closely consistent with observations, although the values of downscaled precipitation were slightly overestimated in some months.
Generally, the statistical relationships built by SDSM are capable
of reproducing daily meteorological variables and can be used to
project the future climate scenarios over the three catchments.
Moreover, the performance of the built downscaling model in constructing Tmean, Tmax, Tmin and Rn are better than that in constructing RHmean and U.
4.1. Performance assessment of SDSM in constructing various
predictands
4.1.1. Performance assessment of SDSM in constructing meteorological
variables
The selection of the predictor is a critical step for constructing
the downscaling model (Wilby et al., 2002). Similar with what is
adopted in previous studies (e.g., Khan et al., 2006; Wang et al.,
2013c), the predictor selection was performed by the screening
of the most relevant predictors set based on the correlation and
partial correlation analysis in the individual predictors and predictand (Huang et al., 2011; Wang et al., 2013c). Taking the mean
temperature in Duolun station (42.18 °N, 116.47 °E) as an example,
Fig. 3 demonstrates the partial correlation coefficient (p-corr) and
the correlation (corr) between daily mean temperature and 26
atmospheric variables. 10 predictor variables can be selected due
to high correlation. Under the same selection criterion with mean
temperature, the predictors identified for other local variables in
Duolun station are summarized in Table 3. Four assessment
criteria, i.e., r, RS, Ens and Bias, were used to evaluate the performance of SDSM in constructing six meteorological predictands
4.1.2. Performance assessment of four approaches in estimating
potential evapotranspiration
For offering intuitionistic information in evaluating the performance of four PET methods, Fig. 6 displays the statistic results of
inter-comparison of model skills during the validation period with
a Taylor diagram, a polar-style graph. It can be seen that the SD-PN
and PN-SD methods agreed with observations that lied closest the
point marked by ‘‘observed”, and the statistic results for SD-PN and
PN-SD were almost identical in all catchments. For the PET estimated by SD-PN and PN-SD methods in three catchments, the correlation between simulated and observed ones exceeded or
equaled to 0.9. The normalized standard deviation ranged from
1.2 to 2.0, and the values of RMSE were lower than 0.8. Compared
with the SD-PN and PN-SD methods, the SD-PT and SD-HM methods showed weaker capability of matching local climatic information with larger RMSE and lower correlation. In addition, the
spatial distributions of the statistic points in the Ganjiang River
Basin with bigger aridity index were closer compared with that
in the other catchments, indicating that the PET was less affected
Fig. 3. The absolute values of correlation coefficient (corr) and partial correlation coefficient (p = corr) between the mean daily air temperature (Tmean) and 26 NCEP
atmospheric variables for Duolun station.
304
W. Wang et al. / Journal of Hydrology 555 (2017) 298–313
Table 3
Selected predictor variables for all predictands including mean temperature (Tmean), maximum temperature (Tmax), minimum temperature (Tmin), net radiation (Rn), wind speed
(U), relative humidity (RHmean), precipitation (P) and potential evapotranspiration (PET) at Duolun station.
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Variable
mslpgl
p1_fgl
p1_ugl
p1_vgl
p1_zgl
p1thgl
p1zhgl
p5_fgl
p5_ugl
p5_vgl
p5_zgl
p500gl
p5thgl
p5zhgl
p8_fgl
p8_ugl
p8_vgl
p8_zgl
p850gl
p8thgl
p8zhgl
prcpgl
s500gl
s850gl
shumgl
tempgl
Description
Predictands
Tmean
p
Mean sea level pressure
1000 hPa Wind Speed
1000 hPa Zonal Velocity
1000 hPa Meridional Velocity
1000 hPa Vorticity
1000 hPa Wind Direction
1000 hPa Divergence
500 hPa Wind Speed
500 hPa Zonal Velocity
500 hPa Meridional Velocity
500 hPa Velocity
500 hPa Geopotential
500 hPa Wind Direction
500 hPa Divergence
850 hPa Wind Speed
850 hPa Zonal Velocity
850 hPa Meridional Velocity
850 hPa Velocity
850 hPa Geopotential
850 hPa Wind Direction
850 hPa Divergence
Total precipitation
Specific humidity at 500 hPa
Specific humidity at 850 hPa
Specific Near Surface Humidity
Mean Temperature at 2 m
p
p
Tmax
p
p
Tmin
p
p
p
Rn
p
p
U
p
p
p
p
RHmean
p
P
p
PET
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
Table 4
Performance of SDSM in constructing meteorological variables over the Duolun
station. The left and right sides of the line represent the statistics in calibration and
validation periods, respectively.
Variables
r
RS
Ens
Bias
Tmean
Tmax
Tmin
Rn
U
RHmean
0.98/0.98
0.98/0.98
0.97/0.97
0.96/0.95
0.79/0.80
0.76/0.76
0.99/1.02
0.99/1.00
0.97/1.02
0.97/1.03
0.80/0.96
0.81/0.81
0.97/0.96
0.96/0.96
0.94/0.94
0.93/0.91
0.62/0.59
0.58/0.56
0.00/-0.22
0.00/-0.15
0.00/-0.31
0.00/0.51
0.00/0.26
-0.01/-2.28
by the estimating methods over humid areas. Generally, all methods are generally acceptable of estimating the PET over the three
catchments. Meanwhile, the SD-PN and PN-SD methods have better performance than the other two methods.
4.2. Calibration and validation of the abcd model
The PET estimated by PN method and observed monthly precipitation were used to drive the abcd model in the three catchments.
Fig. 7 displays the abcd-simulated and observed monthly runoff
series during the calibration and validation periods. It can be seen
that both simulated monthly runoff amount and hydrograph shape
were consistent with observed values except some peak runoff.
The parameter values of the abcd model optimized by genetic algorithm and the assessment criteria are presented in Table 5. The correlation in the three catchments ranged from 0.83 to 0.90 in the
calibration and validation. The values of Ens over calibration and
validation periods exceeded 0.70 and 0.64, respectively. Although
the performance of the abcd model in validation period is not as
good as that in calibration period over the Source Region of the Yellow River and Ganjiang River Basin, the model results shown in
Fig. 7 and Table 5 are overall acceptable, indicating that the calibrated abcd model is applicable to project the future runoff.
p
p
p
p
p
p
p
p
p
p
p
p
p
4.3. Impact of different potential evapotranspiration inputs on runoff
projection
With the future climate variables during the period of 2021–
2050 constructed based on SDSM as inputs, the calibrated abcd
model was driven to simulate hydrological processes under
RCP4.5 scenario output in three catchments. Wherein, the future
daily PET was projected by coupling different formulations and
input data reliabilities (i.e., SD-PN, SD-PT, SD-HM and PN-SD methods). The distributions of the monthly PET values are presented in
the box-and-whisker plots (see Fig. 8a–c). Of the four PET projection approaches used in this study, SD-HM provided larger PET values in summer months and smaller ones in the other months,
especially for the Luanhe River Basin. Compared with the PET series in the Luanhe River Basin and the Source Region of the Yellow
River, a better agreement for four monthly PET series was presented in the Ganjiang River Basin no matter for the quartile, the
maximum/minimum values or the mean values. In addition, the
PET in January, November and December over the period of
2021–2050 showed smaller ranges than that in the other months.
Generally, the obvious differences in PET magnitude still exist in all
catchments, especially for spring and summer months in the
Luanhe River Basin and the Source Region of the Yellow River with
relatively dry climate feature, although the monthly patterns of
PET projected by four approaches are similar. Besides, the PET projected by SD-PN and PN-SD approaches were closer to each other
for each catchment, demonstrating that the simulation of PET
using SDSM by establishing the ‘‘black box’’ relationship between
large-scale climatic data and PET was also an alternative. As the
other extremely important input variable of the hydrological model, the mean monthly precipitation over the three catchments are displayed in Fig. 8d. The monthly distribution for
future precipitation was generally consistent with the observed
pattern in baseline period, suggesting that downscaled precipitation under RCP4.5 scenario can locally reflect the historical climate
conditions.
W. Wang et al. / Journal of Hydrology 555 (2017) 298–313
305
Fig. 4. Observed and downscaled daily Tmean (a), Tmax (b), Tmin (c), Rn (d), U (e) and RHmean (f) in the Duolun station during the validation period (1991–2000).
Fig. 9 demonstrates the results of simulated monthly runoff for
the period of 2021–2050. The overall impression of Fig. 9 is that the
ranges of four runoff series were larger than that of the corresponding monthly PET, indicating that the runoff for the same
month varies greatly in different years. Concurrently, there were
apparent differences in runoff over the three catchments reflected
by obvious discrepancy in magnitude of the quartile, maximum/
minimum values and mean values. The differences between four
runoff series for a given catchment were therefore due to the
diverse PET values. Furthermore, it is worth noting that the differences of runoff series in the Luanhe River Basin and the Source
Region of the Yellow River with relatively dry climate feature were
more obvious than that in the Ganjiang River Basin with humid climate condition.
4.4. Discrepancy of runoff changes caused by different potential
evapotranspiration inputs
By taking the period of 1961–2000 as a reference period, the
changes in seasonal and annual mean runoff across the three
catchments in the future period (2021–2050) under the RCP4.5
scenario are shown in Fig. 10. The results suggested that the PET
projected by coupling different formulations and input data reliabilities can cause the obvious distinctions in the runoff changes,
306
W. Wang et al. / Journal of Hydrology 555 (2017) 298–313
Fig. 5. Observed and downscaled mean monthly precipitation (P) in the validation period (1991–2000) over the Luanhe River Basin (a), the Source Region of the Yellow River
(b) and the Ganjiang River Basin (c).
which may be quite significant in some situations. For example, for
the spring months (MAM) in the Source Region of the Yellow River
the runoff changes using two PET types (SD-PN and PN-SD) had the
opposite sign to that based on the other two PET types (Fig. 10a). A
similar phenomenon can be seen in the Luanhe River Basin, but for
the summer months (JJA). There can also be obvious differences in
the impact of different PET methods on runoff projection even
where the four PET types resulted in the same sign of runoff
change. For instance, for the changes in annual mean runoff over
the Luanhe River Basin, the runoff reductions in the period of
2041–2050 using SD-HM PET was approximately double of that
using the other PET types (Fig. 10b). On the contrary, the runoff
increments in the Luanhe River Basin over the period of 2031–
2040 using SD-HM PET was less than one-fifth of that using the
other PET types. Generally, there are noticeable discrepancy in
the magnitude of change in runoff and even the adverse change
direction over the three catchments. Meanwhile, the more apparent discrepancies can be found in the Luanhe River Basin and the
Source Region of the Yellow River compared with that in the Ganjiang River Basin.
5. Discussion
5.1. Uncertainty in potential evapotranspiration through different
approaches
The results in this study demonstrated that although the
monthly patterns of PET projected by four approaches during the
period of 2021–2050 for each catchment were similar, the noticeable differences in PET magnitude still existed in all catchments,
suggesting there were uncertainties existing. Similar findings were
also made in some previous studies (e.g., McAfee, 2013; Wang
et al., 2015). In other studies (e.g., Kingston et al., 2009; Kay and
Davies, 2008), the different methods used to estimate PET was
even found to cause differences in the direction of projected
changes in PET. The uncertainty in PET during the history periods
derive primarily from the variables used. For instance, temperature
is the only climatic variable used in the HM method. Thus, the
impact of other variables (e.g., radiation, wind speed and humidity)
on PET cannot be reflected directly. Increasing air temperature can
lead to the increase of PET when using temperature-based methods, and the influence of increasing temperature on PET may be
tempered or even counteracted by the impact of changes in radiation, humidity and wind speed (McAfee, 2013; Xing et al., 2017).
Thompson et al. (2014) found that the PET estimated by
temperature-based methods (Linacre, Blaney-Criddle, and Hamon
methods) are larger than that estimated by the other methods during the period of 1961–1990 over the four representative subcatchments located in the Mekong River Basin. While the radiation
is the relatively major factor to affect the PET when using empirical
radiation-based methods (e.g., Priestley-Taylor, Turc and Abtew
methods). In practical applications, the simpler models with less
data requirements (e.g., temperature-based and radiation-based
methods) were recommended by Oudin et al. (2005), especially
in the situations where data were scarce. Moreover, the
aerodynamic-and radiation-based PN method, the data-intensive
one with six input climatic variables, can adequately reflect the
meteorological information. The PN method was suggested as the
preferable one for determining the PET, even though the use of this
method was always prevented by the insufficient input data in
some regions (Donohue et al., 2010; McAfee, 2013).
However, the uncertainty in future projection of PET is caused
not only by the used formulas, but also by the accuracy of the
required meteorological variables. The biases in any of the input
variables could impact the magnitude of PET. As indicated by Yin
et al. (2015), the uncertainty in variables mainly arises from GCMs.
W. Wang et al. / Journal of Hydrology 555 (2017) 298–313
307
Fig. 6. Taylor diagrams for estimated four potential evapotranspiration types (SD-PN, SD-PT, SD-HM and PN-SD) evaluated against observed ones in the Luanhe River Basin
(a), the Source Region of the Yellow River (b) and the Ganjiang River Basin (c) during the validation period (1991–2000). Three metrics of the correlation (curved axis), the
ratio of the standard deviations (x and y axes) and the RMSE (dashed line) are presented in it.
Furthermore, more confidences are always found in downscaling
GCM-derived temperature and radiation than that of the other
meteorological variables (Randall et al., 2007; Wang et al., 2015).
This is also supported by the results of the current study (see
Fig. 4 and Table 4). In fact, it is difficult to establish perfect multiple
regression equation for variables with conditional behavior
(Gulacha and Mulungu, 2016). Taking the precipitation as an
example, the amounts depend on wet/dry day occurrence, and thus
lies on regional scale predictors, e.g., atmospheric pressure. The
intermediate processes exist between regional forcing and some
local meteorological factors in the conditional model (Wilby and
Dawson, 2004). Therefore, not all of meteorological variables
required in evapotranspiration and hydrological model can be perfectly modeled (Goncu and Albek, 2016). Generally, there are
appreciable uncertainties in PET projected by coupling different
formulations and input data reliabilities (i.e., SD-PN, SD-PT, SDHM and PN-SD approaches) in all catchments, especially for spring
and summer months in the Luanhe River Basin and the Source
Region of the Yellow River with relatively smaller aridity index
(Fig. 8a–c).
5.2. Uncertainty in runoff projection caused by the alternative
potential evapotranspiration methods
As the appropriate indicator for the activity of climate change
and hydrological regime, PET is also considered as the important
input for hydrological model, and thus PET method can determine
the direction of projections of future water resources (Hobbins
et al., 2001; Xu and Singh, 2005; Kingston et al., 2009; Wang
et al., 2012b). The different PET magnitudes were given by the four
projection methods when applied to climate model data under the
RCP4.5 emissions scenario over the period of 2021–2050, and this
can have an obvious impact on the subsequent hydrological simulation in all catchments of this study (see Fig. 9). Although many
studies have indicated that the runoff simulations were insensitive
to the different PET inputs in historical period (e.g., Andreassian,
et al., 2004; Oudin et al., 2005; Bai et al., 2016), the future PET projection method can further add substantial uncertainty to the
existing uncertainty associated with the climate change signal
between GCMs (Kingston et al., 2009; Yin et al., 2015). Moreover,
as indicated by Bai et al. (2016), the reasons for the insensitivities
of runoff simulations to PET inputs are attributed to model parameter calibration using all PET types. It is worth noting that the set of
parameters in abcd model was only calibrated by using observed
precipitation, runoff and one PET type (calculated by PN equation)
rather than using four PET types in this study. This can ensure that
the differences of projection future runoff come from the selection
of different PET projection methods. Despite that the PET methodrelated uncertainty is less than the GCM/RCM-related uncertainty
as for future runoff projection (Kay and Davies, 2008; Wang
et al., 2012a; Thompson et al., 2014), the apparent uncertainties
in runoff projection caused by the PET projected by coupling differ-
308
W. Wang et al. / Journal of Hydrology 555 (2017) 298–313
Fig. 7. Comparison between abcd-simulated and observed monthly runoff for calibration (1961–1990) and validation (1991–2000) periods in the Luanhe River Basin (a), the
Source Region of the Yellow River (b) and the Ganjiang River Basin (c).
Table 5
Model parameter and the performance assessment of the abcd model during calibration and validation periods over the three catchments.
Catchment
Luanhe River Basin
Source Region of the Yellow River
Ganjiang River Basin
Parameter values
Calibration
Validation
a
b
c
d
Ens
r
Ens
r
0.981
0.744
0.951
290
421
457
0.446
0.125
0.372
0.089
0.060
0.258
0.700
0.753
0.810
0.838
0.869
0.900
0.715
0.654
0.646
0.870
0.832
0.873
W. Wang et al. / Journal of Hydrology 555 (2017) 298–313
309
Fig. 8. Box-and-whisker plots for the monthly potential evapotranspiration values projected by four approaches (SD-PN, SD-PT, SD-HM and PN-SD) ((a), (b) and (c)) and
monthly mean precipitation (d) for the period of 2021–2050 under RCP4.5 over the three catchments. In the box-and-whisker plots, the outer edges of the boxes and the
horizontal lines within the boxes represent the 25th, 75th, and 50th percentiles of potential evapotranspiration values. The whiskers represent the minimum and maximum
potential evapotranspiration values. The filled circles within the boxes represent the average potential evapotranspiration.
310
W. Wang et al. / Journal of Hydrology 555 (2017) 298–313
Fig. 9. Box-and-whisker plots for the monthly runoff values for the different PET methods in the period of 2021–2050 under RCP4.5 over the Luanhe River Basin (a), the
Source Region of the Yellow River (b) and the Ganjiang River Basin (c). The explanations of the box-and-whisker plots can be found in Fig. 8.
ent formulations and input data reliabilities still exist in the three
catchments, especially for summer months in the Luanhe River
Basin and spring months in the Source Region of the Yellow River
(see Figs. 9 and 10).
5.3. The reasons for different uncertainty of runoff projections between
various climatic regions
The different impacts of PET projection approaches on future
runoff between various climatic regions were found in this study
(Figs. 9 and 10). On the one hand, the obvious uncertainties in runoff projection over the Luanhe River Basin and the Source Region of
the Yellow River were derived from the greater differences of
future PET series. On the other hand, as the other extremely important input variable of the hydrological model, precipitation plays a
dominant role in runoff change (Zhang et al., 2011; Croitoru et al.,
2013; Liu et al., 2013; Thompson et al., 2014; Guo et al., 2015). It is
perhaps not surprising that the higher proportion of precipitation
in the catchments with smaller aridity index (e.g., the Luanhe River
Basin and the Source Region of the Yellow River) are consumed by
evapotranspiration, giving rise to more noticeable impacts of different PET approaches on runoff projection in these catchments
compared with the other catchments with bigger aridity index.
Meanwhile, the impact of precipitation that might temper the
influence of different PET projection approaches on runoff when
the contribution of precipitation to runoff is greater than the contribution of PET to runoff. Here, the contribution analysis based on
climatic elasticity method was used to further explain the different
uncertainty of runoff projections between various climatic regions.
As a hydroclimatic property (Xu et al., 2014), the climate elasticity
of runoff is defined as the proportional change in runoff due to the
proportional change in one or more climatic variables
(Sankarasubramanian et al., 2001; Fu et al., 2007). Budyko hypothesis (Budyko, 1948, 1974) states that the annual water balance can
311
W. Wang et al. / Journal of Hydrology 555 (2017) 298–313
Fig. 10. The seasonal (a) and annual (b) changes in runoff for the different PET methods over the three catchments.
Table 6
The attribution analysis for the three catchments. The variables in this table are calculated based on the Wang et al. (2016b).
Catchment
Luanhe River Basin
Source Region of the Yellow River
Ganjiang River Basin
Breakpoint
1979
1989
1991
Elasticity
Change from period-1 to period-2
Contribution
(%)
eP
ePET
ex
DR
DP
DPET
Dx
gP
gPET
gx
gP - gPET
2.3695
1.4893
1.5723
1.3695
0.4893
0.5723
2.6844
2.2951
0.6317
39.6
26.0
130.8
38.0
22.4
229.3
37.7
24.2
88.2
0.3867
0.0452
0.3429
30.9
50.2
142.7
8.8
7.9
27.9
77.9
44.0
71.0
39.7
42.3
114.8
be expressed as a function of available water and energy. As one
widely used equation in many curves of Budyko assumptions, Fu
equation (Fu, 1981) was selected to quantitatively analyze the
impacts of changes in the climate (precipitation and potential
evapotranspiration) and catchment characteristics (x) on runoff
change in this section.
Table 6 shows the results of the contribution analysis for the
three catchments over the period of 1961–2000. In the Luanhe
River Basin, precipitation and x presented positive contributions
to the runoff change with contribution rates of 30.9% and 77.9%,
respectively, indicating that land use/cover was the main driving
factor for the runoff decrease in the Luanhe River Basin. The value
of the contribution of x to runoff decline in the Luanhe River Basin
was slightly higher than that in the previous studies (Bao et al.,
2012; Wang et al., 2013b). Moreover, the difference of contribution
rates between precipitation and PET to runoff reduce was 39.7%.
While the contributions of precipitation, PET and x to the runoff
decrease in the Source Region of the Yellow River were 50.2%,
7.9% and 44.0%, respectively. In addition, the significant increase
in precipitation (229.3 mm) played a dominating role in the
increase of runoff with contribution rates of 142.7% in the Ganjiang
River Basin with humid climate condition. It also can be seen that
the largest difference of contribution rates between precipitation
and PET to runoff change was occurred in the Ganjiang River Basin
(114.8%), followed by the Source Region of the Yellow River
(42.3%). Generally, the results of contribution analysis suggest that
the more humid the area, the greater impact of precipitation on
runoff change than the influence of PET on runoff change. Therefore, the more noticeable uncertainties were found in the Luanhe
River Basin and the Source Region of the Yellow River with relatively dry climate feature compared with that in the Ganjiang River
Basin.
6. Conclusion
In this study, based on the abcd monthly hydrological model,
we explored the influences of different PET inputs estimated by
four different approaches on runoff projection in three catchments
(i.e., the Luanhe River Basin, the Source Region of the Yellow River
and the Ganjiang River Basin) representing a large climatic diversity. Generally, although the similar monthly patterns of future
PET during the period of 2021–2050 were provided by using the
more physically based Penman (PN) equation with less reliable
input variables, more empirical radiation-based Priestley-Taylor
(PT) equation with relatively dependable downscaled data,
temperature-based Hamon (HM) equation with the most reliable
downscaled temperature, and statistical downscaling method with
directly choosing PET calculated by Penman (PN) equation as predictand, the obvious differences in PET magnitude estimated by
four approaches still existed in all catchments, especially for spring
and summer months in the Luanhe River Basin and the Source
Region of the Yellow River with relatively dry climate feature.
More importantly, there were apparent uncertainties in runoff projection in the three catchments reflected by obvious discrepancy in
magnitude of change in runoff and even the diverse change direction for summer months in the Luanhe River Basin and spring
months in the Source Region of the Yellow River. Meanwhile, the
312
W. Wang et al. / Journal of Hydrology 555 (2017) 298–313
more noticeable uncertainties were generally found in the Luanhe
River Basin and the Source Region of the Yellow River with smaller
aridity index compared with that in the Ganjiang River Basin. Apart
from the relatively smaller uncertainties in more humidity Ganjiang River Basin, the fact that the runoff change was more attributed to the precipitation in Ganjiang River Basin compared with
the other two basins should be the more important reason, which
was revealed by the contribution analysis based on climatic elasticity method.
Acknowledgments
This work was jointly supported by the National Science Foundation of China (51779073, 51379057), the Fundamental
Research Funds for the Central Universities (2017B21414,
2015B14114), the National ‘‘Ten Thousand Program” Youth
Talent, the QingLan Project, and the Priority Academic Program
Development of Jiangsu Higher Education Institutions (PAPD).
Thanks to the National Climatic Centre (NCC) of the China Meteorological Administration (CMA) for providing the valuable meteorological data and also thanks to the Canadian Climate Data and
Scenarios (http://www.cccsn.ec.gc.ca) for offering the dataset of
NCEP reanalysis and CanESM2. Cordial thanks are extended to
the Editor in Chief, Professor Geoff Syme, and two anonymous
referees for their valuable comments, which greatly improved
the quality of the paper.
References
Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop evapotranspiration:
Guidelines for computing crop water requirements, FAO Irrig. Drain. Pap. 56,
Food and Agric. Org., U. N., Rome.
Alley, W.M., 1984. On the treatment of evapotranspiration, soil moisture
accounting, and aquifer recharge in monthly water balance models. Water
Resour. Res. 20 (8), 1137–1149.
Andreassian, V., Perrin, C., Michel, C., 2004. Impact of imperfect potential
evapotranspiration knowledge on the efficiency and parameters of watershed
models.
J.
Hydrol.
286
(1–4),
19–35.
https://doi.org/10.1016/j.
jhydrol.2003.09.030.
Arnold, J.G., Srinivasan, R., Muttiah, R.S., Williams, J.R., 1998. Large area hydrologic
modeling and assessment: part 1. Model development. J. Am. Water Resour.
Assoc. 34 (1), 73–89. https://doi.org/10.1111/j.1752-1688.1998.tb05961.x.
Bai, P., Liu, X., Yang, T., Li, F., Liang, K., Hu, S., Liu, C., 2016. Assessment of the
influences of different potential evapotranspiration inputs on the performance
of monthly hydrological models under different climatic conditions. J.
Hydrometeorol. 17 (8), 2259–2274. https://doi.org/10.1175/JHM-D-15-0202.1.
Bao, Z., Zhang, J., Wang, G., Fu, G., He, R., Yan, X., Jin, J., Liu, Y., Zhang, A., 2012.
Attribution for decreasing streamflow of the Haihe River basin, northern China:
climate variability or human activities? J. Hydrol. 460, 117–129. https://doi.org/
10.1016/j.jhydrol.2012.06.054.
Bates, B.C., Kundzewicz, Z.W., Wu, S., Palutikof, J.P. (Eds.), 2008. Climate change and
water. Tech. Paper Intergovern. Panel on Clim. Change. IPCC Secretariat, Geneva.
Budyko, M.I., 1974. Climate and Life. Academic Press, New York.
Budyko, M.I., 1948. Evaporation under natural conditions. Gidrometeorizdat,
Leningrad, English translation by IPST, Jerusalem.
Chen, L., Frauenfeld, O.W., 2014. Surface air temperature changes over the
twentieth and twenty-first centuries in China simulated by 20 CMIP5 models.
J. Clim. 27 (11), 3920–3937. https://doi.org/10.1175/JCLI-D-13-00465.1.
Clarke, L.E., Edmonds, J.A., Jacoby, H.D., Pitcher, H.M., Reilly, J.M., Richels, R.G., 2007.
Scenarios of greenhouse gas emissions and atmospheric concentrations,
Subreport 2.1A of Synthesis and Assessment Product 2.1 by the U.S. Climate
Change Science Program and the Subcommittee on Global Change Research,
154 pp., Dep. of Energy, Off. of Biol. and Environ. Re., Washington, D.C.
Croitoru, A.E., Chiotoroiu, B.C., Todorova, V.I., Torica, V., 2013. Changes in
precipitation extremes on the Black Sea Western Coast. Global Planet Change
102, 10–19. https://doi.org/10.1016/j.gloplacha.2013.01.004.
Das, J., Umamahesh, N.V., 2016. Downscaling monsoon rainfall over River Godavari
Basin under different climate-change scenarios. Water Resour Manag. 30,
5575–5587. https://doi.org/10.1007/s11269-016-1549-6.
Donohue, R.J., Roderick, M.L., McVicar, T.R., 2007. On the importance of including
vegetation dynamics in Budyko’s hydrological model. Hydrol. Earth Syst. Sci. 11
(2), 983–995. https://doi.org/10.5194/hess-11-983-2007.
Donohue, R.J., McVicar, T.R., Roderick, M.L., 2010. Assessing the ability of potential
evaporation formulations to capture the dynamics in evaporative demand
within a changing climate. J. Hydrol. 386 (1–4), 186–197. https://doi.org/
10.1016/j.jhydrol.2010.03.020.
Du, C., Sun, F., Yu, J., Liu, X., Chen, Y., 2016. New interpretation of the role of water
balance in an extended Budyko hypothesis in arid regions. Hydrol. Earth Syst.
Sci. 20, 393–409. https://doi.org/10.5194/hess-20-393-2016.
Federer, C.A., Vorosmarty, C., Fekete, B., 1996. Intercomparison of methods for
calculating potential evaporation in regional and global water balance models.
Water Resour. Res. 32 (7), 2315–2321. https://doi.org/10.1029/96WR00801.
Fu, B.P., 1981. On the calculation of the evaporation from land surface [in Chinese].
Sci. Atmos. Sin. 5 (1), 23–31.
Fu, G., Charles, S.P., Chiew, F.H.S., 2007. A two-parameter climate elasticity of
streamflow index to assess climate change effects on annual streamflow. Water
Resour. Res. 43 (11), W11419. https://doi.org/10.1029/2007WR005890.
Goncu, S., Albek, E., 2016. Statistical downscaling of meteorological time series and
climatic projections in a watershed in Turkey. Theor. Appl. Climatol. 126, 191–
211. https://doi.org/10.1007/s00704-015-1563-2.
Gulacha, M.M., Mulungu, D.M.M., 2016. Generation of climate change scenarios for
precipitation and temperature at local scales using SDSM in Wami-Ruvu River
Basin Tanzania. Phys. Chem. Earth, 1–11. https://doi.org/10.1016/j.
pce.2016.10.003.
Guo, S., Guo, J., Hou, Y., Xiong, L., Hong, X., 2015. Prediction of future runoff change
based on Budyko hypothesis in Yangtze River basin [in Chinese]. Adv. Water Sci.
26 (2), 151–160. https://doi.org/10.14042/j.cnki.32.1309.2015.02.001.
Haddeland, I., Clark, D.B., Franssen, W., Ludwig, F., Voss, F., Arnell, N.W., Bertrand, N.,
Best, M., Folwell, S., Gerten, D., Gomes, S., Gosling, S.N., Hagemann, S., Hanasaki,
N., Harding, R., Heinke, J., Kabat, P., Koirala, S., Oki, T., Polcher, J., Stacke, T.,
Viterbo, P., Weedon, G.P., Yeh, P., 2011. Multimodel estimate of the global
terrestrial water balance: setup and first results. J. Hydrometeorol. 12 (5), 869–
884. https://doi.org/10.1175/2011JHM1324.1.
Hamon, W.R., 1961. Estimating potential evapotranspiration. J. Hydraul. Div. Am.
Soc. Civ. Eng. 87, 107–120.
Hobbins, M.T., Ramirez, J.A., Brown, T.C., 2001. The complementary relationship in
estimation of regional evapotranspiration: an enhanced Advection-Aridity
model. Water Resour. Res. 37 (5), 1389–1403. https://doi.org/10.1029/
2000WR900359.
Huang, J., Zhang, J., Zhang, Z., Xu, C.Y., Wang, B., Yao, J., 2011. Estimation of future
precipitation change in the Yangtze River basin by using statistical downscaling
method. Stoch. Enviro. Res. Risk Asse. 25, 781–792.
Kannan, N., White, S.M., Worrall, F., Whelan, M.J., 2007. Sensitivity analysis and
identification of the best evapotranspiration and runoff options for hydrological
modelling in SWAT-2000. J. Hydrol. 332 (3–4), 456–466. https://doi.org/
10.1016/j.jhydrol.2006.08.001.
Kay, A.L., Davies, H.N., 2008. Calculating potential evaporation from climate model
data: a source of uncertainty for hydrological climate change impacts. J. Hydrol.
358 (3–4), 221–239. https://doi.org/10.1016/j.jhydrol.2008.06.005.
Khan, M.S., Coulibaly, P., Dibike, Y., 2006. Uncertainty analysis of statistical
downscaling methods. J. Hydro. 319, 357–382.
Kingston, D.G., Todd, M.C., Taylor, R.G., Thompson, J.R., Arnell, N.W., 2009.
Uncertainty in the estimation of potential evaporation under climate change.
Geophys. Res. Lett. 36, L20403. https://doi.org/10.1029/2009GL040267.
Li, N., Tang, G., Zhao, P., Hong, Y., Gou, Y., Yang, K., 2017. Statistical assessment and
hydrological utility of the latest multi-satellite precipitation analysis IMERG in
Ganjiang River basin. Atmos. Res. 183, 212–223. https://doi.org/10.1016/j.
atmosres.2016.07.020.
Liu, B., Chen, X., Lian, Y., Wu, L., 2013. Entropy-based assessment and zoning of
rainfall distribution. J. Hydrol. 490, 32–40. https://doi.org/10.1016/j.
jhydrol.2013.03.020.
Liu, C.M., Wang, Z.G., Zheng, H.X., Zhang, L., Wu, X.F., 2008. Development of hydroinformatic modelling system and its application. Sci. China Ser. E-Tech. Sci. 51
(4), 456–466. https://doi.org/10.1007/s11431-008-0040-x.
Lu, J.B., Sun, G., McNulty, S.G., Amatya, D.M., 2005. A comparison of six potential
evapotranspiration methods for regional use in the southeastern United States.
J. Am. Water Resour. Assoc. 41 (3), 621–633. https://doi.org/10.1111/j.17521688.2005.tb03759.x.
Mahmood, R., Babel, M.S., 2014. Future changes in extreme temperature events
using the statistical downscaling model (SDSM) in the trans-boundary region of
the Jhelum river basin. Weather Clim. Extremes 5–6, 56–66.
McAfee, S.A., 2013. Methodological differences in projected potential
evapotranspiration. Clim. Change 120 (4), 915–930. https://doi.org/10.1007/
s10584-013-0864-7.
McVicar, T.R., Roderick, M.L., Donohue, R.J., Li, L.T., Van Niel, T.G., Thomas, A.,
Grieser, J., Jhajharia, D., Himri, Y., Mahowald, N.M., Mescherskaya, A.V., Kruger,
A.C., Rehman, S., Dinpashoh, Y., 2012. Global review and synthesis of trends in
observed terrestrial near-surface wind speeds: implications for evaporation. J.
Hydrol. 416–417, 182–205. https://doi.org/10.1016/j.jhydrol.2011.10.024.
Moss, R.H., Edmonds, J.A., Hibbard, K.A., Manning, M.R., Rose, S.K., van Vuuren, D.P.,
Carter, T.R., Emori, S., Kainuma, M., Kram, T., Meehl, G.A., Mitchell, J.F.B.,
Nakicenovic, N., Riahi, K., Smith, S.J., Stouffer, R.J., Thomson, A.M., Weyant, J.P.,
Wilbanks, T.J., 2010. The next generation of scenarios for climate change
research and assessment. Nature 463 (7282), 747–756. https://doi.org/
10.1038/nature08823.
Oudin, L., Hervieu, F., Michel, C., Perrin, C., Andreassian, V., Anctil, F., Loumagne, C.,
2005. Which potential evapotranspiration input for a lumped rainfall-runoff
model? Part 2-Towards a simple and efficient potential evapotranspiration
model for rainfall-runoff modelling. J. Hydrol. 303 (1–4), 290–306. https://doi.
org/10.1016/j.jhydrol.2004.08.026.
Penman, H.L., 1948. Natural evaporation from open water, bare soil and grass. Proc.
R. Soc Ser. A 193, 120–145.
W. Wang et al. / Journal of Hydrology 555 (2017) 298–313
Penman, H.L., 1956. Evaporation: an introductory survey. Neth. J. Agric. Sci. 4 (1), 9–
29.
Pervez, M.S., Henebry, G.M., 2014. Projections of the Ganges-Brahmaputra
precipitation-Downscaled from GCM predictors. J. Hydrol. 517, 120–134.
https://doi.org/10.1016/j.jhydrol.2014.05.016.
Priestley, C.H.B., Taylor, R.J., 1972. On the assessment of surface heat flux and
evaporation using large-scale parameters. Mon. Weather. Rev. 100 (2), 81–92.
Randall, D.A., Wood, R.A., Bony, S., Colman, R., Fichefet, T., Fyfe, J., Kattsov, V.,
Pitman, A., Shukla, J., Srinivasan, J., Stouffer, R.J., Sumi, A., Taylor, K., 2007.
Climate models and their evaluation. In: Solomon, S. et al. (Eds.), Climate
Change 2007: The Physical Science Basis. Contribution of Working Group I to
the Fourth Assessment Report of the Intergovernmental Panel on Climate
Change. Cambridge Univ. Press, Cambridge, U.K., pp. 589–662.
Roderick, M.L., Hobbins, M.T., Farquhar, G.D., 2009. Pan evaporation trends and the
terrestrial water balance. II. Energy balance and interpretation. Geogr Compass
3 (2), 761–780. https://doi.org/10.1111/j.1749-8198.2008.00214.x.
Sankarasubramanian, A., Vogel, R.M., 2002. Annual hydroclimatology of the United
States. Water Resour. Res. 38 (6), 1083. https://doi.org/10.1029/
2001WR000619.
Sankarasubramanian, A., Vogel, R.M., Limbrunner, J.F., 2001. Climate elasticity of
streamflow in the United States. Water Resour. Res. 37 (6), 1771–1781. https://
doi.org/10.1029/2000WR900330.
Sheffield, J., Wood, E.F., Roderick, M.L., 2012. Little change in global drought over the
past
60
years.
Nature
491
(7424),
435–438.
https://doi.org/
10.1038/nature11575.
Su, F., Duan, X., Chen, D., Hao, Z., Cuo, L., 2013. Evaluation of the global climate
models in the CMIP5 over the Tibetan Plateau. J. Clim. 26 (10), 3187–3208.
https://doi.org/10.1175/JCLI-D-12-00321.1.
Tatsumi, K., Oizumi, T., Yamashiki, Y., 2015. Effects of climate change on daily
minimum and maximum temperatures and cloudiness in the Shikoku region: a
statistical downscaling model approach. Theor. Appl. Climatol. 120 (1–2), 87–
98. https://doi.org/10.1007/s00704-014-1152-9.
Taylor, K.E., Stouffer, R.J., Meehl, G.A., 2012. An overview of CMIP5 and the
experiment design. Bull. Am. Meteorol. Soc. 93, 485–498.
Thiessen, A.H., 1911. Precipitation averages for large areas. Mon. Weather. Rev. 39,
1082–1084.
Thomas, H.A., 1981. Improved methods for national water assessment, report,
contract WR15249270. U. S. Water Resour. Counc, Washington, D.C..
Thomas, H.A., Marin, C.M., Brown, M.J., Fiering, M.B. (1983), Methodology for water
resource assessment, report to U. S. Geological Survey, Rep. NTIS 84124163,
Natl. Tech. Info. Serv. Springfield, VA.
Thompson, J.R., Green, A.J., Kingston, D.G., 2014. Potential evapotranspirationrelated uncertainty in climate change impacts on river flow: an assessment for
the Mekong River basin. J. Hydrol. 510, 259–279. https://doi.org/10.1016/j.
jhydrol.2013.12.010.
Thornthwaite, C.W., 1948. An approach toward a rational classification of climate.
Geogr. Rev. 38 (1), 55–94.
UNEP (United Nations Environment Programme), 1997. World atlas of
desertification 2ED. UNEP, London.
Vandewiele, G.L., Xu, C.Y., Win, N.L., 1992. Methodology and comparative study of
monthly water balance models in Belgium, China and Burma. J. Hydrol. 134,
315–347.
Vorosmarty, C.J., Federer, C.A., Schloss, A.L., 1998. Potential evaporation functions
compared on US watersheds: Possible implications for global-scale water
balance and terrestrial ecosystem modeling. J. Hydrol. 207 (3–4), 147–169.
https://doi.org/10.1016/S0022-1694(98)00109-7.
Wang, G., Zhang, J., Jin, J., Pagano, T., Calow, R., Bao, Z., Liu, C., Liu, Y., Yan, X., 2012a.
Assessing water resources in China using PRECIS projections and a VIC model.
Hydrol. Earth Syst. Sci. 16, 231–240. https://doi.org/10.5194/hess-16-231-2012.
Wang, G., Zhang, J., He, R., Liu, C., Ma, T., Bao, Z., Liu, Y., 2016a. Runoff sensitivity to
climate change for hydro-climatically different catchments in China. Stoch.
Enviro. Res. Risk Asse. 31 (4), 1011–1021. https://doi.org/10.1007/s00477-0161218-6.
Wang, G., Zhang, J., Xuan, Y., Liu, J., Jin, J., Bao, Z., He, R., Liu, C., Liu, Y., Yan, X., 2013a.
Simulating the impact of climate change on runoff in a typical river catchment
of the Loess Plateau, China. J. Hydrometeorol. 14 (5), 1553–1561. https://doi.
org/10.1175/JHM-D-12-081.1.
313
Wang, Q.J., 1991. The genetic algorithm and its application to calibrating conceptual
rainfall-runoff models. Water Resour. Res. 27 (9), 2467–2471.
Wang, W., Shao, Q., Peng, S., Xing, W., Yang, T., Luo, Y., Yong, B., Xu, J., 2012b.
Reference evapotranspiration change and the causes across the Yellow River
Basin during 1957–2008 and their spatial and seasonal differences. Water
Resour. Res. 48, W05530. https://doi.org/10.1029/2011WR010724.
Wang, W., Shao, Q., Yang, T., Peng, S., Xing, W., Sun, F., Luo, Y., 2013b. Quantitative
assessment of the impact of climate variability and human activities on runoff
changes: a case study in four catchments of the Haihe River basin, China.
Hydrol. Process. 27 (8), 1158–1174. https://doi.org/10.1002/hyp.9299.
Wang, W., Zou, S., Shao, Q., Xing, W., Chen, X., Jiao, X., Luo, Y., Yong, B., Yu, Z., 2016b.
The analytical derivation of multiple elasticities of runoff to climate change and
catchment characteristics alteration. J. Hydrol. 541, 1042–1056. https://doi.org/
10.1016/j.jhydrol.2016.08.014.
Wang, W., Xing, W., Shao, Q., 2015. How large are uncertainties in future projection
of reference evapotranspiration through different approaches? J. Hydrol. 524,
696–700. https://doi.org/10.1016/j.jhydrol.2015.03.033.
Wang, W., Xing, W., Shao, Q., Yu, Z., Peng, S., Yang, T., Yong, B., Taylor, J., Singh, V.P.,
2013c. Changes in reference evapotranspiration across the Tibetan Plateau:
Observations and future projections based on statistical downscaling. J.
Geophys. Res-Atmos. 118, 4049–4068. https://doi.org/10.1002/jgrd.50393.
Wang, W., Yu, Z., Zhang, W., Shao, Q., Zhang, Y., Luo, Y., Jiao, X., Xu, J., 2014.
Responses of rice yield, irrigation water requirement and water use efficiency to
climate change in China: Historical simulation and future projections. Agr.
Water Manage. 146, 249–261. https://doi.org/10.1016/j.agwat.2014.08.019.
Wang, X., Zhou, Y., 2016. Shift of annual water balance in the Budyko space for
catchments with groundwater-dependent evapotranspiration. Hydrol. Earth
Syst. Sci. 20, 3673–3690. https://doi.org/10.5194/hess-20-3673-2016.
Wen, H.Q., Zhang, X., Xu, Y., Wang, B., 2013. Detecting human influence on extreme
temperatures in China. Geophys. Res. Lett. 40, 1171–1176. https://doi.org/
10.1002/grl.50285.
Wilby, R.L., Dawson, C.W., 2004. Using SDSM version 3.1-A decision support tool for
the assessment of regional climate change impacts. User Manual.
Wilby, R.L., Dawson, C.W., 2007. SDSM 4.2-A decision support tool for the
assessment of regional climate change impacts. Lancaster University, UK.
Wilby, R.L., Dawson, C.W., Barrow, E.M., 2002. SDSM-A decision support tool for the
assessment of regional climate change impacts. Environ. Modell. Softw. 17 (2),
147–159.
Xing, W., Wang, W., Shao, Q., Taylor, J., Ding, Y., Fu, J., Feng, X., 2017. Statistical
downscaling of reference evapotranspiration in Haihe River Basin: applicability
assessment and application to future projection. Hydrolog. Sci. J. 62 (1), 15–27.
https://doi.org/10.1080/02626667.2016.1170132.
Xu, C.Y., Singh, V.P., 2005. Evaluation of three complementary relationship
evapotranspiration models by water balance approach to estimate actual
regional evapotranspiration in different climatic regions. J. Hydrol. 308 (1–4),
105–121. https://doi.org/10.1016/j.jhydrol.2004.10.024.
Xu, C., Gong, L., Jiang, T., Chen, D., Singh, V.P., 2006. Analysis of spatial distribution
and temporal trend of reference evapotranspiration and pan evaporation in
Changjiang (Yangtze River) catchment. J. Hydrol. 327 (1–2), 81–93. https://doi.
org/10.1016/j.jhydrol.2005.11.029.
Xu, X., Yang, D., Yang, H., Lei, H., 2014. Attribution analysis based on the Budyko
hypothesis for detecting the dominant cause of runoff decline in Haihe basin. J.
Hydrol. 510, 530–540. https://doi.org/10.1016/j.jhydrol.2013.12.052.
Yang, R., Hu, L., Shi, L., 2003. The analysis of hydrologic characteristics in the
Ganjiang River Basin [in Chinese]. J. Water Resour. Res. 24 (1), 35–38.
Yang, T., Wang, C., Chen, Y., Chen, X., Yu, Z., 2015. Climate change and water storage
variability over an arid endorheic region. J. Hydrol. 529, 330–339. https://doi.
org/10.1016/j.jhydrol.2015.07.051.
Yin, Y., Ma, D., Wu, S., Pan, T., 2015. Projections of aridity and its regional variability
over China in the mid-21st century. Int. J. Climatol. 35 (14), 4387–4398. https://
doi.org/10.1002/joc.4295.
Zhang, Q., Singh, V.P., Sun, P., Chen, X., Zhang, Z., Li, J., 2011. Precipitation and
streamflow changes in China: Changing patterns, causes and implications. J.
Hydrol. 410 (3–4), 204–216. https://doi.org/10.1016/j.jhydrol.2011.09.017.
Zhang, Y., You, Q., Chen, C., Ge, J., 2016. Impacts of climate change on streamflows
under RCP scenarios: A case study in Xin River Basin, China. Atmos. Res. 178–
179, 521–534. https://doi.org/10.1016/j.atmosres.2016.04.018.
Документ
Категория
Без категории
Просмотров
9
Размер файла
3 683 Кб
Теги
2017, 023, jhydrol
1/--страниц
Пожаловаться на содержимое документа