You are on page 1of 10

PUBLICATIONS

Water Resources Research


RESEARCH ARTICLE An error analysis of the Budyko hypothesis for assessing the
10.1002/2014WR015451
contribution of climate change to runoff
Key Points: Hanbo Yang1, Dawen Yang1, and Qingfang Hu2
 Error analysis of the first-order Taylor
expansion of Budyko hypothesis 1
State Key Laboratory of Hydro-Science and Engineering, Department of Hydraulic Engineering, Tsinghua University,
 A new method for assessing
Beijing, China, 2Nanjing Hydraulic Research Institute, Nanjing, China
hydrologic response to climate
change

Abstract Many previous studies have evaluated the hydrologic response to climate change using the first-
Correspondence to: order approximation (first-order Taylor expansion) of the Mezentsev-Choudhury-Yang equation (formulating
H. Yang,
yanghanbo@tsinghua.edu.cn the Budyko hypothesis), which has a parameter n representing catchment characteristics. However, no studies
have paid attention to the error due to the first-order approximation. This study therefore estimates this error
Citation: to improve the theoretical framework for assessing the contribution of climate change to runoff based on the
Yang, H., D. Yang, and Q. Hu (2014), Budyko hypothesis. Specifically, the error increases when precipitation (P) decreases and potential evaporation
An error analysis of the Budyko (E0) increases, and n increases. Therefore, an increasing P or decreasing E0 leads to an underestimate of the cli-
hypothesis for assessing the
contribution of climate change to matic contributions, while a decreasing P or increasing E0 leads to an overestimate. In addition, we suggest a
runoff, Water Resour. Res., 50, new method to accurately estimate the contribution of climate change to runoff.
9620–9629, doi:10.1002/
2014WR015451.

Received 25 FEB 2014


Accepted 12 NOV 2014 1. Introduction
Accepted article online 17 NOV 2014
Published online 23 DEC 2014
During the past half century, significant climate change has been widely reported [Yang et al., 2014a, 2014b;
McVicar et al., 2012; Ren et al., 2013; Sherwood et al., 2010; Wild, 2009], which in turn has impacted the hydrologic
cycle. Consequently, many hydrologists and geoscientists have been focusing on the basic question how much
runoff change a 10% annual precipitation change will lead to [Roderick and Farquhar, 2011; Yang and Yang,
2011]. To answer this question, Schaake [1990] defined the climate elasticity of runoff (R) to precipitation (P) as
dR P
eP ðP; RÞ5  ; (1)
dP R
with dR and dP being changes in runoff and precipitation, respectively. Climate elasticity, quantifying the
sensitivity of runoff to climate variables, has been widely used to evaluate the hydrologic response to cli-
mate change [Yang et al., 2014a; Chiew, 2006; Dooge et al., 1999; Dooge, 1992; Fu et al., 2007; Ma et al., 2010;
Milly and Dunne, 2002; Sankarasubramanian et al., 2001; Sankarasubramanian and Vogel, 2003; Schaake,
1990; Xu et al., 2013; Yang and Yang, 2011; Zheng et al., 2009].
A unified approach for climate elasticity of runoff is being established based on the Budyko hypothesis [Arora,
2002; Budyko, 1948, 1974; Roderick and Farquhar, 2011; Sankarasubramanian et al., 2001; Yang and Yang, 2011],
which is clear in theory and does not depend on a large amount of historical climate and runoff data [Yang and
Yang, 2011]. Consequently, this method has been applied in many studies to evaluate the impact of climate
change on runoff [Arora, 2002; Donohue et al., 2011; Dooge et al., 1999; Liu and McVicar, 2012; Sankarasubramanian
et al., 2001; Wang et al., 2013a; Yang et al., 2014a; Zheng et al., 2009]. Furthermore, Yang and Yang [2011] theoreti-
cally derived climate elasticity based on the M-C-Y equation [Mezentsev, 1955; Choudhury, 1999; Yang et al., 2008]:
E0 P
E5 1=n ; (2)
Pn 1E0n

where E is actual evaporation, E0 denotes potential evaporation, and n represents the effects of catchment
characteristics.
In previous studies, the climate elasticity was estimated as a first-order approximation (the first-order Taylor
expansion) based on the Budyko hypothesis. No studies paid attention to the potential error due to this
approximation. Therefore, to improve the theoretical framework of attribution of hydrologic response to cli-
mate change and human activities, this study aims to evaluate this potential error.

YANG ET AL. C 2014. American Geophysical Union. All Rights Reserved.


V 9620
Water Resources Research 10.1002/2014WR015451

2. Theoretical Analysis of Errors


2.1. Theoretical Derivation
For a long time scale, the water balance equation can be simplified to P 5 E 1 R for a catchment. Based on
equation (2), we can state that:
E0 P
R5P2E5P2 1=n : (3)
P 1E0n
n

E0 P
Denoting that gðE0 ; PÞ5P2 , we derive the complete Taylor expansion as:
1=n
ðPn 1E0n Þ
 
 @ @
gðE 0 1dE0 ; P1dP Þ5gðE 0 ; P Þ1 dE0 1dP gðE 0 ; P Þ
@E0 @P
  (4)
1 @ @ 2  
1 dE0 1dP gðE 0 ; P Þ1    ;
2! @E0 @P

The runoff change due to changes in P and E0 can be estimated as



dR5gðE 0 1dE0 ; P1dP Þ2gðE 0 ; PÞ
    (5a)
@ @ 1 @ @ 2  
5 dE0 1dP gðE 0 ; P Þ1 dE0 1dP gðE 0 ; P Þ1    ;
@E0 @P 2! @E0 @P

And the change ratio can be expressed as:


   2
dR dE0 @E@ 0 1dP @P @
gðE 0 ; P Þ 1 dE0 @E@ 0 1dP @P @
gðE 0 ; P Þ
 5 1 1L    : (5b)
R gðE 0 ; P Þ 2! gðE 0 ; P Þ

The first-order approximations of equations (5a) and (5b) are:


 
@ @
dR5 dE0 1dP gðE 0 ; P Þ; (6a)
@E0 @P

and
 
dR dE0 @E@ 0 1dP @P @
gðE 0 ; P Þ
 5 : (6b)
R gðE 0 ; P Þ

Equations (6a) and (6b) have been used to evaluate the impact of climate change on runoff in many studies
[Donohue et al., 2011; Liu and McVicar, 2012; Roderick and Farquhar, 2011; Tang et al., 2013; Wang et al.,
2013a; Xu et al., 2014; Yang and Yang, 2011; Yang et al., 2014a].
Assuming that long-term water balance of a catchment will follow the existing Budyko curve if only climate
change occurs, equation (5a) estimates the contribution from climate change to runoff as

gðE 0 1dE0 ; P1dP Þ2gðE 0 ; P Þ, denoted as dR**; while equation (6a) is its first-order approximation, denoted
as dR*. Consequently, the relative error (RE) caused by the first-order approximation can be estimated as
ðdR  2dRÞ=dR  , i.e.,
 
@ @
RE512 dE0 1dP gðE 0 ; P Þ=ðgðE 0 1dE0 ; P1dP
 Þ2gðE 0 ; P ÞÞ: (7)
@E0 @P

2.2. Numerical Analysis


The relative error caused by a 10% increase/decrease in P (or E0) can be estimated as
@    ðgðE 0 ; ð160:1Þ  P Þ2gðE 0 ; P ÞÞ;
RE10%P 512 gðE 0 ; P Þ  ð60:1ÞP= (8a)
@P
and

YANG ET AL. C 2014. American Geophysical Union. All Rights Reserved.


V 9621
Water Resources Research 10.1002/2014WR015451

Figure 1. Relative error of the first-order approximation due to a 10% increase/decrease in precipitation. The relative error is defined as
1 2 dR*/dR**, with dR* being estimated by equations (6a) (a first-order approximation) and dR** being estimated by equation (5a). The rel-
ative error is calculated by equation (8a). The black lines of each subplots increment/decrement by 5%.

@
RE10%E0 512 E 0 ; P Þ  ð60:1ÞE 0 =ðgðð160:1Þ  E 0 ; P Þ2gðE 0 ; P ÞÞ:
g ð (8b)
@E0

The results have been plotted in Figures 1 and 2.

3. Case Study
3.1. 33 Catchments in the Hai River Basin
The error analysis was applied in 33 catchments of the Hai River basin, whose information was given by Xu
et al. [2014]. Those 33 catchments, which locate in the upper mountain region, belong to six subbasins,
namely Luan River, Chaobai River, Yongding River, Daqing River, Ziya River, and Zhang River. Their drainage
areas range from 785 to 25,533 km2 and aridity index (E0/P) ranges from 1.8 to 3.4. In these catchments,
there are relatively low human impacts such as dams and irrigation. Monthly discharge data from 1956 to
2005 were obtained from the Hydrological Bureau of the Ministry of Water Resources of China. Daily mete-
orological data from 1956 to 2005 for 53 stations were collected from the China Administration of Meteorol-
ogy. The potential evaporation was estimated using the Penman equation recommended by Shuttleworth
[1993] and Donohue et al. [2010]. More details of the 33 catchments were provided in Xu et al. [2014]. Substi-
tution of mean annual P, E0, and R into equation (3) obtains the parameter n value for each catchment.

3.2. 207 Catchments Across China


The error analysis was also applied in 207 catchments across China, whose information was given by Yang
et al. [2014a, 2014b]. Annual P and E0 data set covers from 1961 to 2010. The catchment extent was deter-
mined according to the third-order basins of China. Their drainage areas range from 3100 to 682,700 km2,

YANG ET AL. C 2014. American Geophysical Union. All Rights Reserved.


V 9622
Water Resources Research 10.1002/2014WR015451

Figure 2. Relative error of the first-order approximation due to a 10% increase/decrease in potential evaporation. The relative error is
defined as 1 2 dR*/dR**, with dR* being estimated by equation (6a) (a first-order approximation) and dR** being estimated by equation
(5a). The relative error is calculated by equation (8b). The black lines of each subplots increment/decrement by 5%.

aridity index (E0/P) ranges from 1.8 to 3.4, and the parameter n ranges from 0.4 to 3.8. Therein, the parameter
n for each catchment was calibrated according to mean annual P, E0, and the runoff ratio [Yang et al., 2014a].

3.3. Trend Analysis


The trends in precipitation, potential evaporation, and runoff were detected by using the Mann-Kendall
(MK) nonparametric test [Maidment, 1993]. To eliminate the effect of a serial correlation in the MK test, we
applied the trend-free prewhitening (TFPW) method, which was proposed by Kulkarni and von Storch [1995]
and modified by Yue and Wang [2002]. Furthermore, we estimated the slope of the trends as follows [Burn
and Elnur, 2002; Gan, 1998]:
 
ðxj 2xi Þ
b5Median ; for all i < j; (9)
ðj2iÞ

where b is the magnitude of the trend, and xi and xj are the ith and jth observations, respectively. A positive
value for b indicates an increasing trend, and a negative value for b indicates a decreasing trend.

3.4. Error Analysis


For each catchment, the climatic contribution to runoff was estimated by using equation (5a), denoted as
dR**; while it was approached by using the first-order approximation (equation (6a)), denoted as dR*.
Therein, the same n was used for the both equation. Furthermore, the error was calculated as dR** 2 dR*
and the relative error (RE) was calculated as (dR** 2 dR*)/dR** 5 1 2 dR*/ dR**. The results for the 33 catch-
ments of the Hai River basin and the 207 catchments across China were shown in Table 1 and Figure 3,
respectively.

YANG ET AL. C 2014. American Geophysical Union. All Rights Reserved.


V 9623
Water Resources Research 10.1002/2014WR015451

Table 1. Error Analysis of the First-Order Approximation in 33 Catchments of the Hai River Basin, China
Subbasin Number P (mm) E0 (mm) R (mm) n dP (mm) dE0 (mm) dR* (mm) dR** (mm) dR**2 dR* (mm) RE (%)
Luan River 1 611 1087 144 1.68 291.6 2163.0 216.1 216.1 0.0 0.0
2 607 1055 127 1.69 51.0 2177.3 44.5 54.3 9.8 18.0
3 583 1067 117 1.31 278.7 2144.1 219.5 219.5 0.0 0.0
4 559 1080 77 1.27 267.1 2129.6 216.5 216.5 0.0 0.0
5 515 1101 88 1.80 272.6 2103.5 212.9 212.7 0.2 22.0
6 459 1126 48 1.71 237.6 292.3 24.0 24.0 0.0 0.0
7 424 1145 39 1.51 0.0 2100.7 6.0 6.6 0.6 8.9
8 463 1132 51 0.96 263.9 252.1 228.4 227.8 0.6 22.2
9 434 1162 35 1.66 261.2 254.6 212.2 211.5 0.7 25.7
10 483 1214 61 1.63 272.4 260.7 216.2 215.3 0.9 25.8
11 486 1187 66 1.49 264.1 252.2 217.0 216.3 0.7 24.3
12 455 1231 61 1.04 239.1 253.0 213.7 213.6 0.1 21.0
Chaobai River 13 402 1107 66 1.56 240.2 255.4 27.6 27.4 0.2 22.3
14 384 1141 31 1.57 238.4 257.0 26.6 26.4 0.2 22.4
15 453 1293 76 1.39 261.2 258.2 215.2 214.6 0.6 24.2
16 432 1282 44 1.73 264.9 264.1 210.4 29.7 0.7 26.7
Yongding River 17 406 1289 34 1.54 254.9 2116.0 27.7 27.6 0.1 21.7
18 379 1318 26 1.70 251.2 2118.6 25.0 24.9 0.1 22.1
19 416 1301 31 1.52 234.1 253.4 25.7 25.6 0.1 21.9
20 398 1246 31 1.65 271.6 2112.1 210.3 29.8 0.5 25.0
21 451 1268 17 1.09 2112.9 263.4 242.9 239.9 3.0 27.5
Daqing River 22 476 1302 111 1.55 2107.0 2117.2 222.3 220.7 1.6 27.5
23 505 1309 149 1.73 2166.0 261.5 236.7 229.7 7.0 223.4
24 507 1320 55 1.53 2156.0 258.1 241.8 235.5 6.3 217.8
25 503 1320 81 1.80 2138.9 260.7 227.8 223.3 4.5 219.2
26 503 1301 63 1.54 2158.5 0.0 245.9 237.3 8.6 223.0
27 539 1390 70 1.69 2177.3 265.3 241.0 233.5 7.5 222.5
28 531 1302 46 1.46 2167.2 258.6 251.1 243.8 7.3 216.7
Ziya River 29 549 1229 83 1.36 2141.6 252.8 250.9 245.9 5.0 210.8
30 555 1208 64 2.23 291.0 299.1 212.7 211.9 0.8 26.4
31 545 1225 159 1.88 2147.1 255.1 234.3 228.8 5.5 218.9
Zhang River 32 533 1071 78 1.83 289.5 245.0 224.1 222.2 1.9 28.4
33 543 1061 47 2.31 2108.7 0.0 226.2 221.6 4.6 221.6

4. Results
Figure 1 shows that jRE10%P j increases with decreasing P and increasing E0, and an increasing n. Therein, n
has a crucial impact on jRE10%P j. jRE10%P j is less than 5% when n < 1, less than 10% when n < 2, and less
than 14% when n < 3. In the arid regions of China, mean annual P ranges from 30 to 200 mm, mean annual

Figure 3. (a) Error and (b) relative error of the first-order approximation in 207 catchments across China during the period from 1961 to 2010. The error is calculated as dR** 2 dR*, with
dR* being estimated by equation (6a) (a first-order approximation) and dR** being estimated by equation (5a). The relative error is defined as 1 2 dR*/dR**. The light grey lines represent
the national boundary or the catchment boundary.

YANG ET AL. C 2014. American Geophysical Union. All Rights Reserved.


V 9624
Water Resources Research 10.1002/2014WR015451

E0 ranges from 1200 to 1800 mm, and n is not greater than 1. Therefore, jRE10%P j ranges from 0 to 5%. In
the humid regions of China, E0 has a range of 800–1200 mm and P has a range of 800–1900 mm, and n
mainly ranges from 1.0 to 2.0. Therefore, jRE10%P j ranges from 1 to 5%. In the subhumid and subarid regions
of China, P has a range of 200–800 mm and E0 has a range of 1000–1800 mm, and n ranges from 2 to 3.
Thus, jRE10%P j ranges from 5 to 16%. RE10%P is positive when P increases and negative when P decreases.
This indicates that the first-order approximation will lead to an overestimate of the contribution to runoff
caused by P decreasing and an underestimate when P increases.
Figure 2 shows that jRE10%E0 j increases with decreasing P and increasing E0, and an increasing n. In the arid
regions of China, jRE10%E0 j is no greater than 10%. In the humid regions of China, it is no greater than 11%,
and in the subhumid and subarid regions of China, it can reach up to 21%. When E0 has a decreasing trend,
RE10%E0 is positive, and when E0 has an increasing trend, RE10%E0 is negative. This indicates that the first-
order approximation will lead to an underestimate of the contribution to runoff when E0 decreases and an
overestimate when E0 increases.
As shown in Table 1, in the 33 catchments of the Hai River basin, the first-order approximation leads to a
1–23% overestimate in 27 catchments with negative climate impacts on runoff and a 9–18% underesti-
mate in 2 catchments with positive climate impacts on runoff.
As shown in Figure 3, in the 207 catchments across China, the error of the climate impact on runoff caused by
the first-order approximation (equation (6a)) has a range from 0 to 20 mm in the period of 1961–2010; therein,
the largest mainly occur in the East China, up to 11–20 mm. The relative error caused by the first-order approx-
imation has a range from 2118% to 174%; therein, larger relative error occurs in northern China.
It is noteworthy that all of the error values (dR** 2 dR*) are positive as shown in both Table 1 and Figure 3.
It is because equation (6a) (first-order approximation) underestimates runoff change (dR* < dR**) when the
climate impact is positive; while (6a) overestimates runoff change (|dR*| > |dR**| and dR* < dR**) when the
climate impact is negative. Therefore, the error dR** 2 dR* is always positive.

5. Discussion
5.1. Climate Elasticity
Equation (5b) can be converted to:
dR dP dE0
 5eP  P 1eE0  E 0 ;
R
(10)

where the elasticity can be estimated as:


  
@ 2 @2
@   1 dP @P 1dE0 @P@E0 gðE 0 ; P Þ
@P gðE 0 ; P Þ   ;
eP 5 P1 P1 (11a)
gðE 0 ; P Þ 2! gðE 0 ; P Þ

and
  2 
@2
@
E 0 ; P Þ
g ð dE0 @
@E0 1dP @P@E E 0 ; P Þ
g ð
1 0
eE0 5 @E0   E 0 1 E 0 1    : (11b)
gðE 0 ; P Þ 2! gðE 0 ; P Þ

Equations (11a) and (11b) show how eP and eE0 vary with changing P and E0.
Bao et al. [2012a] estimated the relationship of eP to precipitation change in four catchments of the Hai River
basin using the Variable Infiltration Capacity (VIC) model, and their results show that a 10% precipitation
increase/decrease causes a 10% increase/decrease in eP . In the four catchments, n equals 3, mean
annual P ranges from 450 to 650 mm, and mean annual E0 ranges from 1200 to 1400 mm. According
to equation (11a), the increment in eP caused by a 10% precipitation increase is
   
@ 2   @ 2
D5 2!1 dP @P @
gðE 0 ; P Þ1    = @P gðE 0 ; P Þ. @P
@
g ranges from 0.04 to 0.18, @P g ranges from 0.0002
 3
to 0.0008, and @P @
g equals 131026. When P equals 500 mm, a 10% increase in P leads to

1
2! 35030:00051 3502 310-6    =0:115 10% increase in D. This is similar to the results of Bao et al.
3!
1

YANG ET AL. C 2014. American Geophysical Union. All Rights Reserved.


V 9625
Water Resources Research 10.1002/2014WR015451


@ 3

@ 2
[2012a]. In addition, the third-order 3!1 d2 P @P g equals 3% of the second-order 2!1 dP @P g. Therefore, the
third and higher order can be neglected.

5.2. Contributions of Climate Change and Human Activities


Roderick and Farquhar [2011] proposed a framework for separating contributions of climatic variations and
human activities, as follows:
@g @g @g
dR5 dE0 1 dP1 dn: (12)
@E0 @P @n

To relate runoff change to climate change and human activities, previous research [Zhan et al., 2013; Ma
@g
et al., 2008; Ye et al., 2013] has considered @E 0
dE0 1 @g
@P dP as the contribution of climate change (dRc), and dR
– dRc as the contribution of human activities (or catchment characteristics). This method leads to an overes-
timate (or underestimate) of the negative (or positive) contribution of climate change to runoff. Conse-
quently, it will lead to an underestimate (or overestimate) of the contribution of human activities. In China,
the negative contribution of climate change can be detected in most catchments [Yang et al., 2014a]. Under
these conditions, the above method will overestimate the contribution of climate change, which results in
an underestimate of the contribution of human activities.
To accurately evaluate the contribution of climate change to runoff (dRc), we suggest the following method:

dRc 5gðE 0 1dE0 ; P1dP Þ2gðE 0 ; P Þ: (13)

Furthermore, the contributions of precipitation and potential evaporation can be separated as



dRP 5gðE 0 ; P1dP Þ2gðE 0 ; P Þ; (14a)

and

E 0 1dE0 ; P Þ2gð
dRE0 5gð E 0 ; P Þ: (14b)

If knowing the total runoff change (dRt), we can estimate the contribution of human activities (dRh), as
follow
dRh 5 dRt 2dRc : (15)

Using equations (6a), (13), and (15), we separated the contributions of climate change and human activities
in the 33 catchments of the Hai River basin, which were compared with those in previous studies (shown in
Table 2). The results show that the major causes for runoff decline, which are evaluated by different meth-
ods, are roughly consistent. Moreover, equation (13) estimates less climatic contribution than equation (6a)
does, especially in the Yongding River basin and Daqing River basin, which is consistent with Table 1.
Previously, Wang and Hejazi [2011] proposed a novel method for quantifying the relative contribution of the
climate and direct human impacts on runoff (WH2011), where a precondition to estimate the contribution of
climate (DRc ) is to detect runoff change (DR), i.e., calculating the contribution of direct human impacts DRh
according to the Budyko hypothesis and detecting DR based on the runoff observations, and then estimating
DRc 5DR2DRh . In some cases, such as predicting the future runoff change, we often focus on the impact of
climate change on runoff and do not have enough information on the impact of human activities. Under
such conditions, equations (13), (14a), and (14b) are better choices than the WH2011 method.

5.3. Uncertainty of This Analysis


In this study, we assumed that long-term water balance of a catchment will follow the Budyko curve (i.e.,
the M-C-Y equation) if only climate change occurs, and further proposed the method for estimating the
contributions of climate change and human activities. Remarkable, as shown in the M-C-Y equation, the
catchment water balance will follow one Budyko curve only when n keeps constant. In fact, n also relatives
to other climate factors, such as climate seasonality [Woods, 2003; Yang et al., 2012] and mean storm depth
[Cong et al., 2014; Donohue et al., 2012]. Therefore, the climate impact estimated using equations (13) and
(10) do not include the contribution from changes in climate seasonality and mean storm depth.
Climate changes not only directly impact catchment water balance, but also indirectly impact it through
altering vegetation [Lei et al., 2014]. The parameter n was related to vegetation coverage [Li et al., 2013;

YANG ET AL. C 2014. American Geophysical Union. All Rights Reserved.


V 9626
Water Resources Research 10.1002/2014WR015451

Table 2. Attribution of Annual Runoff Change in the Hai River Basin, China
Contribution From

Human Climate
River Catchment (Drainage Area) Study Period Method Activities (%) Change (%) Source
Luan River Upstream of Luanxian 1957–2000 Hydrological model, 57–67 33–43 Wang et al. [2013b]
(44,100 km2) sensitivity analysis,
elasticity modela
Upstream of Panjiakou 1956–2005 Distributed hydrological 61 39 Xu et al. [2013]
(33,700 km2) model (GBHM)
Upstream of Taolinkou 1957–2000 VIC model 59 42 Bao et al. [2012b]
(5060 km2)
12 catchments 1956–2005 Elasticity modelb 81c 19c Xu et al. [2014]
(1086–17,100 km2) Elasticity modeld 67c 33c This study
This methode 66c 34c
Chaobai River Upstream of Miyun 1956–2005 GBHM 41 55 Ma et al. [2010]
(15,800 km2)
Chao River 1961–2001 Distributed monthly 69 35 Wang et al. [2009b]
Bai River water balance model 70 31
(DTVGM)
Chao River 1961–2001 Decomposition method 65 35 Wang and Hejazi [2011]
Bai River 53 47
Bai River 1954–2004 VIC model 60 40 Bao et al. [2012b]
(8,506 km2)
Chao River 1957–2000 Hydrological model, 54–66 34–46 Wang et al. [2013b]
(4701 km2) sensitivity analysis,
elasticity modela
4 catchments 1956–2005 Elasticity modelb 80c 20c Xu et al. [2014]
(1600–4266 km2) Elasticity modeld 72c 28c This study
This methode 73c 27c
Yongding River Upstream of Guanting 1957–2000 SWAT model 35 65 Zhang et al. [2013]
(47,016 km2)
5 catchments 1956–2005 Elasticity modelb 92c 8c Xu et al. [2014]
(785–25,533 km2) Elasticity modeld 17c 83c This study
This methode 23c 77c
Daqing River 7 catchments 1956–2005 Elasticity modelb 61c 39c Xu et al. [2014]
(1,611–11,936 km2) Elasticity modeld 43c 57c This study
This methode 51c 49c
Ziya River 3 catchments 1956–2005 Elasticity modelb 75c 25c Xu et al. [2014]
(2521–6420 km2) Elasticity modeld 74c 26c This study
This methode 76c 24c
Zhang River Upstream of Guantai 1956–2000 VIC model 74 26 Bao et al. [2012b]
(17,800 km2) Hydrological model, 64–69 31–36 Wang et al. [2013b]
sensitivity analysis,
elasticity modela
2 catchments 1956–2005 Elasticity modelb 72c 28c Xu et al. [2014]
(5060–11,250 km2) Elasticity modeld 75c 25c This study
This methode 78c 22c
a
Using a simple hydrological water balance model, sensitivity analysis, and elasticity model (a first-order approximation).
b
Dividing the annual runoff time series into two subseries using a breakpoint test and then separating the contributions from human activities and climate to the change between
the two subseries by using an elasticity model (a first-order approximation).
c
Estimating the contribution for each catchment and then evaluating that for the whole subbasin by the area weighted average method.
d
Using an elasticity model for climatic contribution, equation (6a) (a first-order approximation).
e
Using equation (13) for climatic contribution.

Yang et al., 2009] or effective rooting depth and plant root characteristics [Cong et al., 2014; Donohue et al.,
2012]. The causes for vegetation change include climate change and human activities, such as the ‘‘Grain-
for-Green’’ revegetation project in China [McVicar et al., 2007; Liu and McVicar, 2012], and the impact on veg-
etation due to increasing CO2 concentration from human activities [Donohue et al., 2013; Bond and Midgley,
2012; Iversen, 2010].
In summary, catchment water balance can be expressed using the Budyko hypothesis, i.e., E 5 E (P, E0, n),
with E (P, E0, n) being a function of P, E0, and n. Therein, n can be approached using a function of climate
and vegetation, i.e., n 5 n (C, V), with C, V representing the impacts of climate (such as climate seasonality
and mean storm depth) and vegetation, respectively. Furthermore, vegetation (V) can be related to climate
and human activities (such as the ‘‘Grain-for-Green’’ and large amount of CO2 emission), i.e., V 5 V (C, H),

YANG ET AL. C 2014. American Geophysical Union. All Rights Reserved.


V 9627
Water Resources Research 10.1002/2014WR015451

with H being the impact of human activities. Consequently, the climatic contribution to runoff should
include not only the impacts from P and E0 change, but also the impacts from the changes in other climate
characteristics (such as climate seasonality and mean storm depth) and vegetation change caused by cli-
mate change.

6. Summary
Previous researches [Roderick and Farquhar, 2011; Tang et al., 2013; Xu et al., 2014; Yang and Yang, 2011;
Yang et al., 2014a] have evaluated the hydrological response to climate change using a first-order Taylor
expansion of the Mezentsev-Choudhury-Yang equation [Choudhury, 1999; Mezentsev, 1955; Yang et al.,
2008]. To improve the framework used to evaluate hydrologic response to climate change, this study esti-
mated the errors caused by the first-order approximation (ignoring the higher orders of the Taylor expan-
sion). The results show that this method will underestimate climatic contribution to runoff when P increases
or E0 decreases. Conversely, a decrease in P or an increase in E0 will lead to an overestimate of the climatic
contribution to runoff.
In addition, we estimated the error caused by a 10% precipitation (RE10%P) or potential evaporation
(RE10%E0 ) change. jRE10%P j and jRE10%E0 j increase with decreasing P or increasing E0, and an increasing
n. In theory, the contributions of climate change to runoff can be accurately estimated as dRc 5gðE 0 1

dE0 ; P1dPÞ2g ðE 0 ; P Þ. Furthermore, the contributions of precipitation and potential evaporation can be sep-
arated as dRP 5gðE 0 ; P1dP Þ2gðE 0 ; P Þ and dRE0 5gðE 0 1dE0 ; P Þ2gðE 0 ; P Þ, where gðE0 ; PÞ5P2 n E0 Pn 1=n , with
ðP 1E0 Þ
n representing catchment characteristics. Remarkable, the parameter n also incorporates the impacts
from the changes in other climate characteristics (such as climate seasonality and mean storm
depth) and vegetation change caused by climate change. Therefore, dRc that was estimated by
keeping n constant doesn’t include these impacts from climate change.

Acknowledgments References
This research was supported by
Arora, V. K. (2002), The use of the aridity index to assess climate change effect on annual runoff, J. Hydrol., 265(1-4), 164–177.
funding from the National Natural
Bao, Z. X., J. Y. Zhang, J. F. Liu, G. Q. Wang, X. L. Yan, X. J. Wang, and L. R. Zhang (2012a), Sensitivity of hydrological variables to climate
Science Foundation of China
change in the Haihe River basin, China, Hydrol. Processes, 26(15), 2294–2306.
(51379098, 91225302, and 51025931)
Bao, Z. X., J. Y. Zhang, G. Q. Wang, G. B. Fu, R. M. He, X. L. Yan, J. L. Jin, Y. L. Liu, and A. J. Zhang (2012b), Attribution for decreasing stream-
and the Beijing Higher Education
flow of the Haihe River basin, northern China: Climate variability or human activities?, J. Hydrol., 460, 117–129.
Young Elite Teacher Project
Bond, W. J., and G. F. Midgley (2012), Carbon dioxide and the uneasy interactions of trees and savannah grasses, Philos. Trans. R. Soc. B,
(YETP0080). In addition, the
367(1588SI), 601–612.
meteorological and streamflow data
Budyko, M. I. (1948), Evaporation Under Natural Conditions, Gidrometeorizdat, Leningrad, Russia.
were provided by the China
Budyko, M. I. (1974), Climate and Life, Academic, N. Y.
Meteorological Data Sharing Service
Burn, D. H., and M. Elnur (2002), Detection of hydrologic trends and variability, J. Hydrol., 255(1-4), 107–122.
System (http://cdc.cma.gov.cn/) and
Chiew, F. H. S. (2006), Estimation of rainfall elasticity of streamflow in Australia, Hydrol. Sci. J., 51(4), 613–625.
the Bureau of Hydrology of China (not
Choudhury, B. J. (1999), Evaluation of an empirical equation for annual evaporation using field observations and results from a biophysical
for public), respectively. The authors
model, J. Hydrol., 216(1-2), 99–110.
would like to express their
Cong, Z. T., X. Y. Zhang, D. Li, H. B. Yang, and D. W. Yang (2014), Understanding hydrological trends by combining the Budyko hypothesis
appreciation to the Editors, Scott
and a stochastic soil moisture model, Hydrol. Sci. J., doi:10.1080/02626667.2013.866710, in press.
Mackay and Martyn Clark, and the
Donohue, R. J., T. R. McVicar, and M. L. Roderick (2010), Assessing the ability of potential evaporation formulations to capture the dynamics
three anonymous reviewers, whose
in evaporative demand within a changing climate, J. Hydrol., 386(1-4), 186–197.
comments and suggestions led to
Donohue, R. J., M. L. Roderick, and T. R. McVicar (2011), Assessing the differences in sensitivities of runoff to changes in climatic conditions
significant improvements in the
across a large basin, J. Hydrol., 406(3-4), 234–244.
manuscript.
Donohue, R. J., M. L. Roderick, and T. R. McVicar (2012), Roots, storms and soil pores: Incorporating key ecohydrological processes into
Budyko’s hydrological model, J. Hydrol., 436, 35–50.
Donohue, R. J., M. L. Roderick, T. R. McVicar, and G. D. Farquhar (2013), Impact of CO2 fertilization on maximum foliage cover across the
globe’s warm, arid environments, Geophys. Res. Lett., 40, 3031–3035, doi:10.1002/grl.50563.
Dooge, J. C., M. Bruen and B. Parmentier (1999), A simple model for estimating the sensitivity of runoff to long-term changes in precipita-
tion without a change in vegetation, Adv. Water Resour., 23(2), 153–163.
Dooge, J. C. I. (1992), Sensitivity of runoff to climate change: A Hortonian approach, Bull. Am. Meteorol. Soc., 73(12), 2013–2024.
Fu, G., S. P. Charles, and F. H. S. Chiew (2007), A two-parameter climate elasticity of streamflow index to assess climate change effects on
annual streamflow, Water Resour. Res., 43, W11419, doi:10.1029/2007WR005890.
Gan, T. Y. (1998), Hydroclimatic trends and possible climatic warming in the Canadian Prairies, Water Resour. Res., 34(11), 3009–3015.
Iversen, C. M. (2010), Digging deeper: Fine-root responses to rising atmospheric CO2 concentration in forested ecosystems, New Phytol.,
186(2), 346–357.
Kulkarni, A., and H. von Storch (1995), Monte Carlo experiments on the effect of serial correlation on the Mann–Kendall test of trend, Mete-
orol. Z., 4(2), 82–85.
Lei, H., D. Yang, and M. Huang (2014), Impacts of climate change and vegetation dynamics on runoff in the mountainous region of the
Haihe River basin in the past five decades, J. Hydrol., 511, 786–799.

YANG ET AL. C 2014. American Geophysical Union. All Rights Reserved.


V 9628
Water Resources Research 10.1002/2014WR015451

Li, D., M. Pan, Z. Cong, Z. Lu, and E. Wood (2013), Vegetation control on water and energy balance within the Budyko framework, Water
Resour. Res., 49, 969–976, doi:10.1002/wrcr.20107.
Liu, Q., and T. R. McVicar (2012), Assessing climate change induced modification of Penman potential evaporation and runoff sensitivity in
a large water-limited basin, J. Hydrol., 464, 352–362.
Ma, H. A., D. W. Yang, S. K. Tan, B. Gao, and Q. F. Hu (2010), Impact of climate variability and human activity on streamflow decrease in the
Miyun Reservoir catchment, J. Hydrol., 389(3-4), 317–324.
Ma, Z. M., S. Z. Kang, L. Zhang, L. Tong, and X. L. Su (2008), Analysis of impacts of climate variability and human activity on streamflow for a
river basin in arid region of northwest China, J. Hydrol., 352(3-4), 239–249.
Maidment, D. R. (1993), Handbook of Hydrology, McGraw-Hill, N. Y.
McVicar, T. R., et al. (2007), Developing a decision support tool for China’s re-vegetation program: Simulating regional impacts of afforesta-
tion on average annual streamflow in the Loess Plateau, Forest Ecol. Manag., 251(1–2), 65–81.
McVicar, T. R., et al. (2012), Global review and synthesis of trends in observed terrestrial near-surface wind speeds: Implications for evapora-
tion, J. Hydrol., 416, 182–205.
Mezentsev, V. S. (1955), More on the calculation of average total evaporation, Meteorol. Gidrol., 5, 24–26.
Milly, P. C., and K. A. Dunne (2002), Macroscale water fluxes. 2: Water and energy supply control of their interannual variability, Water
Resour. Res., 38(10), 1206, doi:10.1029/2001WR000760.
Ren, L., P. Arkin, T. M. Smith, and S. Shen (2013), Global precipitation trends in 1900–2005 from a reconstruction and coupled model simu-
lations, J. Geophys. Res. Atmos., 118, 1679–1689, doi:10.1002/jgrd.50212.
Roderick, M. L., and G. D. Farquhar (2011), A simple framework for relating variations in runoff to variations in climatic conditions and
catchment properties, Water Resour. Res., 47, W00G07, doi:10.1029/2010WR009826.
Sankarasubramanian, A., and R. M. Vogel (2003), Hydroclimatology of the continental United States, Geophys. Res. Lett., 30(7), 1363, doi:
10.1029/2002GL015937.
Sankarasubramanian, A., R. M. Vogel, and J. F. Limbrunner (2001), Climate elasticity of streamflow in the United States, Water Resour. Res.,
37(6), 1771–1781.
Schaake, J. C. (1990), From climate to flow, in Climate Change and U.S. Water Resources, edited by P. E. Waggoner, pp. 177–206, John Wiley, N. Y.
Sherwood, S. C., R. Roca, T. M. Weckwerth, and N. G. Andronova (2010), Tropospheric water vapor, convection, and climate, Rev. Geophys.,
48, RG2001, doi:10.1029/2009RG000301.
Shuttleworth, W. J. (1993), Evaporation, in Handbook of Hydrology, pp. 4.1–4.53, edited by D. R. Maidment, McGraw-Hill, N. Y.
Tang, T., Q. Tang, F. Tian, Z. Zhang, and G. Liu (2013), Responses of natural runoff to recent climatic variations in the Yellow River basin,
China, Hydrol. Earth Syst. Sci., 17, 4471–4480.
Wang, D. B., and M. Hejazi (2011), Quantifying the relative contribution of the climate and direct human impacts on mean annual stream-
flow in the contiguous United States, Water Resour. Res., 47, W00J12, doi:10.1029/2010WR010283.
Wang, G. S., J. Xia, and J. Chen (2009), Quantification of effects of climate variations and human activities on runoff by a monthly water bal-
ance model: A case study of the Chaobai River basin in northern China, Water Resour. Res., 45, W00A11, doi:10.1029/2007WR006768.
Wang, S. P., Z. Q. Zhang, T. R. McVicar, J. T. Guo, Y. Tang, and A. K. Yao (2013a), Isolating the impacts of climate change and land use change
on decadal streamflow variation: Assessing three complementary approaches, J. Hydrol., 507, 63–74.
Wang, W. G., Q. X. Shao, T. Yang, S. Z. Peng, W. Q. Xing, F. C. Sun, and Y. F. Luo (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. Processes,
27(8), 1158–1174.
Wild, M. (2009), Global dimming and brightening: A review, J. Geophys. Res., 114, D00D16, doi:10.1029/2008JD011470.
Woods, R. (2003), The relative roles of climate, soil, vegetation and topography in determining seasonal and long-term catchment dynam-
ics, Adv. Water Resour., 26(3), 295–309.
Xu, X. Y., H. B. Yang, D. W. Yang, and H. Ma (2013), Assessing the impacts of climate variability and human activities on annual runoff in the
Luan River basin, China, Hydrol. Res., 44(5), 940–952.
Xu, X. Y., D. W. Yang, H. B. Yang, and H. M. Lei (2014), Attribution analysis based on the Budyko Hypothesis for detecting the dominant
cause of runoff decline in Haihe basin, J. Hydrol., 510, 530–540.
Yang, D. W., W. W. Shao, P. Yeh, H. B. Yang, S. Kanae, and T. Oki (2009), Impact of vegetation coverage on regional water balance in the
nonhumid regions of China, Water Resour. Res., 45, W00A14, doi:10.1029/2008WR006948.
Yang, H. B., and D. W. Yang (2011), Derivation of climate elasticity of runoff to assess the effects of climate change on annual runoff, Water
Resour. Res., 47, W07526, doi:10.1029/2010WR009287.
Yang, H. B., D. W. Yang, Z. D. Lei, and F. B. Sun (2008), New analytical derivation of the mean annual water-energy balance equation, Water
Resour. Res., 44, W03410, doi:10.1029/2007WR006135.
Yang, H. B., H. F. Lv, D. W. Yang, and Q. F. Hu (2012), Seasonality of precipitation and potential evaporation and its impact on catchment
water-energy balance [in Chinese], J. Hydroelectr. Eng., 31(4), 54–59.
Yang, H. B., J. Qi, X. Y. Xu, D. W. Yang, and H. F. Lv (2014a), The regional variation in climate elasticity and climate contribution to runoff
across China, J. Hydrol., 517, 607–615.
Yang, H. B., D. W. Yang, Q. F. Hu, and H. F. Lv (2014b), Spatial variability of the trends in climatic variables China during 1961–2010, Theor.
Appl. Climatol., doi:10.1007/s00704-014-1208-x, in press.
Ye, X. C., Q. Zhang, J. Liu, X. H. Li, and C. Y. Xu (2013), Distinguishing the relative impacts of climate change and human activities on varia-
tion of streamflow in the Poyang Lake catchment, China, J. Hydrol., 494, 83–95.
Yue, S., and C. Y. Wang (2002), Applicability of prewhitening to eliminate the influence of serial correlation on the Mann-Kendall test, Water
Resour. Res., 38(6), doi:10.1029/2001WR000861.
Zhan, C. S., C. W. Niu, X. M. Song, and C. Y. Xu (2013), The impacts of climate variability and human activities on streamflow in Bai River
basin, northern China, Hydrol. Res., 44(5), 875–885.
Zhang, L. P., S. Y. Yu, Y. B. Duan, L. J. Shan, X. C. Chen, and Z. X. Xu (2013), Quantitative assessment of the effects of climate change and
human activities on runoff in the Yongding River basin [in Chinese], Adv. Clim. Change Res., 9(6), 391–397.
Zheng, H., L. Zhang, R. Zhu, C. Liu, Y. Sato, and Y. Fukushima (2009), Responses of streamflow to climate and land surface change in the
headwaters of the Yellow River Basin, Water Resour. Res., 45, W00A19, doi:10.1029/2007WR006665.

YANG ET AL. C 2014. American Geophysical Union. All Rights Reserved.


V 9629

You might also like