Water Research 177 (2020) 115767

Contents lists available at ScienceDirect

Water Research
journal homepage: www.elsevier.com/locate/watres

CN-China: Revised runoff curve number by using rainfall-runoff

events data in China
Huishu Lian a, b, Haw Yen c, Jr-Chuan Huang d, Qingyu Feng e, Lihuan Qin a,
Muhammad Amjad Bashir a, Shuxia Wu a, A-Xing Zhu f, g, Jiafa Luo h, Hongjie Di i,
Qiuliang Lei a, *, Hongbin Liu a, **
a
Key Laboratory of Nonpoint Source Pollution Control, Ministry of Agriculture, Institute of Agricultural Resources and Regional Planning, Chinese Academy
of Agricultural Sciences, Beijing, 10081, China
b
School of Environmental Science and Engineering, Southern University of Science and Technology, Shenzhen, Guangdong, 518055, China
c
Blackland Research and Extension Center, Texas A&M Agrilife Research, Texas A&M University, Temple, TX, 76502, USA
d
Department of Geography, National Taiwan University, Taipei, 10617, Taiwan
e
Research Center for Eco-Environmental Sciences, Chinese Academy of Sciences, Beijing, 100085, China
f
Key Laboratory of Virtual Geographic Environment, Nanjing Normal University, Ministry of Education, Nanjing, 210023, China
g
Department of Geography, University of Wisconsin-Madison, Madison, WI, 53706, USA
h
AgResearch Limited, Ruakura Research Centre, Hamilton, 3240, New Zealand
i
Centre for Soil and Environmental Research, Lincoln University, Lincoln, Christchurch, 7647, New Zealand

a r t i c l e i n f o a b s t r a c t

Article history: The curve number (CN) method developed by the United States Department of Agriculture (USDA) in
Received 22 September 2019 1954 is the most common adopted method to estimate surface runoff. For years, applicability of the CN
Received in revised form method is a conundrum when implementing to other countries. Specifically, countries with more
2 March 2020
complex natural environment may require more dedicated adjustments. Therefore, the current CN look-
Accepted 28 March 2020
Available online 2 April 2020
up table provided by USDA might not be appropriate and could be questionable to be applied directly to
regions elsewhere. Some studies have been conducted to modify CN values according to specified natural
characteristics in scattered regions of mainland China. However, an integral and representative work is
Keywords:
Hydrology
still not available to address potential concerns in general matters. In this study, a large set of rainfall-
Curve number runoff monitoring data were collected to adjust CN values in 55 study sites across China. The results
Surface runoff estimation showed that the revised CN values are largely different from CN look-up table provided by USDA, which
Modeling would lead to huge errors in runoff estimation. In this study, the revised CN (dubbed CN-China) provides
better reference guidelines that are suitable for most natural conditions in China. In addition, scientists
and engineers from other parts of the world can take advantage of the proposed work to enhance the
quality of future programs related to surface runoff estimation.
© 2020 Elsevier Ltd. All rights reserved.

1. Introduction
The high-density construction of meteorological stations around
the world makes it easy to acquire historical and real-time rainfall
data, but there are few stations for runoff monitoring. Runoff moni-
toring is often carried out for short-term and limited area in order to
accomplish specific research projects (Fauvel et al., 2016; Gwynne and
Glover, 1961; Hvitvedjacobsen and Yousef, 1988; Kim and Sansalone,
* Corresponding author.
** Corresponding author.
E-mail addresses: leiqiuliang@caas.cn (Q. Lei), liuhongbin@caas.cn (H. Liu).
2008). Surface runoff estimation serves as the most important role in
hydrology related research (Hawkins,1993; Jiang et al., 2012; Kim and
Sansalone, 2008; Kirchner et al., 2000; Muche et al., 2019; Steenhuis
et al.,1995; Tyagi et al., 2008; Wang and Wang, 2018). Fortunately, the
relationship between runoff and rainfall allows us to use mathe-
matical methods to estimate runoff based on readily available rainfall
data (Clyde and Work, 1943; Guo et al., 2017). The Green-Ampt
infiltration curve (Freyberg et al., 1980; Li et al., 2015; Stewart,
2018), the Philip infiltration curve (Triadis and Broadbridge, 2012),
and the Horton infiltration curve methods are all for calculating runoff
(Esen, 1987; Grimaldi et al., 2013). Applications of these methods are
very limited because they require many parameters and detailed soil

https://doi.org/10.1016/j.watres.2020.115767
0043-1354/© 2020 Elsevier Ltd. All rights reserved.
2 H. Lian et al. / Water Research 177 (2020) 115767
attribute data that are difficult to acquire. In 1949, Mockus proposed a
framework to predict runoff based on land use, soil, antecedent pre-
cipitation, storm duration, and annual mean temperature based on
mechanism of storage-excess flow (Mockus, 1949). The USDA Soil
Conservation Service (SCS) now the Natural Resources Conservation
Service (NRCS) proposed CN method in 1954 according to this
framework (Boughton, 1989; Hawkins et al., 2019), which was pub-
lished in the National Engineering Handbook (USDA,1954) in order to
meet the design needs of the Watershed Protection and Flood Pre-
vention Act (Public Law 83-566) (Hawkins et al., 2019). The CN
method has been widely used since then.
The conceptual and empirical CN method is simple with only
one parameter CN, making it the most widely used method for
calculating runoff based on rainfall events (D’Asaro et al., 2014;
Garen and Moore, 2005; Gaume et al., 2004; Lal et al., 2019; Wilson
et al., 2017). Besides, the CN method has been widely adopted in
many hydrological models (Hawkins et al., 2019; Soulis et al., 2017).
The CN method will be performed well if model users obtain CN
values from measured rainfall-runoff data when available (Lal et al.,
2017). CN parameters reflect the effect of soil infiltration capacity
(Ponce and Hawkins, 1996; Zoure et al., 2019). Key impact factors of
CN include soil properties, land use type, slope, antecedent mois-
ture conditions, vegetation coverage, land management practices
(Boughton, 1989, Chin, 2017, Hawkins, 1978a, b, Lal et al., 2019, Shi
et al., 2017, Zoure et al., 2019). CN values have been derived
experimentally from rainfall-runoff events measurement over a
wide range of geographic, soil, and land management conditions.
Based on the monitoring data of 150 watersheds in the U.S., sci-
entists organized a CN look-up table (USDA, 1954), which is widely
used around the world. Subsequently, Hawkins found that the
runoff calculated by this method was very sensitive to CN values
even more than rainfall depth (Hawkins, 1975). Previous study has
shown that a variation of ±10% of CN value leads to a variation
from
45% to þ55% runoff (Boughton, 1989). Therefore, the accu-
racy of the CN value plays a crucial role in the estimation of runoff.
Various applications and modifications of CN method have been
implemented in China, including finding the local initial abstrac-
tion (Ia) and CN values based on rainfall-runoff monitoring data (Fu
et al., 2011; Jiao et al., 2015; Shi et al., 2009), the effects of different
soil, land use, slope on CN values (Dong et al., 2015; Huang et al.,
2007; Huang et al., 2006), relationship between CN value and the
rainfall depth under long-term sequence and using remote sensing
to study the spatial heterogeneity of CN values (Chen et al., 2017;
Zhang et al., 2019). There is a common conundrum that scientists
could not find the corresponding CN values from CN look-up table.
The primary reasons are the complexity and interactions among
land use type, soil infiltration capacity, and climate characteristics
which could result in an enormous error between CN from look-up
table and reality. For instance, nearly 70% of the rainfall in the U.S.
will seep into the soil, while farmlands are fragmented and densely
drained in China. Infiltration rate is low due to the dense surface
irrigation and drainage system (He et al., 2001). In addition, topo-
graphic variations in China are substantially more complicated than
in the U.S. Therefore, parameters are being over-calibrated in order
to achieve better simulation performance, especially when users
adopt CN values from look-up table to hydrological models. The
embedded uncertainty with potentially large errors in surface
runoff estimation is going to affect the corresponding scientific
credibility in the following analytical processes such as the simu-
lation of nitrogen and phosphorus in watershed scale (Yen et al.,
2014). Thus, a national-scale research is urgently needed to alle-
viate the tremendous differences in CN value and the correspond-
ing predominant controlling factors.
The primary goal of this study was to adjust CN values based on
the monitored rainfall-runoff data in China. In order to provide
revised CN values that can be better representative to reflect actual
hydrological conditions in China, three objectives were defined to:
(i) calculate CN values using the monitored rainfall-runoff data
under different climate, soil, land use, and slope conditions across
mainland China (contiguous China); (ii) compare CN values ob-
tained from CN look-up table with the calculated CN values under
corresponding field conditions (soil, land use and slope); and, (iii)
identify the key factors differentiating CN values under Chinese
field conditions from those obtained by NRCS.
2. Materials and methods
2.1. Overview of SCS-CN method
The CN method was developed by USDA in 1950s originally
intended for estimating the depth of surface runoff in an ungauged
small catchment and later implemented also to non-agricultural
watersheds with other extended applications (Boughton, 1989;
Hawkins et al., 2019; Ponce and Hawkins, 1996). The simple
structure approach relies on the parameter CN, a lumped expres-
sion of a watershed potential hydrologic response. The CN method
is based on a water balance function (Eq. (1)) and two fundamental
hypotheses. Eq. (1) shows the rainfall depth equals to the summery
of Ia, cumulative infiltration (F), and direct runoff (Q). The first
hypothesis is that the ratio of direct runoff to the maximum po-
tential runoff is equal to the ratio of the amount of actual infiltration
to the potential maximum retention (Eq. (2)). The second hypoth-
esis states that the amount of initial abstraction is some fraction of
the potential maximum retention (Eq. (3)) (Mishra and Singh,
1999). The method has interpreted empirical relationships be-
tween initial abstraction (Ia) and potential maximum retention or
infiltration (S) through extensive rainfall-runoff experimental data:
Ia ¼ 0.2S. Functions for S (Eq. (4)) and CN (Eq. (5)) are as follow:
P ¼ Ia þ F þ Q (1)
Q F
¼ (2)
P
Ia S
Ia ¼ lS (3)

 1=2 
S ¼ 5 P þ 2Q
4Q 2 þ 5PQ (4)
CN ¼ 25400=ðS þ 254Þ (5)
where, P ¼ rainfall depth (mm); Ia ¼ initial abstraction of the
rainfall (mm); F ¼ cumulative infiltration excluding Ia (mm);
Q ¼ runoff depth (mm); S ¼ potential maximum retention or
infiltration (mm); and, l is the initial abstraction coefficient,
empirically l is treated as a constant 0.2, the relationship between
Ia and S is Ia ¼ lS; CN ¼ curve number, which is a dimensionless
parameter with the range from 0 to 100, and the higher the CN
value, the greater the potential of surface runoff.
The SCS model has no strict physical theory, and its runoff
calculation formula is summarized by measured data from more
than 150 small watersheds in the U.S. It reflects the empirical law of
rainfall-runoff events. The key to affect the accuracy of results is
whether the parameter CN value can accurately reflect the water-
shed characteristics. The range of CN values is 0e100. In this study,
spatial variable CN values are calculated into a single value to
represent the average condition of study sites. The arithmetic
means is used to determine the CN value. When CN values are
calculated from real storm data as outlined in the preceding
 H. Lian et al. / Water Research 177 (2020) 115767
section, a secondary relationship always happened between
decreased CN and increased rainfall depth, and finally CN approach
a constant (Hawkins, 1993). There are many methods to determine
CN values from observed rainfall-runoff data, while no agreement
has been settled because no one method shows outstanding
advantage (Lal et al., 2017). In addition, the fact that we use mean
CN value otherwise asymptotic CN value is the lack of abundant
rainfall-runoff data at each study site. What’s more, frequency
matching is used to calculate the chosen return-period runoff from
the same return-period rainfall depth (Hawkins, 1993). The rainfall
and runoff depths are collected separately and then arranged on a
rank-order to produce rainfall-runoff pairs during equal return
periods (Hawkins et al., 2009). In this work, we used the actual
rainfall-runoff events data to calculate CN values other than the
rank-order data, because we don’t have enough rainfall-runoff
events in some study sites to match rainfall-runoff data in the
same return period. Therefore, no obvious difference identified
between the two forms of rainfall-runoff data.
2.2. Data collection
In this study, rainfall-runoff monitoring data were collected
from 31 monitoring plots and 24 watersheds to calculate the CN
value. These sites located in 21 provinces in China. The distribution
of all study sites is shown in Fig. 1. Total 55 sites were selected
depending on available monitoring rainfall-runoff event data. The
land use, soil and plant type, slope, and other attributes for these
sites are showed in Table 1. In addition, the detailed data sources for
each study site are showed in Supplementary Material Table S2. The
soil and land use data are derived from soil surveys conducted by
Soil Testing and Formula Fertilization Program, while the slope and
climatic data are collected from literature review. The rainfall-
runoff data used to determine CN values were collected from
literature and Chinese Ecosystem Research Network (CERN). The
time scale of rainfall-runoff data is the day.
Fig. 1. The distribution of monitoring sites for rainfall and runoff. There are totally 55 sites, including 24 for watershed scale and 31 for plot scale.
3
2.3. Validation of the revised CN
In order to verify whether CN values calculated based on the
monitored data can well simulate the runoff in the study area, four
research sites were selected with abundant data and divided the
rainfall event into two parts. The first part was used to determine
CN values, and the second part was used to verify the validity of the
CN value by using the calculated CN value to estimate the runoff
depth and compare with the measured runoff depth. The Nash-
Sutcliffe Efficiency (NSE) (Nash and Sutcliffe, 1970) was used to
evaluate the performance of the method. The NSE was used as
indices of the agreement between the calculated and observed
values of runoff depth (perfect fit when NSE equals to 1, and the
worse value of NSE is -∞). When the Nash coefficient is above 0.6,
the calculated values were deemed qualified.
PN
ðQ
Qc Þ2i
NSE ¼ 1
P i¼1 0 (6)
N 2
i¼1 ðQ0
Q0 Þi
where, Qo ¼ observed runoff depth; Qc ¼ calculated runoff depth.
2.4. CN values obtained from the table CN entries
The CN values for these 55 study sites also obtained from the
lookup table according to land use type, management practice,
hydrological condition (HC) and hydrologic soil groups (HSG). HSG
(A/B/C/D) is classified according to soil saturated hydraulic con-
ductivity Ks (mm/h). A (Ks >180), B (18< Ks 180), C (1.8< Ks 18), D
(Ks <1.8). Ks is calculated according to the soil properties data
(Table 1) from the Soil Testing and Formula Fertilization Program.
The equation for calculating Ks is as follows:
Ks ¼ 0:056C þ 0:016s þ 0:231Om
0:693 (7)
where, C is the percentage of clay content in soil; s is the percentage
of sand in soil, and Om is the percentage of organic matter in soil.
Table 1

4
Properties for all 55 study sites. Including the number of rainfall-runoff events N, annual rainfall (mm), altitude (m), soil clay content (%), soil organic matter (SOM) (%), Ks, HSG (hydrological soil group), land use type and slope ().

ID Province Watershed Area/ Plot Area/ Time N Annual Rainfall/ Altitude/ Clay Sand SOM Ks (mm/ HSG CN CN Looked Slope Lon Lat Land use
km2 m2 mm m % % % h) Calculated up

1 Sichuan 100.00 2011 371 2033 3850 12.82 56.15 2.84 158.58 B 94.73 58 NA 101.96 29.99 alpine meadow
e2014
2 Beijing 3.22 2004 42 669 660 6.13 62.78 7.44 52.36 B 97.45 63 NA 117.18 40.58 multiple land use type
3 Yunnan 100.00 2011 115 1500 670 66.18 24.89 0.94 232.99 A 77.14 45 NA 100.68 21.59 broadleaf evergreen forest
e2014
4 Jiangsu 100.00 2004 42 1030 50 43.34 19.91 2.04 162.75 B 90.54 60 NA 120.60 28.10 broadleaf evergreen forest
e2006
5 Jiangxi 100.00 2014 60 1465 70 14.64 55.84 2.86 93.67 B 88.43 60 10 114.92 25.85 forest
6 Guangdong 100.00 2011 136 1613 NA 56.89 17.89 2.34 184.31 A 68.86 45 NA 113.33 23.33 broadleaf evergreen & deciduous scrub
e2014
7 Guizhou 1441.00 2003 5 1220 670 21.89 58.03 1.27 121.01 B 82.12 60 NA 107.29 26.09 multiple land use type
8 Guangdong 25.00 2002 5 1523 40 8.60 84.78 2.40 80.03 B 87.34 66 2 110.30 21.20 forest
9 Beijing 176.00 1982 1 537 NA 13.20 63.34 2.14 93.26 B 77.60 58 NA 117.07 40.53 multiple land use type
e2011
10 Heilongjiang 100.00 2000 36 455 176 44.89 28.54 0.82 176.00 B 85.41 66 10 123.48 47.90 pine forest
11 Beijing 50.00 1993 3 484 660 6.13 62.78 0.94 52.36 B 97.20 78 14.6 117.18 40.58 dryland
e2006
128.10 42.40 broadleaf deciduous & needleleaf evergreen

H. Lian et al. / Water Research 177 (2020) 115767

12 Jilin 100.00 2011 67 800 1090 26.06 30.94 4.06 132.00 B 84.89 66 NA
e2014 forest
13 Beijing 59.10 2006 20 473 1320 22.21 48.50 5.13 150.72 B 78.24 60 8 116.34 40.28 multiple land use type
̊
14 Fujian 100.00 2011 33 1478 5 15.12 31.58 1.79 64.37 B 83.73 66 15 119.14 26.16 orchard
15 Hubei 16.70 1989 21 1013 250 27.04 49.45 0.96 109.98 B 76.71 60 20 110.46 31.13 forest
e1996
16 Hunan 5600.00 1990 28 1474 150 27.93 29.15 3.82 133.13 B 86.59 72 NA 112.72 25.73 multiple land use type
e2009
17 Guangdong 100.00 2011 201 1737 NA 3.15 89.95 0.25 58.84 B 80.27 66 NA 112.52 23.10 broadleaf evergreen & deciduous scrub
e2014
18 Shaanxi 3.70 1958 8 500 1300 7.86 73.66 0.47 62.04 B 71.88 58 30 110.32 39.12 montane steppe
e1966
19 Shaanxi 2.00 1958 18 500 1200 13.47 55.65 1.81 82.24 B 94.60 81 15 109.30 36.80 dryland
e1966
20 Sichuan 100.00 2006 42 863 1800 41.03 31.75 1.23 143.82 B 90.26 77 5 102.18 27.54 dryland
21 Shanxi 1146.00 1982 4 510 1525 7.79 78.86 0.36 65.27 B 91.00 78 NA 112.00 39.03 multiple land use type
e1987
22 Beijing 158.00 1981 1 537 NA 13.20 63.34 2.14 93.26 B 72.89 60 NA 117.07 40.53 multiple land use type
e1994
23 Zhejiang 1444.00 2013 5 1425 650 18.98 57.59 0.66 108.21 B 87.88 76 NA 121.16 30.04 multiple land use type
24 Beijing 50.00 2001 9 508 NA 24.71 43.52 1.40 92.42 B 88.68 78 NA 116.04 40.59 dryland
e2006
25 Guangdong 100.00 1995 13 1720 NA 7.24 60.44 0.75 60.15 B 75.63 66 NA 113.37 23.07 orchard
26 Ningxia 100.00 1993 16 428 3940 9.21 62.06 0.81 90.73 B 94.75 86 10 105.30 36.00 bare land
27 Anhui 50.00 2015 6 995 30 36.35 26.11 0.89 116.06 B 87.85 81 NA 117.55 31.75 dryland
28 Shaanxi 0.50 2015 16 635 580 21.98 26.59 2.22 69.01 B 86.90 81 10 108.04 34.17 dryland
29 Hubei 0.14 2016 12 850 35 28.78 24.40 3.02 90.92 B 90.70 86 5 117.15 31.70 bare land
30 Shaanxi 3.39 1959 3 458 477 19.66 38.10 0.72 70.35 B 80.85 78 3 109.17 34.76 dryland
e2005
31 Shandong 10336.00 1997 28 671 NA 27.21 53.00 1.08 110.71 B 83.37 81 NA 118.50 36.42 multiple land use type
32 Sichuan 100.00 2013 25 826 570 13.17 67.92 0.74 82.86 B 76.29 74 6.5 105.32 31.49 cropland
33 Shandong 323.10 2004 11 612 NA 45.49 30.29 2.12 150.60 B 82.82 81 NA 117.05 36.42 multiple land use type
e2005
34 Beijing 100.00 2011 35 626 1300 20.73 41.92 0.61 97.75 B 67.71 66 NA 115.31 40.11 broadleaf deciduous scrub
e2014
35 Anhui 2670.00 1982 17 1600 100 20.49 47.11 3.36 81.00 B 75.47 75 NA 118.40 29.76 multiple land use type
36 Guizhou 100.00 2012 5 863 1380 47.34 20.06 0.78 183.36 A 44.40 45 5 105.43 27.00 broadleaf evergreen & deciduous scrub
e2014
37 Gansu 8.48 1989 5 300 1970 12.15 53.92 0.28 61.86 B 73.35 74 35 104.26 35.62 dryland
38 Jiangxi 100.00 24 1881 NA 18.75 68.25 1.97 90.81 B 77.27 78 NA 116.77 28.34 Subtropical park woodland
 H. Lian et al. / Water Research 177 (2020) 115767 5

3. Results

101.20 26.58 broadleaf deciduous & needleleaf evergreen

109.71 26.86 broadleaf deciduous & needleleaf evergreen

3.1. Relationships between rainfall and runoff events

104.02 31.67 broadleaf evergreen & deciduous scrub

There is a linear correlation between rainfall and runoff data
from 401 events (Fig. 2). Distribution of scatters in the graph is

109.71 26.86 needleleaf evergreen forest

highly concentrated. Slope of the trend line in the graph reflects
100.18 25.35 multiple land use type

multiple land use type

the runoff coefficient, which is around 0.62. The intercept repre-
sents the minimum rainfall that produces runoff, which is
20.94 mm. It was found that the SCS-CN model has better simu-
lation effect on study areas with runoff coefficient greater than 0.5

cropland
than areas with runoff coefficient less than 0.5 (Peng and You,
105.57 31.06 dryland
108.19 35.21 dryland

106.06 34.65 dryland

109.73 34.91 dryland

110.22 39.23 dryland

dryland

dryland
110.32 39.12 pasture

fallow
forest

forest

forest
2006).
In different study areas with same rainfall depth, the amount

30.07
38.26
30.07
30.84
40.97
26.98
of runoff produced was substantially different. In the same study
107.38 area, the same amount of rainfall at different times or seasons also
110.33
107.38
116.36
122.27
112.39
had a large difference in runoff generation. These two phenomena
indicated that influencing factors of the CN value are complex.
Interactions in external natural geographical conditions will cause
NA

NA

NA
NA

NA

NA

NA

NA

NA

NA
6.5

30

40

41

10

10

10
the difference in regional runoff generation ability, while the in-
ternal seasonal variation and rainfall duration will also cause the
difference in the runoff.
Relationships between CN values and runoff depth are shown
60
78
78

77
60

66

78

69

55

55

78

71
78
86
78
75
60

in Fig. 3 under different rainfall depths. According to the SCS-CN

method, larger CN values may generate greater runoff under the
same rainfall event. However, small rainfall events can still
58.78
76.29
76.01

74.69
57.68

63.06

74.03

64.07

49.47

48.89

70.74

53.03
58.80
64.16
53.23
40.92
12.80

generate runoff by proper amount of initial abstraction and the

soil infiltration ability. It indicates the research area has a strong
ability to produce runoff. Therefore, the corresponding CN value
B
B
B

B
B

B
B
B
B
B
B

could be large.
35.74 28.60 1.01 127.23

36.04 19.61 3.17 142.31

36.04 19.61 3.17 142.31

18.26 55.21 2.18 103.05

102.78

102.78

115.27
124.49
21.32 35.27 1.24 91.27
13.39 63.71 1.03 81.75
28.85 24.94 0.93 93.59

20.71 35.72 2.93 75.15

13.10 46.77 1.53 87.93

20.10 34.33 0.67 72.92

73.66 0.47 62.04

58.47

94.07

3.2. Comparison of calculated and table CN entries CN values

0.34
1.10
1.10
1.42
1.76
1.25

CN values of the 55 study sites were calculated from the

23.35
77.00
23.35
68.83
50.03
44.13

measured data of rainfall and runoff events, and the average CN

value of different events in the same study area represents the CN
31.77

31.77
13.43
27.24
30.97
7.86

6.37

parameters of this site. The calculated CN values have showed

clear differences compared to the CN values obtained from the
Note: the references for these 55 study sites are added in Supplementary Material Table S2.

table CN entries (USDA, 1954) for the study area (Fig. 4, Supple-
1997

1230

1625
2300

1300

1250
965

410

430

430

519
900
519
910

110

mentary materials Table S1). Results revealed that the calculated

NA

10

CN value ranges 12.80e97.45, while the looked-up table indicates

as 36e91. The difference between calculated and looked-up
values ranges between
40.41e34.08. Comparisons of average
value and error at each point indicated that the CN value acquired
1200

1335

1335

1071

1071
1400

1500

by table CN entries have reduced its variation. It could be sub-

547
900
556

522

909

561

440

440

440

480

stantially different by regions, which largely ignores the study

area characteristics. Moreover, the CN table CN entries provided
15

14
90

38

52

51

60

12

18
10
3

by USDA-SCS were summarized from a limited set of watershed

observations (Bartlett et al., 2016; Ogden et al., 2017). Thus, these
e2006

e1995

e2014

e2014

e2008

e1964

e2014

e2014

e1972
2004

2013
2013
1959

1987
2011

2011

1959

1959

2011

2011

1959

2006
1959
2006
1987
2008
2009

differences have supported our doubts on the applicability of the

existing table CN entries in China, and for the areas with different
characteristics in different regions, the existing looked-up table
100.00

100.00

100.00

100.00

100.00

100.00

100.00
60.00

for CN is not suitable for runoff estimation.

5.30

The blue line represents the calculated value of the monitoring

point CN, and the orange line represents the CN value obtained by
the existing look- up the table (Fig. 4). It can be found that CN
value based on the measured data has a large variation range.
Differences between the monitoring points can be clearly found,
219.00

984.00
0.41

1.14

0.33

2.54

0.86

2.54

while the orange line has a small fluctuation range, floating be-
tween 50e80. In literature, it is nearly impossible to see that the
Chongqing

Chongqing

CN value is less than 30 under actual environmental conditions,

Liaoning
40 Sichuan
41 Shaanxi

44 Sichuan

45 Shaanxi

46 Shaanxi

49 Shaanxi

Shaanxi
39 Yunnan

43 Yunnan

and the range of variation is between 30e100. Here, some study

47 Hunan

48 Hunan

Hunan
42 Gansu

Anhui

sites with CN values more than 80 and some are less than 30. After
repeated inspections of the data, the occurrence of extreme CN
50
51
52
53
54
55

values is because of the different forest type and local characters.

6 H. Lian et al. / Water Research 177 (2020) 115767
3.3. Model performance
The purpose of validation conducted in the study was to ensure
the driven results are reasonable in scientific measures. One should
keep in mind that the derived results could be biased by changing
validation standard (Yen et al., 2016). In this study, 0.6 is the value
of NSE set to be the statistical guideline of satisfactory. To figure out
the reliability of CN method, the simulated runoff was compared
with observed runoff depths. For this purpose, four sites with large
set of rainfall-runoff data were selected to validate the performance
of this model. The freedom of rainfall-runoff events to validate the
accuracy of CN parameters of these four sites is 7, 14, 8, and 14
respectively. The NSE results are 0.98, 0.76, 0.86, and 0.84 respec-
tively, far more than 0.6, which indicates that the CN method is
accurate enough to simulate runoff depth (Fig. 5).
4. Discussions
4.1. Differences of CN values between the United States and China
Studies have shown that soil moisture contents, hydrological
processes, infiltration, and other losses through rainfall have sig-
nificant spatial and temporal variability (Ponce and Hawkins, 1996;
Tramblay et al., 2010; Wang, 2018; Zeng et al., 2017). It indicated
that the forestland has lowest runoff production capacity and the
paddy field has highest (Choi et al., 2019). Other researchers also
found that the land use types with high CN values are mainly urban
area, farmland and water bodies, and the low CN values are in
forests, orchards, shrubs and grasslands (Choi et al., 2019;
Egodawatta et al., 2007; Li et al., 2018; Soulis et al., 2017; Zhang
et al., 2018). This might be associated with less vegetation cover
in urban area, farmland and water bodies compared to forests.
Although CN-method may introduce large uncertainty in runoff
estimation of paddy field, many scholars still treat paddy filed as
one kind of land use type with low infiltration capacity in order to
estimate runoff depth easily. The main character of curve number
method is its base on accumulated experience in use, mainly in the
U.S. There are many weaknesses and limitations when the original
CN values were applied in China (Boughton, 1989). Furthermore,
some watersheds performed quite differently from the basic CN
runoff response patterns, leading to great differences between
model and the actual processes. The major weakness and utmost
potential source of error is the sensitivity of estimated runoff in the
selection of the curve number.
Fig. 2. Relationships between rainfall and runoff of all monitoring sites. A total of 600
rainfall events were collected across the country.
Fig. 3. Relationships between CN values and runoff depth under different rainfall
depths.
4.2. Impact factors of the revised CN
Failure to evaluate the size of regional scale might affect the
corresponding error between actual runoff and CN parameters
simulated runoff. By summarizing the previous studies, it was
found that this model is suitable for the study of small and
medium-scale watersheds, with an area of about 0.25e1000 km2.
Even the CN method was appropriate still the dominant behavior of
CN value’s response to rainfall depth is important to calculate the
runoff depth (Muche et al., 2019; Soulis et al., 2009). Results indi-
cated a decrease in CN value with increase in rainfall but stable at
high rainfall intensity (Fig. 6). It indicates that the rainfall depth is
the primary factor to determine the CN value, and the asymptotic
constant value is the best way to identify CN parameters for a
watershed. And other studies have showed that CN values calcu-
lated from measured rainfall-runoff data vary systematically with
the rainfall depth, the determination of a single asymptotic CN
value observed for very high rainfall depths to characterize the
watershed’s runoff response (Soulis and Valiantzas, 2012). Standard
behavior of CN response for the rainfall-runoff relationship found in
our study with a distinct bias for high CN at lower rainfall depths
indicates that rainfall is another key factor in determining CN.
Rainfall duration could be an important phenomenon to in-
crease soil water content, which will be at maximum after certain
rainfall period (Thorndahl and Willems, 2008). After attaining
maximum soil water contents, CN values will be increased nearly
up to 100 indicating runoff generation capacity is also increased. As
the previous soil moisture condition (AMC) has a great influence on
the results (Lal et al., 2019), the study sites with detailed monitoring
data can be more reliable. For same test sites, runoff coefficient is
different under different rain intensity. Therefore, CN value as the
parameter of regional runoff needs to be able to represent the
runoff property of the region, which needs to be calculated. Long-
term monitoring of rainfall in different seasons and intensity can
reflect the overall situation of the area (Yeh et al., 2018).
Curve number varies spatially with the change in land use type,
soil type and soil humidity in the early stage (Ross et al., 2018).
Urban land and water bodies have higher CN value compared to
 H. Lian et al. / Water Research 177 (2020) 115767 7

Fig. 4. The calculated CN value and looked-up CN value for 55 study sites. Only one third of total study sites can use table CN entries (USDA and Soil Conservation Service 1954) from
USDA to gain satisfied (±5%) CN values.

Fig. 5. The results of SCS-CN method validation. Four study sites with enough data are selected to validate the accuracy of the model. Qob is the observed runoff depth, Qca is the
calculated runoff depth, and n represent the number of rainfall-runoff events.

forests, mountains, and hills (Yang and Toor, 2017). The effect of soil
properties on CN is divided into four categories according to the
permeability of soil. American soil experts initially based on more
than 14,000 soil data classified soils with similar runoff generation
capacity called the hydrological soil group (HSG) (Ross et al., 2018;
Stewart et al., 2012). The runoff generation capacities from low to
high are divided into four categories: A, B, C, and D. Detailed clas-
sification criteria showed in the US National Engineering Hand-
book, 1972 (USDA, 1972).
4.3. Shortcomings and future perspectives
The main limitation of our study is the ignorance of antecedent
moisture condition (AMC). Because the rainfall-runoff event data
we used did not have the rainfall data of the days before the event
occurred, it was impossible to know the soil moisture conditions in
the early stage of the event. Another limitation is the number of
rainfall-runoff events is still not large enough to reflect the more
detailed hydrological conditions for some study sites. In this study,
we have tried our best to collect more data in order to cover more
regions of China and collect more rainfall-runoff events to calculate
accurate results.
The runoff calculated by the SCS-CN method is known to be
sensitive to CN values, which is the largest potential source of error.
Low frequency and poor resolution of monitoring data will result in
poor estimation of runoff. Even with years of data available for
calibration of the curve numbers, substantial errors may still occur
(Boughton, 1989). Obviously, we are getting more and more accu-
rate and high-resolution rainfall data through advanced observa-
tion devices. Thus, a strong runoff estimation method is urgently
needed. While it is very difficult to establish the mechanism
equation of rainfall and runoff because infiltration capacity is
affected by many factors such as soil properties, vegetation
coverage, land use type, climate, etc. Fortunately, we have such a
simple empirical method to estimate runoff and meet the needs of
massive simulations. There are several studies mapping the global
8 H. Lian et al. / Water Research 177 (2020) 115767
Fig. 6. Response of CN values to rainfall intensity. Decreasing CN value with increasing
rainfall but stable at high rainfall intensity.
CN values based on Moderate Resolution Imaging Spectroradi-
ometer (MODIS), Harmonized World Soil Database (HWSD), and
Normalized Difference Vegetation Index (NDVI) data sets (Lin et al.,
2017; Zeng et al., 2017). Those methods are efficient to produce
high-resolution CN values. While the huge error will be introduced
by the original CN lookup table which may not suitable for other
countries. Our work is going to be very useful to resolve this issue
with relatively minor effort. And future research should focus on
the establishment of a dynamic runoff estimation model. The input
data of the model should include meteorological, topographical,
soil type, land use, and vegetation cover data sets.
For ungauged watersheds, CN parameters are usually estimated
by well-known handbook tables (USDA, 1972), according to the
hydrologic soil group (HSG), land use, surface condition and ante-
cedent moisture condition (AMC). In the presence of variable
conditions (soil type, land cover, and land use) within the basin, an
area-weighted average CN is often used. According to our results, it
is urgently needed a CN handbook table for different regions, which
can be established based on local rainfall runoff monitoring
network data.
5. Conclusion
The Curve Number approach is an effective, simple and widely
used method to estimate surface runoff volume. The revised CN
values calculated by using monitoring data from 55 sites in China
are considerably different from the CN values acquired from US-
based CN value lookup-table. It has been shown that the moni-
toring data can provide valuable information to calculate the CN
value. In this study, the modified CN values (CN-China) derived
from the actual rainfall-runoff data play an important role and a
very useful reference in future runoff estimation. The results of this
study provide new insight into the establishment of the CN table
that is more suitable for China and possibly many other countries
around the world. One can take advantage of the given work to
incorporate more details of the curve number approach to further
improve the corresponding accuracy and applicability.
Declaration of competing interest
The authors declare that they have no known competing
financial interests or personal relationships that could have
appeared to influence the work reported in this paper.
Acknowledgments
This study was supported by funding from the National Natural
Science Foundation of China (Grant No.: 31572208); the Special
Fund for Agro-scientific Research in the Public Interest (Grant No.:
201303089); and the Newton Fund (Grant Ref: BB/N013484/1). And
acknowledgment for the data support from Chinese Ecosystem
Research Network (CERN).
Appendix A. Supplementary data
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.watres.2020.115767.
References
Bartlett, M.S., Parolari, A.J., McDonnell, J.J., Porporato, A., 2016. Beyond the SCS-CN
method: a theoretical framework for spatially lumped rainfall-runoff response.
Water Resour. Res. 52 (6), 4608e4627.
Boughton, W.C., 1989. A review of the USDA SCS curve number method. Aust. J. Soil
Res. 27 (3), 511e523.
Chen, H.L., Liang, Z.Y., Liu, Y., Liang, Q.H., Xie, S.G., 2017. Integrated remote sensing
imagery and two-dimensional hydraulic modeling approach for impact evalu-
ation of flood on crop yields. J. Hydrol. 553, 262e275.
Chin, D.A., 2017. Estimating the parameters of the curve number model. J. Hydrol.
Eng. 22 (6), 06017001.
Choi, D., Park, H., Kim, Y.J., Jung, J.W., Choi, W.J., Her, Y.G., Yoon, K.S., 2019. Curve
numbers for rice paddies with different water management practices in korea.
J. Irrigat. Drain. Eng. 145 (5), 06019003.
Clyde, G.D., Work, R.A., 1943. Precipitation runoff relationships as a basis for water-
supply forecasting. Trans. Am. Geophys. Union 24, 43e49.
D’Asaro, F., Grillone, G., Hawkins, R.H., 2014. Curve number: empirical evaluation
and comparison with curve number handbook tables in sicily. J. Hydrol. Eng. 19
(12), 04014035.
Dong, L.H., Xiong, L.H., Lall, U., Wang, J.W., 2015. The effects of land use change and
precipitation change on direct runoff in Wei River watershed, China. Water Sci.
Technol. 71 (2), 289e295.
Egodawatta, P., Thomas, E., Goonetilleke, A., 2007. Mathematical interpretation of
pollutant wash-off from urban road surfaces using simulated rainfall. Water
Res. 41 (13), 3025e3031.
Esen, I.I., 1987. Least-squares estimates of the Horton infiltration parameters. Soil
Sci. 144 (1), 6e10.
Fauvel, B., Cauchie, H.M., Gantzer, C., Ogorzaly, L., 2016. Contribution of hydrological
data to the understanding of the spatio-temporal dynamics of F-specific RNA
bacteriophages in river water during rainfall-runoff events. Water Res. 94,
328e340.
Freyberg, D.L., Reeder, J.W., Franzini, J.B., Remson, I., 1980. Application of the green-
ampt model to infiltration under time-dependent surface-water depths. Water
Resour. Res. 16 (3), 517e528.
Fu, S., Zhang, G., Wang, N., Luo, L., 2011. Initial abstraction ratio in the scs-cn method
in the loess plateau of China. Transact. Asabe 54 (1), 163e169.
Garen, D.C., Moore, D.S., 2005. Curve number hydrology in water quality modeling:
uses, abuses, and future directions. J. Am. Water Resour. Assoc. 41 (2), 377e388.
Gaume, E., Livet, M., Desbordes, M., Villeneuve, J.P., 2004. Hydrological analysis of
the river Aude, France, flash flood on 12 and 13 November 1999. J. Hydrol. 286
(1e4), 135e154.
Grimaldi, S., Petroselli, A., Romano, N., 2013. Green-Ampt Curve-Number mixed
procedure as an empirical tool for rainfall-runoff modelling in small and
ungauged basins. Hydrol. Process. 27 (8), 1253e1264.
Guo, A.J., Chang, J.X., Liu, D.F., Wang, Y.M., Huang, Q., Li, Y.Y., 2017. Variations in the
precipitation-runoff relationship of the weihe river basin. Nord. Hydrol 48 (1),
295e310.
Gwynne, M.D., Glover, J., 1961. Light rainfall and plant survival - meaurement of
stem flow run-off. Nature 191 (479), 1321. &.
Hawkins, R.H., 1975. The importance of accurate curve numbers in the estimation of
storm runoff. Water Resour. Bull. 11 (5), 887e891.
Hawkins, R.H., 1978. Infiltration formula based on scs curve number. J. Irrigation
and Drainage Division-Asce 104 (4), 464e467.
Hawkins, R.H., 1978. Runoff curve numbers with varying site moisture. J. Irrigation
and Drainage Division-Asce 104 (4), 389e398.
Hawkins, R.H., 1993. Asymptotic determination of runoff curve numbers from data.
J. Irrigat. Drain. Eng. 119 (2), 334e345.
Hawkins, R.H., Theurer, F.D., Rezaeianzadeh, M., 2019. Understanding the basis of
the curve number method for watershed models and TMDLs. J. Hydrol. Eng. 24
(7), 06019003.
Hawkins T.W., R.H., Ward, T.J., Woodward, D.E., Van Mullem, J.A., 2009. Curve
Number Hydrology State of the Practice. American Society of Civil Engineers,
Reston Va, 106pp.
He, B., Zhou, N., Gao, X., 2001. Precipitation-runoff relationship in farmland
nonpoint source pollution research: amending coefficient of SCS hydrologic
 H. Lian et al. / Water Research 177 (2020) 115767
method. Res. Environ. Sci. 14 (3), 49e51.
Huang, M.B., Gallichand, J., Wang, Z.L., Goulet, M., 2006. A modification to the Soil
Conservation Service curve number method for steep slopes in the Loess
Plateau of China. Hydrol. Process. 20 (3), 579e589.
Huang, M.B., Gallichand, J., Dong, C.Y., Wang, Z.L., Shao, M.G., 2007. Use of soil
moisture data and curve number method for estimating runoff in the Loess
Plateau of China. Hydrol. Process. 21 (11), 1471e1481.
Hvitvedjacobsen, T., Yousef, Y.A., 1988. Analysis of rainfall series in the design of
urban drainage control-systems. Water Res. 22 (4), 491e496.
Jiang, W.Y., Haver, D., Rust, M., Gan, J., 2012. Runoff of pyrethroid insecticides from
concrete surfaces following simulated and natural rainfalls. Water Res. 46 (3),
645e652.
Jiao, P.J., Xu, D., Wang, S.L., Yu, Y.D., Han, S.J., 2015. Improved SCS-CN method based
on storage and depletion of antecedent daily precipitation. Water Resour.
Manag. 29 (13), 4753e4765.
Kim, J.Y., Sansalone, J.J., 2008. Event-based size distributions of particulate matter
transported during urban rainfall-runoff events. Water Res. 42 (10e11),
2756e2768.
Kirchner, J.W., Feng, X.H., Neal, C., 2000. Fractal stream chemistry and its implica-
tions for contaminant transport in catchments. Nature 403 (6769), 524e527.
Lal, M., Mishra, S.K., Pandey, A., Pandey, R.P., Meena, P.K., Chaudhary, A., Jha, R.K.,
Shreevastava, A.K., Kumar, Y., 2017. Evaluation of the Soil Conservation Service
curve number methodology using data from agricultural plots. Hydrogeol. J. 25
(1), 151e167.
Lal, M., Mishra, S.K., Kumar, M., 2019. Reverification of antecedent moisture con-
dition dependent runoff curve number formulae using experimental data of
Indian watersheds. Catena 173, 48e58.
Li, J., Wang, Z.G., Liu, C.M., 2015. A combined rainfall infiltration model based on
Green-Ampt and SCS-curve number. Hydrol. Process. 29 (11), 2628e2634.
Li, C.L., Liu, M., Hu, Y.M., Shi, T., Zong, M., Walter, M.T., 2018. Assessing the impact of
urbanization on direct runoff using improved composite CN method in a large
urban area. Int. J. Environ. Res. Publ. Health 15 (4), 775.
Lin, W., Yang, F., Zhou, L., Xu, J.G., Zhang, X.Q., 2017. Using modified Soil Conser-
vation Service curve number method to simulate the role of forest in flood
control in the upper reach of the Tingjiang River in China. J. Mt. Sci. 14 (1), 1e14.
Mishra, S.K., Singh, V.P., 1999. Another look at scs-cn method. J. Hydrol. Eng. 4 (3),
257e264.
Mockus, V., 1949. Estimation of total (and peak rates of) surface runoff for individual
storms. Exhibit A of Appendix B, Interim Survey Report, Grand (Neosho) River
Watershed. USDA Soil Conservation Service.
Muche, M.E., Hutchinson, S.L., Hutchinson, J.M.S., Johnston, J.M., 2019. Phenology-
adjusted dynamic curve number for improved hydrologic modeling. J. Environ.
Manag. 235, 403e413.
Nash, J.E., Sutcliffe, J.V., 1970. River flow forecasting through conceptual models part
I- A discussion of principles. J. Hydrol. 10 (3), 282e290.
Ogden, F.L., Hawkins, R., Walter, M.T., Goodrich, D.C., 2017. Comment on "Beyond
the SCS-CN method: a theoretical framework for spatially lumped rainfall-
runoff response. In: Bartlett, M.S. (Ed.), Water Resources Research vol. 53(7),
pp. 6345e6350 et al.
Peng, D., You, J., 2006. Application of modified SCS model into runoff simulation.
J. Water Res. Water Eng. 17, 20e24.
Ponce, V.M., Hawkins, R.H., 1996. Runoff curve number: has it reached maturity?
J. Hydrol. Eng. 1 (1), 11e19.
Ross, C.W., Prihodko, L., Anchang, J., Kumar, S., Ji, W.J., Hanan, N.P., 2018. HYS-
OGs250m, global gridded hydrologic soil groups for curve-number-based runoff
modeling. Scientific Data 5, 150091.
Shi, Z.H., Chen, L.D., Fang, N.F., Qin, D.F., Cai, C.F., 2009. Research on the SCS-CN
initial abstraction ratio using rainfall-runoff event analysis in the Three
Gorges Area, China. Catena 77 (1), 1e7.
Shi, W.H., Huang, M.B., Gongadze, K., Wu, L.H., 2017. A modified SCS-CN method
incorporating storm duration and antecedent soil moisture estimation for
runoff prediction. Water Resour. Manag. 31 (5), 1713e1727.
Soulis, K.X., Valiantzas, J.D., 2012. SCS-CN parameter determination using rainfall-
runoff data in heterogeneous watersheds - the two-CN system approach.
Hydrol. Earth Syst. Sci. 16 (3), 1001e1015.
9
Soulis, K.X., Valiantzas, J.D., Dercas, N., Londra, P.A., 2009. Investigation of the direct
runoff generation mechanism for the analysis of the SCS-CN method applica-
bility to a partial area experimental watershed. Hydrol. Earth Syst. Sci. 13 (5),
605e615.
Soulis, K.X., Ntoulas, N., Nektarios, P.A., Kargas, G., 2017. Runoff reduction from
extensive green roofs having different substrate depth and plant cover. Ecol.
Eng. 102, 80e89.
Steenhuis, T.S., Winchell, M., Rossing, J., Zollweg, J.A., Walter, M.F., 1995. Scs runoff
equation revisited for variable-source runoff areas. J. Irrigation Drain. Eng. Asce
121 (3), 234e238.
Stewart, R.D., 2018. A dynamic multidomain green-ampt infiltration model. Water
Resour. Res. 54 (9), 6844e6859.
Stewart, D., Canfield, E., Hawkins, R., 2012. Curve number determination methods
and uncertainty in hydrologic soil groups from semiarid watershed data.
J. Hydrol. Eng. 17 (11), 1180e1187.
Thorndahl, S., Willems, P., 2008. Probabilistic modelling of overflow, surcharge and
flooding in urban drainage using the first-order reliability method and
parameterization of local rain series. Water Res. 42 (1e2), 455e466.
Tramblay, Y., Bouvier, C., Martin, C., Didon-Lescot, J.F., Todorovik, D., Domergue, J.M.,
2010. Assessment of initial soil moisture conditions for event-based rainfall-
runoff modelling. J. Hydrol. 387 (3e4), 176e187.
Triadis, D., Broadbridge, P., 2012. The Green-Ampt limit with reference to infiltra-
tion coefficients. Water Resour. Res. 48. W07515.
Tyagi, J.V., Mishra, S.K., Singh, R., Singh, V.P., 2008. SCS-CN based time-distributed
sediment yield model. J. Hydrol. 352 (3e4), 388e403.
USDA, Soil Conservation Service (SCS), 1954. National Engineering Handbook.
Section 4. Hydrology. U.S Development of agriculture, Washington, D.C.
USDA, Soil Conservation Service (SCS), 1972. National Engineering Handbook. Sec-
tion 4. Hydrology. U.S Development of agriculture, Washington, D.C.
Wang, D.B., 2018. A new probability density function for spatial distribution of soil
water storage capacity leads to the SCS curve number method. Hydrol. Earth
Syst. Sci. 22 (12), 6567e6578.
Wang, S., Wang, H., 2018. Extending the Rational Method for assessing and devel-
oping sustainable urban drainage systems. Water Res. 144, 112e125.
Wilson, L.E., Ramirez-Avila, J.J., Hawkins, R.H., 2017. Runoff Curve Number Estima-
tion for Agricultural Systems in the Southern Region of USA. World Environ-
mental and Water Resources Congress 2017: Watershed Management,
Irrigation and Drainage, and Water Resources Planning and Management,
pp. 353e366.
Yang, Y.Y., Toor, G.S., 2017. Sources and mechanisms of nitrate and orthophosphate
transport in urban stormwater runoff from residential catchments. Water Res.
112, 176e184.
Yeh, T.C., Liao, C.S., Chen, T.C., Shih, Y.T., Huang, J.C., Zehetner, F., Hein, T., 2018.
Differences in N loading affect DOM dynamics during typhoon events in a
forested mountainous catchment. Sci. Total Environ. 633, 81e92.
Yen, H., Wang, X.Y., Fontane, D.G., Harmel, R.D., Arabi, M., 2014. A framework for
propagation of uncertainty contributed by parameterization, input data, model
structure, and calibration/validation data in watershed modeling. Environ.
Model. Software 54, 211e221.
Yen, H., Hoque, Y.M., Wang, X.Y., Harmel, R.D., 2016. Applications of explicitly
incorporated/post-processing measurement uncertainty in watershed
modeling. J. Am. Water Resour. Assoc. 52 (2), 523e540.
Zeng, Z.Y., Tang, G.Q., Hong, Y., Zeng, C., Yang, Y., 2017. Development of an NRCS
curve number global dataset using the latest geospatial remote sensing data for
worldwide hydrologic applications. Remote Sense. Lett. 8 (6), 528e536.
Zhang, W., Villarini, G., Vecchi, G.A., Smith, J.A., 2018. Urbanization exacerbated the
rainfall and flooding caused by hurricane Harvey in Houston. Nature 563,
384e388.
Zhang, D.J., Lin, Q.Y., Chen, X.W., Chai, T., 2019. Improved curve number estimation
in SWAT by reflecting the effect of rainfall intensity on runoff generation. Water
11, 163.
Zoure, C., Queloz, P., Koita, M., Niang, D., Fowe, T., Yonaba, R., Consuegra, D.,
Yacouba, H., Karambiri, H., 2019. Modelling the water balance on farming
practices at plot scale: case study of Tougou watershed in Northern Burkina
Faso. Catena 173, 59e70.