DOI: 10.12357/cjea.20220121
MUSHIMIYIMANA C, LIU L L, YANG Y H, LI H L, WANG L N, SHENG Z P, ITANGISHAKA A C. Drivers of evapotranspira-
tion increase in the Baiyangdian Catchment[J]. Chinese Journal of Eco-Agriculture, 2023, 31(4): 598?607
Drivers of evapotranspiration increase in the Baiyangdian Catchment
Christine Mushimiyimana1,2, LIU Linlin1,2, YANG Yonghui1,2, LI Huilong1,2, WANG Linna2, SHENG Zhuping3,
Auguste Cesar Itangishaka1,2
(1. Center for Agricultural Resources Research, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences / Key Labor-
atory of Agricultural Water Resources, Chinese Academy of Sciences / Hebei Key Laboratory of Water-saving Agriculture, Shijiazhuang
050022, China; 2. University of Chinese Academy of Sciences, Beijing 100049, China; 3. Texas A & M AgriLife Research Center, El Paso,
Texas 79927, USA)
Abstract: The Baiyangdian Catchment is facing a growing shortage of water resources. Identifying the sensitive drivers of evapotran-
spiration (ET) changes from land and crop management will be critical to understanding the reasons for mountainous runoff reduc-
tion and depletion of groundwater resources in the plain. It will also be important for making Xiong’an become a Future Example
City for green and sustainable development. In this study, remotely sensed ET data from PML V2 products with a spatial resolution of
500 m was used to analyze the trend of ET at the pixel level and to understand its influence on vegetation such as GPP (Gross Primary
Production) and NDVI (Normalized Difference Vegetation Index) under different land-use types for 2002?2018. Results showed that
there was a significant increase in ET in mountain regions and a slight increase in plain regions of the catchment. The spatial pattern
of mean annual ET was very much relevant to the changing trend of GPP and NDVI. For the whole catchment, the average increases
of ET, GPP, and NDVI were respectively 2.4 mm?a?1, 9.8 g?cm?2?a?1, and 0.0021 at an annual rate. In the mountainous region, changes
in annual precipitation and vegetation recovery together caused a total increase of ET by 56.5 mm over the period and negatively af-
fected the runoff. In the plain region, there were 3 factors influencing the change of ET. While intensification of urbanization and re-
duction in the cultivation of wheat, the water consumptive crop, had both resulted in the decrease of ET and water consumption, ET or
water consumption in most irrigated fields increased. Since the beneficial effects from urbanization and crop adjustment were not
enough to offset the increase of ET in irrigated fields, an overall ET increase of 6.4 mm over the period was found. In conclusion,
both in the mountainous and plain regions, ET increased. And therefore, more efforts are needed to control the ET increase in natural
vegetation and cropland for a green and sustainable catchment.
Keywords: Evapotranspiration; Vegetation change; Urbanization; Winter wheat; Irrigated land; Baiyangdian Catchment
Chinese Library Classification: P426.2 Open Science Identity:
白洋淀流域蒸散发增加的驱动因素
Christine Mushimiyimana1,2, 刘林林1,2, 杨永辉1,2, 李会龙1,2, 王林娜2, 盛祝平3,
Auguste Cesar Itangishaka1,2
(1. 中国科学院遗传与发育生物学研究所农业资源研究中心/中国科学院农业水资源重点实验室/河北省节水农业重点实验室
石家庄 050022 中国; 2. 中国科学院大学 北京 100049 中国; 3. 德州农工大学埃尔帕索研究中心 德克萨斯州 79927
美国)
摘 要: 白洋淀流域位于雄安新区上游, 山区植被和下垫面变化、平原区农业灌溉加大了区域蒸散发, 造成山区产
流减少和平原区地下水超采。研究区域蒸散发(ET)时空格局的演变趋势、甄别植被、作物种植结构、城市化等
对蒸散发变化的影响, 对深入揭示白洋淀流域水资源枯竭的成因, 建设绿色雄安“未来之城”具有重要意义。本研究
基于500 m空间分辨率的PML_V2遥感蒸散产品, 从像元尺度分析了2002—2018年研究区ET的变化趋势和显著
This study was financially supported by the Project from the Ministry of Science & Technology of China (2018YFE0110100) and the National Natur-
al Science Foundation of China (42171046).
Corresponding author, E-mail: yonghui.yang@sjziam.ac.cn
Received Feb. 22, 2022; accepted Sep. 2, 2022
中国生态农业学报 (中英文) ?2023年4月 ?第?31?卷 ?第?4?期
Chinese?Journal?of?Eco-Agriculture,?Apr.?2023,?31(4):?598?607
http://www.ecoagri.ac.cn
性, 揭示植被变化、冬小麦压采、城市化等对山区和平原区ET的影响。结果表明, 1)研究时段内白洋淀流域ET
和植被总初级生产力(GPP)及归一化植被指数(NDVI)均呈增加趋势, 平均增长量为2.4 mm?a?1、9.8 g?cm?2?a?1和
0.0021?a?1。2)降雨和植被恢复带来的GPP、NDVI增长是山区ET增加的主要因素, ET与GPP和NDVI的趋势变
化在空间分布上具有很好的相似性, 研究时段内山区ET增加56.5 mm。3)平原区ET受快速城市化、小麦种植面
积压减和农田ET增加3个因素影响, 虽然城市化和小麦压减都带来蒸散发减少, 但仍无法抵消农田ET增加的效
果, 平原区ET总体增长了6.4 mm。就整个流域而言, 减少山区植被和灌溉农田带来的ET增加对维持区域水资源
可持续利用和绿色发展至关重要。
关键词: 蒸散发; 植被变化; 城市化; 冬小麦; 灌区; 白洋淀流域
Evapotranspiration (ET) is the main element of the
hydrologic cycle responsible for distributing energy, wa-
ter, and carbon cycles in a terrestrial ecosystem (Jung et
al., 2010; Wegehenkel et al., 2005). It is the second-
largest part of the water cycle after precipitation (Rodell
et al., 2015; Trenberth et al., 2007). Globally, more than
two-thirds of global terrestrial precipitation is returned
into the atmosphere via ET (Oki and Kanae, 2006). In
water-limited basins where agriculture, environment, and
social economy compete for water, the ratio of water lost
through ET can be considerably high (Li et al., 2016). ET
varies with the properties of the land surface, vegetation,
and topography (e.g., slope and aspect) and directly af-
fects region and catchment water availability (Sun et al.,
2018; Zhou et al., 2015). Globally, the hydrological cycle
is modified by changing climate and human activities
(Oki and Kanae, 2006; Sherwood and Fu, 2014). There-
fore, there is an urgent need to determine the effects of
land-use change and vegetation on ET.
Recent developments in remote sensing technology
make it possible to detect where and when ET changes
across the globe (Ai et al., 2020; Anderson et al., 2012;
Pascolini-Campbell et al., 2021) and at the regional level
(Yang et al., 2014). For instance, it is noted that change
in vegetation is the main driver of ET change and hydro-
logy. Zeng et al. (2018) analyzed multiple global ET
products and noted that significant increase in ET. More
than 50% of the increase in global terrestrial ET was
driven by the greening of the earth. Chen et al. (2019)
observed that one-third of global greening was due to re-
forestation and agricultural activities in China and India.
In China, forest cover has increased from 12.0% in 1980
to 22.9% in 2018 (China Forest and Grassland Statistical
Year Book, 2019). Such a large increase could have
heavily influenced the local hydrological cycle. Li et al.
(2020) reported that due to vegetation recovery, total ET
in North China increased at 13 km3?a?1. Jiang et al. (2020)
studied the driving factors of ET in the Yellow River
Basin using the improved Shuttle Worth-Wallace model
and noted that NDVI (Normalized Difference Vegeta-
tion Index) had the greatest impact on ET.
Globally, 80% of the world population is threatened
by severe water security (V?r?smarty et al., 2010). Yang
et al. (2021) studied runoff change in 11 067 rivers and
5172 human-impacted rivers and found that 62% of the
runoff in human-influenced catchments was non-station-
ary. Aside from climate change, which can be simulated
by hydrological models (Mueller et al., 2011; Sun et al.,
2013; Yang et al., 2013), detecting the driving forces of
hydrological changes from human activity has remained
challenging (Allen et al., 2011; Anderson et al., 2012).
Remote sensing is becoming increasingly promising in
detecting a human effect (especially as land-use changes)
on ET (Feng et al., 2016; Li et al., 2017). Thus, remote
sensing can be used to analyze spatial and temporal
changes in ET to deepen the existing understanding of
human influences.
In 2017, Xiong’an was selected as a Model Future
City for sustainable and green development. However,
there is a severe water shortage in the Baiyangdian Lake
and its upstream Baiyangdian Catchment. In the 1950s
and 1960s, the Baiyangdian Lake was the largest fresh-
water lake in the Haihe Catchment and always had the
problem of floodwater control. Persistent human inter-
vention, especially in recent decades, has led to a drastic
shrink in runoff from the upstream mountain region and
groundwater depletion in the plains (Hu et al., 2012;
Moiwo et al., 2010). Since the 1990s, the lake has heav-
ily relied on over 2 km3 of external water delivery from
the Yellow River and recently from the South-to-North
Water Transfer Project from the Yangtze River. Cur-
rently, it is still unclear as to what extent of each human
第 4 期 Christine Mushimiyimana, et al: Drivers of evapotranspiration increase in the Baiyangdian Catchment 599
http://www.ecoagri.ac.cn
intervention has influenced the increase of ET and feed-
back on the water resources. Thus, the objective of the
paper was to clarify the drivers of ET increase from agri-
cultural irrigation, vegetation recovery, and other land
use influence.
1 Materials and Methods
1.1 Study area
The Baiyangdian Catchment is in the up- and
middle-reaches of Daqing River Catchment in the Haihe
River Basin. It has a total drainage area of 3.12 × 104 km2
with a mainly mountainous region in the upstream or the
west part and a plain area in the east part for agricultural
production and dense residential cities (Fig. 1a).
Agriculture is the major user of water resources in
the catchment. In the plain regions, two main crops,
winter wheat and summer maize are planted in an annual
rotation, while winter wheat grows from October to May
and summer corn from June to September. In the moun-
tainous regions, grassland and artificial forest are distrib-
uted. In the valley, irrigation land is developed mainly
for the growth of maize (see land use distribution in
Fig. 1b). Influenced by the East Asia monsoon climate,
July and August have the highest precipitation and ET.
1.2 Data sources
Even though ET data are increasingly being avail-
able, Anderson et al. (2012), by comparing MODIS ET
(1 km resolution) with Landsat ET (100 m resolution),
found that high spatial resolution products were prefer-
able for water resources assessment. Recently, several
studies (Gan et al., 2018; Zhang et al., 2019) high-
lighted the advantages of PML V2 ET product with
the spatial resolution of 500 m for accurate trend ana-
lysis. Thus, the PMLV2 ET and GPP (Gross Primary
Production) data for 2002—2018 at 500 m spatial res-
olution at an eight-day interval from the Google Earth
Engine was used for this study (Zhang et al., 2019).
The primary goal of the PM-based framework is to
get an accurate estimate of surface conductance (Gs),
which defines soil-canopy water flux. The biophysical
model for Gs was developed by Leuning et al. (2008) and
Zhang et al. (2010) to account for canopy physiological
processes and soil evaporation (PML V1 or version 1).
Then, Gan et al. (2018) coupled vegetation transpiration
with GPP using the biophysical canopy conductance (Gc)
model called PML-V2. PML-V2 ET is estimated as the
sum of evaporation from soil (Es), transpiration from
plant canopy (Et), and evaporation of intercepted precip-
itation by vegetation (Ei). PML-V2 GPP has sub-
sequently become to be known as GPP. PML-V2 is well-
calibrated using 8-day measurements at 95 widely dis-
tributed flux towers for 10 plant function types (Gan et
al., 2018).
Remote sensing NDVI at annual and monthly scales
at a spatial resolution of 1.0 km derived from the Moder-
ate Resolution Imaging Spectroradiometer (MODIS) data
was obtained for 2002?2018 (https://www.resdc.cn/).
The NDVI data was then resampled to a spatial resolu-
tion of 500 m to match the resolution of ET data.
Land use data, with a spatial resolution of 30 m ×
30 m for 2020, was obtained from the website of http://
www.globallandcover.com.
Precipitation data (2002—2018) from 21 meteor-
ological stations of the Baiyangdian Catchment was
obtained from the China Meteorological Data Service
Center (http://data.cma.cn/).
1.3 Data analysis
Considering the fact that the land use types are so
different, the whole analysis for the study was separated
into the mountainous region (as Region Ⅰ) and plain re-
gion (as Region Ⅱ), and the catchment as a whole (as
BYD). The test for trend and significance of trend was
done at pixel level and at annual level for the Regions of
Ⅰ, Ⅱ, and BYD.
Linear regression was used to detect the trend of
ET, NDVI, and GPP at pixel level. It is determined from
the slope of the least-square regression line (Ai et al.,
2020):
Slope =
n
∑n
i=1
(i Vi)
∑n
i=1
i
∑n
i=1
Vi
n
∑n
i=1
i2
∑n
i=1
i
(1)
where Vi is the value of the variable (ET, GPP, NDVI,) in
the ith year, and n is the length of time series which is
2002—2018.
In general, a slope > 0 implies an increased trend
and vice versa.
The Mann-Kendall (MK) statistical test was used to
test the significance of the trend at P≤0.05 (Kendall,
600 中国生态农业学 报 (中英文 )?2023 第 31 卷
http://www.ecoagri.ac.cn
1975; Mann, 1945).
2 Results
2.1 General distribution of ET, GPP, and NDVI in
the catchment
Fig. 2 shows the spatial distribution of average an-
nual ET, GPP, and NDVI for 2002?2018 for
RegionsⅠandⅡin the Baiyangdian Catchment. In gener-
al, ET is high in the plains, especially in irrigated crop-
lands, and low in the mountains. The mean ET values
were 516.9 mm, 575.8 mm, and 541.2 mm respectively
for the mountains (RegionⅠ), plains (RegionⅡ), and the
whole catchment region (BYD). By contrast, the average
annual precipitation were 523.7 mm, 498.9 mm, and
508.8 mm, respectively, for RegionⅠ, Ⅱ, and BYD
(Fig. 3).
Spatially, ET, GPP, and NDVI show pretty much
the same distribution. For instance, in the irrigated crop-
DEM (m)
High: 2793
Low: 0
Meteorological station
40°0′0″N
39°0′0″N
38°0′0″N
114°0′0″E 115°0′0″E 116°0′0″E 114°0′0″E 115°0′0″E 116°0′0″E
114°0′0″E 115°0′0″E 116°0′0″E 114°0′0″E 115°0′0″E 116°0′0″E
40°0′0″N
39°0′0″N
38°0′0″N
40°0′0″N
39°0′0″N
38°0′0″N
a b
0 25 50 100 km 0 25 50 100 km
Land use
Artificial surfaces
Bare land
Cultivated land
Forest
Grassland
Shrubland
Water bodies
Wetland
Fig. 1 Location of the Baiyangdian Catchment [ a: map of elevation with meteorological stations; b: map of land use types in
2020; the map is separated into the mountain region (Region Ⅰ) and plain region (Region Ⅱ) by the elevation of
100 m asl]
a b c
ET (mm?a?1 ) GPP (g?cm
?2 ?a?1 )
<400 <400
>800
400~500 400~800 0 25 50 100 km
NDVI
<0.6
0.6~0.7
0.7~0.8
>0.8
800~1200
1200~1600
1600~2000
>2000
500~600
600~700
700~800
N
Fig. 2 Spatial distribution of annual average evapotranspiration (ET, a), gross primary production (GPP, b), and
normalized difference vegetation index (NDVI, c) for the period of 2002–2018 in the Baiyangdian Catchment
第 4 期 Christine Mushimiyimana, et al: Drivers of evapotranspiration increase in the Baiyangdian Catchment 601
http://www.ecoagri.ac.cn
land (RegionⅡ), ET, GPP, and NDVI all have the
highest value. On the other hand, in the mountainous re-
gion and urban, the low-value zones of ET, GPP, and
NDVI are the same. Both in the mountainous region and
the plain, water bodies, rivers and wetlands are found to
have the highest ET value.
2.2 Comparison on the temporal trend of ET and
precipitation
Fig. 3 shows the comparison between ET and pre-
cipitation for RegionⅠ, RegionⅡ, and BYD for
2002–2018. Precipitation heavily influenced annual
variations of ET, both in the mountainous (Ⅰ) and
plain regions (Ⅱ). For example, in the wet years of 2008
and dry years like 2005 and 2006, ET followed the trend
of annual precipitation change. Driven by groundwater
irrigation, aside from 4 rich precipitation years such as
2008, 2012, 2013, and 2016, ET in the plain area was
higher than precipitation. In contrast, in most years, ET
in the mountainous or headwater region was lower than
annual precipitation, suggesting the water recharging ser-
vice from the mountains.
MK test and correlations also suggest the signific-
ant influence of precipitation on ET in the mountains of
the catchment (P≤0.05). However, there was no signi-
ficant correlation between ET and precipitation for the
plain and the whole catchment (BYD) (Fig. 3).
For the 17-years period, the mean annual ET in-
creased by 56.5 mm, 6.4 mm, and 38.1 mm respectively
for the mountain region, plain region, and the whole
catchment. This suggested an overall increase in ET de-
mand.
2.3 Temporal trend of GPP and NDVI
Fig. 4 shows the trend and significance of GPP
and NDVI for 2002–2018. Similar to ET, both GPP
and NDVI show significant increasing trends in
mountain areas (P<0.05), but not for the plain area and
the whole catchment. In the plain area, even a decreasing
trend of NDVI was found since 2008 possibly driven by
the urbanization (Fig. 4b). In the whole catchment, the
annual NDVI increased marginally by 6.9%.
2.4 Spatial distribution and temporal trends of ET,
GPP, and NDVI
Fig. 5 shows the distribution of trends of ET,
GPP, and NDVI for 2002–2018. Overall, ET had no-
ticeable increasing trends, which were significant
(3.53 mm?a?1) for the mountain area (Ⅰ), slight but not
significant (0.4 mm?a?1) for the plain area (Ⅱ), and medi-
um increase (2.4 mm?a?1) for the whole catchment. Such
increases were mainly contributed from ET increases in
the mountain area (Fig. 5 and Fig. 3).
Spatially, the distribution of temporal trends in GPP
and NDVI were very similar to that of ET. Most of the
decreasing zones in ET, GPP, and NDVI took place in
urbanized areas. In the valleys of the mountainous re-
gion, decreasing areas of ET, GPP, and NDVI were also
observable possibly caused the conversion from arable
land to forest or grassland.
2.5 Temporal trends in monthly ET, GPP, and
NDVI
Fig. 6 shows the temporal trends of monthly ET,
300
350
400
450ET and P
(mm) 500
550
600
650
700
2002 2003 2004 2005 2006 2007 2008 2009
Year
2010 2011 2012 2013 2014 2015 2016 2017 2018
P<0.05 P>0.05 P>0.05
Ⅰ: P Ⅰ: ETⅡ: P Ⅱ: ETBYD: P BYD: ET
y=3.5377x+485.1
R2 =0.2728
y=0.4018x+572.17
R2 =0.0032
y=2.3799x+519.77
R2 =0.1409
Fig. 3 Variation and trend in annual evapotranspiration (ET) and precipitation (P) in mountain areas (Ⅰ), plain areas (Ⅱ),
and the whole catchment (BYD)
602 中国生态农业学 报 (中英文 )?2023 第 31 卷
http://www.ecoagri.ac.cn
GPP, and NDVI and the comparisons among the moun-
tainous region, plain region, and the whole catchment.
For the mountain region (Ⅰ), there was a good agree-
ment between ET, GPP, and NDVI for nearly all the
growing seasons. The only exception was for the cold
season (October to mid-March) when ET, GPP, and
NDVI were very low and natural vegetation withered. It
suggested that the recovery or greening of natural vegeta-
tion was responsible for the increase of ET in the moun-
tainous areas.
In the plain area (Ⅱ), the trends in ET, GPP, and
NDVI were a little complicated. Aside from the strong
urbanization effect (Fig. 5), there could be other influen-
cing factors. For instance, there was a general agreement
between the decrease in ET and NDVI; especially from
October to May which was the winter wheat growing
season. In the summer or maize growing season from
June to September, ET and GPP increase but not NDVI.
2.6 ET response to vegetation and land-use
Our results showed an increase in ET followed
by a stable increase in NDVI, both at the catchment
level and for the mountain region during 2002–2018.
There was still a need to more clearly determine the
main drivers of change in ET.
Analysis of precipitation data for 2002–2018
showed that on average, precipitation for the early
5 years (464.7 mm in 2002–2006) was quite similar to
that for the late 5 years (472.2 mm in 2014–2018). This
provided a valuable time to observe the influence of
NDVI changes (NDVI2014–2018 minus NDVI2002–2006) on the
0.64
0.66
0.68
0.70
0.72NDVI
0.74
0.76
0.78
0.80
0.82
R2=0.65
P>0.05 P<0.05
R2=0.359
P>0.05
2002 2003 2004 2005 2006 2007 2008 2009 2010 201
1
2012 2013 2014 2015 2016 2017 2018
b1250
1150
1050
950
GPP
(g?cm
?2
?a?1
)
850
750
Ⅰ: y=12.50x+857.4
P<0.05
Ⅱ: y=2.70x+1019
R2=0.019
P>0.05
BYD: y=9.37x+919.06
R2=0.216
P>0.05
Year Year
2002 2003 2004 2005 2006 2007 2008 2009 2010 201
1
2012 2013 2014 2015 2016 2017 2018
Ⅰ Ⅱ BYDa
R2=0.3334
I: y=0.0042x+0.7063 : y=?1E?04 x+0.7597
R2=0.0006
BYD: y=0.0021x+0.7327
Fig. 4 Trend and significance of Gross Primary Production (GPP) (a) and Normalized Difference Vegetation Index (NDVI,
b) in mountainous areas (Ⅰ), plain areas (Ⅱ), and the whole catchment (BYD) from 2002 to 2018
a b c
N
ET trend
(mm?a?1 )
GPP trend
(g?cm?2 ?a?1 )
?0.01~?0.005
?0.005~0
0~0.005
0.005~0.01
>0.01
?5~?2 ?15~?5
?5~0
0~5
5~15
>15
?2~0
0~2
2~5
>5
0 25 50 100 km
NDVI trend (?a?1 )
Fig. 5 Temporal trends in evapotranspiration (ET, a), gross primary production (GPP, b), and normalized differ-
ence vegetation index (NDVI, c) in Baiyangdian Catchment for the period 2002–2018
第 4 期 Christine Mushimiyimana, et al: Drivers of evapotranspiration increase in the Baiyangdian Catchment 603
http://www.ecoagri.ac.cn
change of ET at pixel-level under similar precipitation
conditions. Fig. 7 shows the correspondence of NDVI
changes to the changes of ET for the mountains area
(Fig. 7a) and for the urban area (Fig. 7b) and cropland
(Fig. 7c) in the plain region between the two periods.
For the mountain area, both ET and NDVI in-
creased in most pixels in the late 5 years with average in-
crease of 50.1 mm?a?1 in ET and 0.04 in NDVI. This
could be explained by the lunch of afforestation program
in the Taihang Mountain since 1986 and returning of ag-
ricultural land to the forest at the end of the 1990s. This
policy has largely increased forest cover from 13.1% at
the start of 1980 to the current value of 26.8% in the
Taihang Mountain, where our study area is located. The
vegetation recovery was at least one of the major drivers
of the increase in ET.
Given the large difference in ET between cropland
and urban area in the plain region, the urban area was
separated from cropland for further analysis. In Fig. 7b,
ET in more urban pixels shifts to the negative value in
correspondence to the development of urbanization and
decline of NDVI. Although there were seemingly more
pixels with a positive trend of ET and NDVI in cropland,
there was still a large amount of pixels with declining ET
and NDVI as in Fig. 5. In some cropland areas, espe-
cially on the north side of the plain near Beijing and cro-
pland near the mountains, ET and NDVI decreased. This
was likely driven by the government policy on reducing
the cultivation of high water-consuming crops like wheat.
2.7 ET response to the reduction of wheat planta-
tion
In the plain area, two staple crops are heavily cultiv-
ated. While summer maize is planted in the rainy season
from June to September with less irrigation, winter wheat
is grown in the dry season from October to early June
and relies heavily on groundwater irrigation. Crop adjust-
a
b
c
1.5
1.0
0.5
0
ET
trend (mm?a
?1
)
?0.5
?1.0
?1.0
?0.006
?0.004
?0.002
0
0.002
0.004
0.006
0.008
0.010
NDVI trend (?a
?1
)
?0.5
0
0.5
1.0
1.5
2.0
2.5
3.0
GPP
trend (g?cm
?2
?a?1
)
1 2 3 4 5 6 7
Month
8 9 10 11 12
Ⅰ Ⅱ BYD
Fig. 6 Temporal trends in monthly evapotranspiration
(ET, a), gross primary production (GPP, b), and
normalized difference vegetation index (NDVI, c) in
the mountain region (Ⅰ), plain region (Ⅱ), and the
whole catchment (BYD) for the period
2002–2018
a
b
c
?150
?100
?0.30 ?0.20 ?0.10 0 0.10 0.20?50 0
50
100
150
200
250
?150
?100
?0.30 ?0.20 ?0.10 0 0.10 0.20?50 0
50
100
150
200
250
?150
?100
?0.30 ?0.20 ?0.10 0 0.10 0.20?50 0
50
100
150
200
250
NDVI change
ET change (g?cm?2 )
Fig. 7 Correspondence of evapotranspiration (ET)
changes (ET2014–2018 minus ET2002–2006) and
Normalized Difference Vegetation Index (NDVI)
changes (NDVI2014–2018 minus
NDVI2002–2006) in pixels of the mountainous
region (Region Ⅰ, a), urban (b) and cropland (c)
in the plain region (Region Ⅱ)
604 中国生态农业学 报 (中英文 )?2023 第 31 卷
http://www.ecoagri.ac.cn
ment is one of the policies the governments use to re-
duce the unsustainable use of groundwater since 2014.
As shown in Fig. 8, ET in maize and wheat sea-
son in the plain without urban pixels performs very
differently. While ET in maize growing season gener-
ally increased for most of the pixels, it was totally dif-
ferent in wheat season. The area with increasing ET
was mostly on the east side of the plain. On average,
ET for winter wheat decreased by ?1.03 mm?a?1, while
ET for summer maize increased by 2.5 mm?a?1. On aver-
age, ET in the plains increased slowly driven by the de-
crease of ET during the wheat growing season.
Maize ET
(mm?a?1 )
Wheat ET
(mm?a?1 )
?2~0
0~2
2~5
>5
?2~0
0~2
2~5
>5
a b
?5~?2 ?5~?2
Fig. 8 Spatial distributions of evapotranspiration (ET
trend) changes for summer maize season from
June to September (a) and winter wheat grow-
ing season from October to May (b) in the plain
region for the period 2002–2018
3 Discussions
3.1 Drivers of ET change
Remote sensed ET is increasingly becoming a use-
ful tool to detect the effect of natural and human influ-
ence on the water cycle and water resources (Anderson et
al., 2012). Our study confirmed several drivers influen-
cing the changes of ET in the mountainous and plain re-
gions of the Baiyangdian Catchment.
One critical driver is the negative effect of vegeta-
tion recovery in the mountains. There is a significant vis-
ible increase in ET at an annual rate of 3.54 mm?a?1
(Fig. 3) or 56.5 mm from 2002?2018. Such trend was
similar to recent studies on the effect of greening on ET
increase. For instance, Feng et al. (2016) in the Loess
Plateau found that new planting has caused both net
primary productivity (NPP) and ET to increase and res-
ult in a decrease in the runoff. Considering the fact that
the annual runoff in the Tanghe Catchment, one of major
rivers in the mountainous region, varies from 13.5?225.0
mm with an average of 63.1 mm from 1960?2002 (Hu et
al., 2009), 56.5 mm of change in ET for the study period
in the Baiyangdian Catchment was huge. Bai (2017)
studied runoff change in three sub-catchments, Fuping,
Daomaguan, and Zijingguan, at the upstream of the Bai-
yangdian Catchment and found that runoff decreased
from 839 million m3 in the 1990s to 425 million m3 in
2000?2015, which is equivalent to 34.5 mm of runoff de-
crease. The runoff decrease of 34.5 mm in 2000?2015 is
already very close to 56.5 mm of ET increase. Besides,
from Fig. 3, variation of annual precipitation is another
factor significantly influencing the annual variation of
ET, although it is still not enough to conclude to how
much extent the ET increase is driven by precipitation
and vegetation recovery separately.
In the plain region, there are three factors influen-
cing the trend of ET. Firstly, urbanization is good for the
decrease of ET. As much land is shifted to a concreted
surface, it is easy to understand that less available water
for evaporation causes the decrease of ET. Secondly,
from Fig. 8, not like the maize growing season, ET in the
winter wheat growing season decreased (Fig. 8). It sug-
gested that the government’s effort to reduce high
groundwater use in winter wheat since 2014 was making
effect. The above two factors are beneficial for con-
trolling groundwater depletion in the plain. However,
even though the two factors were causing the decrease in
ET, cropland irrigation still caused ET increase in the
whole plain. In other words, the two beneficial factors
are not enough to offset the ET increase from irrigation.
A stabilization of ET in the plain is still not achieved. A
previous study by Gao et al. (2017) and Chi et al. (2022)
also showed that even after the implementation of ecolo-
gical water recharge from South-to-North Water Trans-
fer projects, groundwater levels in the plain continuously
declined caused by over-exploitation of groundwater.
This is another example of the “irrigation paradox”
(Grafton et al., 2018), which means that agricultural wa-
ter saving at the field level can hardly solve water short-
age in the basin or catchment level. The visible increase
in GPP in the plain area also suggests that an increase in
ET is beneficial for the increase of crop yield. For in-
stance, Ai et al. (2020) investigated the global trend of
ET, GPP, and water use efficiency (WUE) and found the
第 4 期 Christine Mushimiyimana, et al: Drivers of evapotranspiration increase in the Baiyangdian Catchment 605
http://www.ecoagri.ac.cn
increase of ET, GPP, and WUE in the North China Plain
or more water for a higher crop yield.
3.2 ET as an indicator for catchment hydrology and
the reliability
While our results suggested that different drivers
were responsible for the hydrological change in the
mountain and plain regions, ET insignificantly increased
at an annual rate of 2.4 mm?a?1 or 38.4 mm, mainly con-
tributed from the upstream mountainous region. Even
though ET increase in the plain is limited, as there is less
and less runoff available for the plain and Baiyangdian
Lake, groundwater depletion is likely to accelerate (Moi-
wo et al. 2010; Hu et al. 2012). For instance, even with
only 1 billion m3 of water from South-to-North Water
Transfer Projects between 2014 and 2018 (Sohu News,
2019), Wang (2020) simulated that around 489 million
m3 of groundwater was over-drafted per year in
2015–2017. And during 2006–2011, over 0.51 billion
m3 of water was delivered from the Yellow River to the
Baiyangdian Lake (Wang et al., 2018). And after the
lunch of the Xiong’an New Area, much water has been
delivered to Xiong’an City and Baiyangdian Lake. Con-
sidering the catchment area is around 31 200 km2, 34.5
mm of overuse of water equals around 1.07 km3 of water
shortages each year in the catchment. It cannot be con-
clusively stated that the 1.07 km3 water deficit was the
precise value of water shortage in the catchment, since
both ET value from PML-V2 (Fisher et al., 2017;
Pascolini-Campbell et al., 2021) and precipitation from
limited stations especially in the mountains are not accur-
ate enough for a precise evaluation of water shortage.
References
AI Z P, WANG Q X, YANG Y H, et al. 2020. Variation of gross
primary production, evapotranspiration and water use efficiency
for global croplands[J]. Agricultural and Forest Meteorology,
287: 107935
ALLEN R G, PEREIRA L S, HOWELL T A, et al. 2011. Evapotran-
spiration information reporting: I. Factors governing measure-
ment accuracy[J]. Agricultural Water Management, 98(6):
899?920
ANDERSON M C, ALLEN R G, MORSE A, et al. 2012. Use of Land-
sat thermal imagery in monitoring evapotranspiration and man-
aging water resources[J]. Remote Sensing of Environment, 122:
50?65
BAI Z. 2017. Spatial and temporal changes of agricultural irriga-
tion requirements in the upper reaches of Baiyangdian Lake
and its impact on runoff of the mountainous areas[D].
Beijing: University of Chinese Academy of Sciences, 1–118
CHEN C, PARK T, WANG X, et al. 2019. China and India lead in
greening of the world through land-use management[J]. Nature
Sustainability, 2(2): 122?129
CHI G Y, SU X S, LYU H, et al. 2022. Prediction and evaluation of
groundwater level changes in an over-exploited area of the Baiy-
angdian Lake Basin, China under the combined influence of cli-
mate change and ecological water recharge[J]. Environmental
Research, Pt A: 113104
FENG X, FU B, PIAO S, et al. 2016. Revegetation in China’s Loess
Plateau is approaching sustainable water resource limits[J].
Nature Climate Change, 6(11): 1019?1022
FISHER J B, MELTON F, MIDDLETON E, et al. 2017. The future of
evapotranspiration: Global requirements for ecosystem function-
ing, carbon and climate feedbacks, agricultural management, and
water resources[J]. Water Resources Research, 53(4): 2618?2626
GAN R, ZHANG Y Q, SHI H, et al. 2018. Use of satellite leaf area in-
dex estimating evapotranspiration and gross assimilation for Aus-
tralian ecosystems[J]. Ecohydrology, 11(5): e1974
GAO Y C, WANG J F, FENG Z M. 2017. Variation trend and re-
sponse relationship of temperature, precipitation and runoff in
Baiyangdian Lake Basin[J]. Chinese Journal of Eco-Agriculture,
25(4): 467?77
GRAFTON R Q, WILLIAMS J, PERRY C J, et al. 2018. The paradox
of irrigation efficiency[J]. Science, 361(6404): 748?750
HU S S, LIU C M, ZHENG H X, et al. 2012. Assessing the impacts of
climate variability and human activities on streamflow in the wa-
ter source area of Baiyangdian Lake[J]. Journal of Geographical
Sciences, 22(5): 895?905
HU S S, ZHAO F, ZHANG G Y. 2009. Change of rainfall-runoff and
its causes in the upper Tang watershed in recent 40 years[J].
South-to-North Water Transfers and Water Science & Techno-
logy, 7(5): 73?75
JIANG Z Y, YANG Z G, ZHANG S Y, et al. 2020. Revealing the spa-
tio-temporal variability of evapotranspiration and its components
based on an improved Shuttleworth-Wallace model in the Yel-
low River Basin[J]. Journal of Environmental Management, 262:
110310
JUNG M, REICHSTEIN M, CIAIS P, et al. 2010. Recent decline in the
global land evapotranspiration trend due to limited moisture sup-
ply[J]. Nature, 467(7318): 951?954
KENDALL G M. 1975. Rank Correlation Measures[M]. London:
Charles Griffin
LEUNING R, ZHANG Y Q, RAJAUD A, et al. 2008. A simple sur-
face conductance model to estimate regional evaporation using
MODIS leaf area index and the Penman-Monteith equation[J].
Water Resources Research, 44(10): W10419
606 中国生态农业学 报 (中英文 )?2023 第 31 卷
http://www.ecoagri.ac.cn
LI C C, ZHANG Y Q, SHEN Y J, et al. 2020. LUCC-driven changes in
gross primary production and actual evapotranspiration in North-
ern China[J]. Journal of Geophysical Research: Atmospheres,
125(6). DOI:10.1029/2019JD031705
LI G, ZHANG F M, JING Y S, et al. 2017. Response of evapotranspir-
ation to changes in land use and land cover and climate in China
during 2001?2013[J]. The Science of the Total Environment,
596/597: 256?265
LI H D, WANG A Z, YUAN F H, et al. 2016. Evapotranspiration dy-
namics over a temperate meadow ecosystem in eastern Inner
Mongolia, China[J]. Environmental Earth Sciences, 75(11): 1?11
MANN H B. 1945. Nonparametric tests against trend[J]. Econometrica,
13(3): 245
MOIWO J P, YANG Y H, LI H L, et al. 2010. Impact of water re-
source exploitation on the hydrology and water storage in Baiy-
angdian Lake[J]. Hydrological Processes, 24(21): 3026?3039
MUELLER B, SENEVIRATNE S I, JIMENEZ C, et al. 2011. Evalu-
ation of global observations-based evapotranspiration datasets
and IPCC AR4 simulations[J]. Geophysical Research Letters,
(38): L06402
OKI T, KANAE S. 2006. Global hydrological cycles and world water
resources[J]. Science, 313(5790): 1068?1072
PASCOLINI-CAMPBELL M, REAGER J T, CHANDANPURKAR H
A, et al. 2021. A 10 per cent increase in global land evapotran-
spiration from 2003 to 2019[J]. Nature, 593(7860): 543?547
RODELL M, BEAUDOING H K, L’ECUYER T S, et al. 2015. The
observed state of the water cycle in the early twenty-first
century[J]. Journal of Climate, 28(21): 8289?8318
SHERWOOD S, FU Q. 2014. A drier future?[J]. Science, 343(6172):
737?739
SOHU NEWS. 2019. The 5th anniversary of the water opening of the
middle route of the South-to-North Water Diversion
Project[N/OL]. Baoding Evening News, (2019-12-12). https://
www.sohu.com/a/359943235_439193
SUN C, REN L. 2013. Assessment of surface water resources and
evapotranspiration in the Haihe River Basin of China using
SWAT model[J]. Hydrological Processes, 27(8): 1200?1222
SUN J M, YU X X, WANG H N, et al. 2018. Effects of forest structure
on hydrological processes in China[J]. Journal of Hydrology,
561: 187?199
TRENBERTH K E, SMITH L, QIAN T T, et al. 2007. Estimates of the
global water budget and its annual cycle using observational and
model data[J]. Journal of Hydrometeorology, 8(4): 758?769
V?R?SMARTY C J, MCINTYRE P B, GESSNER M O, et al. 2010.
Global threats to human water security and river biodiversity[J].
Nature, 467(7315): 555?561
WANG K, LI H, L, WU A, et al. 2018. An analysis of the evolution of
Baiyangdian wetlands in Hebei Province with artificial
recharge[J]. Acta Geoscientica Sinica, 39(5): 549?558
WANG K. 2020. Groundwater resources and wetland sustainability in
Xiong’an City[D]. Beijing: Chinese Geology University
(Beijing)
WEGEHENKEL M, JOCHHEIM H, KERSEBAUM K C. 2005. The
application of simple methods using remote sensing data for the
regional validation of a semidistributed hydrological catchment
model[J]. Physics and Chemistry of the Earth, Parts A/B/C,
30(8/9/10): 575?587
YANG Y M, YANG Y H, LIU D L, et al. 2014. Regional water bal-
ance based on remotely sensed evapotranspiration and irrigation:
An assessment of the Haihe Plain, China[J]. Remote Sensing,
6(3): 2514?2533
YANG Y T, LONG D, SHANG S H. 2013. Remote estimation of ter-
restrial evapotranspiration without using meteorological data[J].
Geophysical Research Letters, 40(12): 3026?3030
YANG Y, RODERICK M. L, YANG D, et al 2021. Streamflow sta-
tionarity in a changing world[J]. Environmental Research Letters,
16(6)
ZENG Z Z, PENG L Q, PIAO S L. 2018. Response of terrestrial evapo-
transpiration to earth’s greening[J]. Current Opinion in Environ-
mental Sustainability, 33: 9?25
ZHANG Y Q, KONG D D, GAN R, et al. 2019. Coupled estima-
tion of 500?m and 8-day resolution global evapotranspira-
tion and gross primary production in 2002–2017[J]. Remote
Sensing of Environment, 222: 165?182
ZHANG Y Q, LEUNING R, HUTLEY L B, et al. 2010. Using long-
term water balances to parameterize surface conductances and
calculate evaporation at 0.05° spatial resolution[J]. Water Re-
sources Research, 46(6): W05512
ZHOU G, WEI X, CHEN X, et al. 2015. Global pattern for the effect of
climate and land cover on water yield[J]. Nature Communica-
tions, 6: 5918
第 4 期 Christine Mushimiyimana, et al: Drivers of evapotranspiration increase in the Baiyangdian Catchment 607
http://www.ecoagri.ac.cn
|
|
