Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
RESEARCH ARTICLE
Time-series remote sensing of rice paddy expansion in the
Yellow River Delta: Towards sustainable ecological
conservation in the context of water scarcity
Chong Huang
1,2,†
& Chenchen Zhang
1,3,†
1
State Key Lab of Resources and Environmental Information System, Institute of Geographic Sciences and Natural Resources Research, Chinese
Academy of Sciences, Beijing 100101, China
2
CAS Engineering Laboratory for Yellow River Delta Modern Agriculture, Institute of Geographic Sciences and Natural Resources Research,
Chinese Academy of Sciences, Beijing 100101, China
3
College of Resources and Environment, University of Chinese Academy of Sciences, 100049, Beijing, China
Keywords
Conservation, phenological metrics, rice
paddy, rice-wetland coexistent area, Sentinel-
2, time series, Yellow River Delta
Correspondence
Chong Huang, State Key Lab of Resources
and Environmental Information System,
Institute of Geographic Sciences and Natural
Resources Research, Chinese Academy of
Sciences, Beijing 100101, China.
E-mail: huangch@leris.ac.cn
Funding Information
CAS Earth Big Data Science Project
(XDA19060302).
Editor: Nathalie Pettorelli
Associate Editor: Henrike Schulte to Buhne
Received: 14 August 2022; Revised: 2
November 2022; Accepted: 8 November
2022
†
Chong Huang and Chenchen Zhang
contributed equally to this work.
doi: 10.1002/rse2.320
Abstract
Large river deltas are usually ecologically important wetland habitats, but also
fertile agricultural exploitation areas, creating a conflict of long-running sub-
stantial interest between agricultural expansion and wetland conservation. Over
the past several years, large-scale cultivation of water-consuming rice has been
growing rapidly in the semi-arid climate of the Yellow River Delta (YRD).
Timely monitoring of rice cultivation dynamics is of great significance for sus-
tainable ecological conservation of the delta, which has insufficient freshwater
resources. This study proposed a stratified metrics-based method that integrates
statistical spectral indices and phenological metrics at different growing stages
to improve the accuracy of rice paddy classification in areas where rice and
wetlands coexist. We applied the method to time-series Sentinel-1/2 images to
produce annual rice paddy maps of the YRD from 2016 to 2021. Together with
rice paddy data from 2011 to 2015 from Statistical Yearbooks of Dongying
Bureau of Statistics, we investigated the expansion dynamics over the past dec-
ade and in this paper discuss the advantages and disadvantages of rice cultiva-
tion expansion over wetland ecosystem conservation. Rapid expansion of rice
cultivation intensifies water conflicts, and adversely affects wetland restoration
in the YRD. Considering the important ecological services of rice paddies as
alternative habitats, we argue for maintaining a reasonable scale of rice paddies
and optimizing their distribution as a potential solution to achieving the overall
sustainable conservation of the YRD in the context of water scarcity.
Introduction
A river delta is the final portion of the fluvial system, par-
ticularly rich in wetlands, such as lagoons, marshes, and
tidal flats, which are incredibly diverse and ecologically
important habitats for wintering and migratory birds and
other animals (Aiello-Lammens et al., 2011; Murray
et al., 2019). In addition, river deltas often have a rich
accumulation of silt, so they are usually fertile agricultural
areas (Clauss et al., 2018). Many large river deltas in trop-
ical and subtropical areas (e.g., Mekong Delta, Yangtze
River Delta, and Nile Delta) are important rice-producing
areas due to being situated in areas with low terrain,
plentiful precipitation, and fertile soil (Clauss et al., 2018;
Farig et al., 2022; Shi & Huang, 2015). However, many
studies have shown that intensive irrigation agriculture
has a significant impact on wetland conservation within
river deltas. Agricultural exploitation directly encroaches
ª2022 The Authors. Remote Sensing in Ecology and Conservation published by John Wiley & Sons Ltd on behalf of Zoological Society of London.
This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and
distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.
1
on original natural wetlands and probably influences the
estuary wetlands through increased demands on freshwa-
ter (Xie et al., 2011). Thus, water conflicts between agri-
cultural expansion and wetland conservation have
received substantial interest, especially for river deltas in
arid and semi-arid regions with insufficient freshwater
resources (Brisco et al., 2013; Elliott et al., 2014; Mao
et al., 2018; Xu et al., 2019).
The Yellow River Delta (YRD) is very deficient in water
resources. In recent years, the quantity of water has sharply
decreased in the lower Yellow River due to climate change
and human activity (Wang et al., 2017). This has led to a
deficiency in suitable water needed by the ecological system
of the estuarine delta (Cui et al., 2009). As a result, fresh-
water wetlands in the delta are facing a high risk of decline
and degradation. In order to protect the rare estuarine wet-
land ecosystem, diverting water from the Yellow River to
recharge the degraded wetlands has been an efficient mea-
sure to restore and maintain the wetland ecosystem health
(Wang, Qi, et al., 2012). At the same time, local govern-
ments in the YRD are facing urgent demands for economic
development, among which agriculture plays an important
role as high-intensity agricultural development has been
seen across the delta. Over the past several years, large-scale
cultivation of water-consuming rice has been growing
rapidly. Timely monitoring of rice cultivation dynamics
and analyzing the impact on ecological conservation are of
great significance for sustainable development of the YRD
in the context of a water shortage.
Numerous studies have proven the potential of satellite-
based remote sensing for rice paddy monitoring (Boschetti
et al., 2017; Guan et al., 2016; Li, Fu, et al., 2020; Singha
et al., 2019). Supervised or unsupervised algorithms are
traditionally employed to map rice paddies based on the
spectral discrimination of different bands from a single-
date image (Li et al., 2014; Roberts et al., 2002). However,
single-date images only represent instantaneous spectral
characteristics of the land surface at a single point in time.
Moreover, land cover categories may show a similar spec-
tral reflectance due to the limitations imposed by broad
spectral bands, which can produce inadequate classifica-
tion results (G´omez et al., 2016). Recent studies have
emphasized the capability of time-series remote sensing
images to identify rice paddies according to their distinct
phenological features, compared to other crops (Dong
et al., 2015; Jeong et al., 2012; Sakamoto et al., 2009; Son
et al., 2013; Xiao, Boles, et al., 2005). The time-series
approaches can be grouped into three categories according
to the way temporal information is used. The first
approach employs stacking time-series images for rice
paddy classification (Singha et al., 2019). This approach
assumes that the spectral values of the pixels in the time-
series images are temporally independent of each other
and the time sequence of image presentation has no effect
on the results, ignoring any temporal dependencies that
may be found in the data (Belgiu & Csillik, 2018). The
second approach is based on a time-series similarity mea-
surement using Euclidean distance or dynamic time warp-
ing (Petitjean et al., 2012). This approach measures the
degree of (dis)similarity between the target pixels and the
reference categories by profile matching, thus identifying
rice paddies from other land cover types (Guan
et al., 2016; Gumma et al., 2014). The main difficulty in
this method is in reconstructing time-series profiles when
large time-series gaps exist due to missed or contaminated
images. The third approach is metrics-based, using statisti-
cal spectral indices to describe the key characteristics,
especially in the flooding and transplanting phases. Since
rice, unlike most dry land crops, is first seeded and trans-
planted in flooded fields, statistical analysis on different
spectral indices [e.g., Normalized Difference Vegetation
Index (NDVI), Enhanced Vegetation Index (EVI), Land
Surface Water Index (LSWI)] during this phase can effec-
tively detect the unique features of rice paddy cultivation
(Dong & Xiao, 2016; Xiao, Boles, et al., 2005; Zhang
et al., 2020). However, a metrics-based approach remains
challenging when applied to rice monitoring in rice-
wetland coexistent areas like the YRD, as both rice paddies
and vegetated wetlands have a flooding stage, then similar
statistical spectral indices in this phase often lead to large
amounts of misclassification (Huang et al., 2020;Ni
et al., 2021; Zhou et al., 2016). Considering their obvious
difference in life spans, phenological metrics in the grow-
ing season (Walker et al., 2014; Zhao et al., 2020) might
have the potential to improve rice paddy mapping in rice-
wetland coexistent areas.
In this study, we used all the available time-series
Sentinel-2 and Sentinel-1 images of the YRD from 2016
to 2021 to: (1) develop a robust method combining sta-
tistical spectral indices and phenological metrics at differ-
ent phases for stratified rice paddy mapping in rice-
wetland coexistent areas; (2) monitor the rice cultivation
changes in the delta over the past decade; and (3) analyze
the impact of rice cultivation expansion on ecological
conservation in the YRD in the context of water scarcity.
Materials and Methods
Study area
The Yellow River Delta (YRD) is located at the estuary of
the Yellow River in Dongying City, Shandong Province,
China (see Fig. 1). The region belongs to the warm tem-
perate zone and has a semi-humid and semi-arid conti-
nental monsoon climate. The average annual temperature
is 12.3°C. The mean annual evaporation is greater than
2ª2022 The Authors. Remote Sensing in Ecology and Conservation published by John Wiley & Sons Ltd on behalf of Zoological Society of London.
Rice Paddy Expansion Impact on YRD Conservation C. Huang & C. Zhang
the mean annual precipitation, which are approximately
1885.0 and 537.4 mm, respectively.
The YRD is an aggradational delta with a weak tide,
much sediment transport, and frequent displacements.
Having a history of only 150 years, the delta forms the
most complete and extensive young wetland ecological
system in China. Pioneer wetlands and habitats of rare
birds in the delta have an important status in biodiversity
conservation in China and the world (Zhang et al., 2016).
It is a wintering, stop-over, and breeding site for migra-
tory birds of inland Northeast Asia and around the Paci-
fic Ocean. There are 283 species of important, protected
Chinese birds, among of them 9 species birds are nation
grade I protection birds (Wang, Lian, et al., 2012). There-
fore, the Chinese government defined a national wetland
nature protection area of 1530 km
2
in the YRD in 1992.
At the same time, the delta has a large area of saline
land resources, which provide preferable conditions for
saline agricultural development. In order to meet increas-
ing food requirements, agriculture has been intensively
developed in the delta. According to the Statistical Year-
book of Dongying Bureau of Statistics (http://dystjj.
dongying.gov.cn/col/col36583/index.html), typical agricul-
tural cultivations include double cropping rotation of
winter wheat and summer corn/soybean as well as single-
season dryland crops such as spring maize and cotton.
Rice paddies can also be cultivated over the delta if the
irrigation water can be guaranteed. Considering the
scarcity of freshwater resources, coordinating the conflict
between agricultural development and ecological conser-
vation has long been difficult in the YRD.
Data
Sentinel-1 remote sensing data
The Sentinel-1A/B Level 1 Ground Range Detected
(GRD) product in the Interferometric Wide swath mode
from January 1, 2016, to December 31, 2021, was col-
lected via the Google Earth Engine (GEE) platform.
Sentinel-1A/B together have a 6-day repeat cycle at the
equator. The satellites provide dual-polarized VV (Vertical
transmit Vertical receive) and VH (Vertical transmit Hor-
izontal receive) data (Torres et al., 2012). Sentinel-1 data
in GEE were pre-processed with the Sentinel-1 Toolbox
using the orbit metadata update, GRD border noise
removal, thermal noise removal, radiometric calibration,
and terrain correction. The final terrain-corrected values
were converted to decibels (dB) in each pixel via log scal-
ing 10log (DN).
Sentinel-2 remote sensing data
Sentinel-2A and Sentinel-2B Multi Spectral Instrument
images from 2016 to 2021 were used in this study.
Sentinel-2A/B images together have a 5-day temporal res-
olution at the equator. Clouds and cloud shadows were
removed using the QA60 bitmask band with cloud mask
information. Snow and ice were identified and removed
using Normalized Difference Snow Index (NDSI), which
can be described as near-infrared (NIR) >0.11 and
NDSI >0.4 (Hall & Riggs, 2011).
Normalized Difference Vegetation Index (Tucker, 1979)
and LSWI (Xiao et al., 2004; Xiao, Zhang, et al., 2005)
were calculated for each image using Equations (1) and
(2). NDVI is the most common index used to detect veg-
etation vigor and track its changes (Huete et al., 1997,
2002; Tucker, 1979). LSWI is a good indicator that can
capture the signal of moisture in vegetation and soil
(Xiao et al., 2004; Xiao, Zhang, et al., 2005).
NDVI ¼NIRRed
NIR þRed (1)
LSWI ¼NIRSWIR
NIR þSWIR (2)
where Red, NIR, and SWIR are the surface reflectance
values of red, near-infrared (NIR), and shortwave-
infrared (SWIR) bands for Sentinel-2.
Normalized Difference Vegetation Index time-series
data were further aggregated into 10-day composite data
Figure 1. Location of the study area.
ª2022 The Authors. Remote Sensing in Ecology and Conservation published by John Wiley & Sons Ltd on behalf of Zoological Society of London. 3
C. Huang & C. Zhang Rice Paddy Expansion Impact on YRD Conservation
by calculating mean values of the valid Sentinel-2 obser-
vations (Wang et al., 2020). Data gaps were filled by lin-
ear interpolation to achieve full coverage throughout the
temporal domain (Griffiths et al., 2019). 10-Day NDVI
time-series data were smoothed using the Savitzky–Golay
filter with a moving window size of 7 and the second-
order polynomial (Liu et al., 2020). The 10-day NDVI
time-series data were used to extract phenological indica-
tors reflecting the characteristics of the rice growth cycle,
which is described in detail in Section “Rice paddy dis-
crimination based on phenological metrics in the growing
season”.
Ground reference data
Field surveys were conducted from 2016 to 2021 to col-
lect geo-referenced field sample points for the major land
cover types in Dongying. After the field survey, all the
ground samples were visually checked using very high
spatial resolution images in Google Earth and three
Sentinel-2 false-color composites (R: NIR, G: Red, B:
Green), including false-color composites from early Jan-
uary to late March, late May to early June, and early
September to late October. The integration of field
points, very high spatial resolution satellite images in
Google Earth, and Sentinel-2 false-color composites
served as the background reference to create regions of
interest (ROI). Finally, ROIs were collected for each year
from 2016 to 2021 for each land cover type: 70 ROIs for
rice paddy samples, 48 ROIs for reed wetland, 120 ROIs
for cotton, 72 ROIs for spring maize, 44 ROIs for dou-
ble season crops, and 42 ROIs for forest. Half of the
samples were randomly selected for training and classifi-
cation, and the remaining half was used to validate the
resultant annual rice paddy map for the corresponding
year.
Methods
Figure 2shows the stratified workflow for producing
annual rice paddy maps from 2016 to 2021 using time-
series Sentinel-1/2 images. First, we used annual statistical
indices (annual NDVImax) to identify vegetated areas
and non-vegetated areas, and seasonal statistical indices
(NDVImax between Jan. and Apr.) to eliminate double-
season vegetation and obtain the single-season vegetation
layer. Then, statistical indices of the rice transplanting
period were applied to identify potential rice paddies (rice
paddies and some vegetated wetlands) within the single-
season vegetation layer. At the last step, rice paddies were
further delineated within the potential rice layer using
phenological metrics of the growing season derived from
time-series Sentinel-2 NDVI data.
Annual maps of single-season vegetation
Figure 3shows the time-series of NDVI, LSWI, and VH
for typical land cover types in the study area. Effective
discrimination of rice paddies starts with the accurate
single-season vegetation map, as the paddies are cultivated
in a single cropping system in the YRD (Wei et al., 2022)
(see Fig. 3A). First, vegetation pixels were identified using
the annual NDVI statistical feature. Compared to non-
vegetation areas (water, built-up, and tidal) which have
very low greenness values throughout the year (see
Fig. 3h–j), vegetated areas have obvious annual greenness
dynamics (see Fig. 3A–G). We extracted vegetated areas
by identifying pixels with annual NDVI
max
>0.5 (Qin
et al., 2015). Second, single-season vegetation (rice paddy,
reed wetland, cotton, spring maize, and forest) were
extracted from the vegetation layer using the seasonal
NDVI statistical feature. In the YRD, the winter wheat is
the first crop of the double cropping system, which is
Figure 2. Flowchart for stratified rice paddy mapping using time-series Sentinel-2 images.
4ª2022 The Authors. Remote Sensing in Ecology and Conservation published by John Wiley & Sons Ltd on behalf of Zoological Society of London.
Rice Paddy Expansion Impact on YRD Conservation C. Huang & C. Zhang
planted in October and harvested in May–June of the
next year. So, compared to single-season vegetation (see
Fig. 3A–D and G), pixels of double-season vegetation
have high NDVI values in winter and early spring as the
wheat seedlings turn green (see Fig. 3E and F). Therefore,
the maximum values of NDVI between January and April
Figure 3. Time-series of Normalized Difference Vegetation Index, Land Surface Water Index, and VH for (A) rice paddy, (B) wetland, (C) cotton,
(D) spring maize, (E) winter wheat–maize, (F) winter wheat–soybean, (G) forest, (H) water, (I) built-up, and (J) tidal.
ª2022 The Authors. Remote Sensing in Ecology and Conservation published by John Wiley & Sons Ltd on behalf of Zoological Society of London. 5
C. Huang & C. Zhang Rice Paddy Expansion Impact on YRD Conservation
[NDVI
(Jan.–Apr.)
] (see Fig. 4A) were calculated for map-
ping single and double-season vegetation. The lowest
value of NDVI
(Jan.–Apr.)
for double-season vegetation is
0.6, and the highest value of NDVI
(Jan.–Apr.)
for single-
season vegetation is 0.4. We set the mean value of the
two values as the threshold [NDVI
(Jan.–Apr.)
≤0.5] to gen-
erated a single-season vegetation map, which was used as
a baseline map for the phenology-based mapping of rice
paddies in the next step.
Potential rice paddy identification using statistical
indices in transplanting phase
According to prior knowledge of rice phenology (Le Toan
et al., 1997) and time-series analysis (see Fig. 3A), the
temporal dynamics of rice paddy fields can be character-
ized by three phases: (1) transplanting period, (2) grow-
ing and maturity period, and (3) fallow period after
harvest. Both rice paddies and parts of natural wetlands
have a mixture of water and green plants during the
transplanting phase, which leads to a higher water content
than other single-season vegetation. LSWI was selected to
distinguish rice paddies due to its excellent performance
in identifying the flood signals in the transplanting period
(Xiao et al., 2006; Xiao, Boles, et al., 2005). We distin-
guished potential rice paddies from single-season vegeta-
tion using the maximum values of LSWI (LSWI_TP) in
the transplanting period (TP, DOY 140 to 170) since rice
paddies feature higher LSWI_TP values than other crops.
Considering that forests also have relative high LSWI_TP
values (see Fig. 3G), we first identified forest category
using information from the VH signal (Borlaf-Mena
et al., 2021). A map of the annual frequencies of VH val-
ues >−20 (see Fig. 3G) in a year was generated and pix-
els with values greater than 0.6 were identified as forest
(see Fig. 4B). In this study, the 2.5% percentile at the
95% confidence level (Qin et al., 2015; Yang et al., 2021)
was selected as the threshold (i.e., LSWI_TP >0.26) to
create a potential rice paddy map (see Fig. 5A), which still
contained a certain percentage of reed wetlands.
Rice paddy discrimination based on phenological
metrics in the growing season
In the early growing season, wild wetland plants are more
tolerant of low temperatures than rice, so the start of the
growing season is a little earlier than that of rice. Simi-
larly, the end of the growing season of wetland plants is a
little later than rice (da Cruz et al., 2013; Zhou
et al., 2016). As a result, the growing season of wetland
plants is longer than that of rice paddies. Another obvi-
ous growing feature difference is that rice after trans-
planted from seedlings will begin greening up faster than
wetlands at the beginning of the growing season due to
the effect of fertilization on the growth of juvenile planta-
tion (Zhou et al., 2016). Based on the phenology differ-
ence in the growing season between rice paddies and
vegetated wetlands, we employed five phenological met-
rics as the main variables to discriminate the rice paddies
from vegetated wetlands in rice-wetland coexistent areas,
including start of the season (SOS), end of the season
(EOS), growing season length (GSL), start date of peak
season (SDPS), and green-up speed (GUS). SOS and
SDPS are respectively defined as the times of NDVI
reaching 30% and 90% of the NDVI amplitude from the
left minimum. EOS is the time of NDVI having 30%
amplitude higher than the right minimum. The GSL rep-
resents the days between SOS and EOS. The GUS is
defined as the ratio of the NDVI change over the number
of days between SOS and SDPS (Wang et al., 2020). Indi-
cators reflecting the characteristics of the rice growth
cycle (SOS, SDPS, EOS, GSL, and GUS) were extracted
Figure 4. Distribution of (A) Normalized Difference Vegetation Index during January to April for single-season and double-season vegetation and
(B) the frequency of VH >−20 in a year.
6ª2022 The Authors. Remote Sensing in Ecology and Conservation published by John Wiley & Sons Ltd on behalf of Zoological Society of London.
Rice Paddy Expansion Impact on YRD Conservation C. Huang & C. Zhang
using a 10-day NDVI composite time series. We used the
training data and the box line plots to explore and visual-
ize the separability between rice paddies and vegetated
wetlands (see Fig. 5B–F). As the five phenological metrics
derived from the NDVI time series were good indicators
to separate rice paddies from wetlands, all of them were
selected as classification features to achieve complemen-
tary strengths and obtain the maximum separability.
A Random Forest (RF) classifier embedded in the
Google Earth Engine (Gorelick et al., 2017) was used to
identify rice paddies from potential rice paddy pixels for
each year from 2016 to 2021 using the phenological
metrics derived from growing season. RF classifiers have
high efficiency and high accuracy in processing high-
dimensional, massive data compared with other machine
learning algorithms (Belgiu & Dr˘agut
ß,2016). The
number of decision trees was set to 50. The other
parameters were set by default in order to avoid overfit-
ting, as recommended by Liaw and Wiener (Liaw &
Wiener, 2002).
Validation and comparison with other available
rice paddy datasets
The validation ROIs described in Section “Ground refer-
ence data” was used to quantitatively assess the accuracies
of the resultant rice paddy maps from 2016 through
2021. We grouped the validation ROIs into rice paddy
and non-rice paddy, and calculated assessment indicators
using confusion matrices (Foody, 2002), including overall
accuracy (OA), kappa coefficient, producer’s accuracy
(PA), and user’s accuracy (UA).
Dongying Bureau of Statistics publishes annual reports
on the planting areas of the main crops in each county
(http://dystjj.dongying.gov.cn). The rice planting areas in
2016–2020 from the Statistical Yearbook were used for
inter-comparison with our results at the county level. In
addition, we also compared our results with China’s third
national land survey report which represented the rice
paddy planting area in 2019.
Results
Accuracy assessment of the rice paddy maps
from 2016 to 2021
Accuracy assessments for annual maps of rice paddies
from 2016 through 2021 were carried out based on the
validation polygons (Table 1). The OAs and the Kappa
coefficients were greater than 96% and 0.9 from 2016 to
2021, respectively. Rice paddy had PAs and UAs greater
than 90% in each year. High values of assessment metrics
indicated that the resultant rice paddy maps from 2016 to
2021 had high accuracies.
Figure 5. Signature analysis of (A) maximum values of Land Surface Water Index for rice paddies and other single-season vegetation and (B–F)
temporal statistical indicators for rice paddies and wetlands.
ª2022 The Authors. Remote Sensing in Ecology and Conservation published by John Wiley & Sons Ltd on behalf of Zoological Society of London. 7
C. Huang & C. Zhang Rice Paddy Expansion Impact on YRD Conservation
Annual dynamics of rice paddies at the
municipal and county scales
Figure 6shows the spatial distribution of rice paddies in
Dongying City from 2016 to 2021. A total of
26 253.64 ha of rice paddies was found in Dongying City
in 2021. Kenli County had the largest rice paddy area of
17 431.98 ha, accounting for 66% of the total rice paddies
in Dongying City. The rice paddy area in Lijin County
and Hekou County was relatively large, covering about
5234.36 ha and 2681.13 ha, respectively, accounting for
about 20% and 10% of the total area, respectively. About
860.11 ha of rice paddies were found in Dongying
County, accounting for about 3% of the total area. Guan-
grao County had the least rice paddy area at 46.05 ha,
accounting for less than 1% of the total area.
TABLE 1. Summary of the accuracy assessment for annual maps of
rice paddies from 2016 to 2021.
Year Class PA (%) UA OA Kappa
2016 Rice paddy 92.31 92.44 96.93% 0.90
Non-rice paddy 98.09 98.06
2017 Rice paddy 94.22 94.83 97.23 0.93
Non-rice paddy 98.25 98.03
2018 Rice paddy 92.29 92.09 96.67 0.90
Non-rice paddy 97.86 97.91
2019 Rice paddy 93.14 94.37 97.32 0.92
Non-rice paddy 98.47 98.12
2020 Rice paddy 94.81 95.12 97.28 0.93
Non-rice paddy 98.20 98.08
2021 Rice paddy 95.32 95.14 97.20 0.93
Non-rice paddy 97.98 98.05
OA, overall accuracy; PA, producer’s accuracy; UA, user’s accuracy.
Figure 6. Spatial distribution of rice paddies in Dongying City from (A–F) 2016 to 2021.
8ª2022 The Authors. Remote Sensing in Ecology and Conservation published by John Wiley & Sons Ltd on behalf of Zoological Society of London.
Rice Paddy Expansion Impact on YRD Conservation C. Huang & C. Zhang
Rice paddies in Dongying City are mainly distributed
along both sides of the Yellow River and near reservoirs
with good irrigation conditions. Before 2018, rice paddies
are mainly distributed near the estuary of the Yellow River.
Then, rice paddies were gradually expanded to the upper
sections of the Yellow River (see Fig. 6). Between 2016 and
2021, rice paddies expanded across the delta; additionally,
in areas where they occurred, rice paddies became more
concentrated and tended to cover a larger area. There was
also an obvious increase in rice paddy area in the southern
part of the Yellow River compared to the northern part.
Figure 7shows the rice paddy area variation during the
period 2011–2021. As Sentinel-2 data were available in
the study area from 2016, rice paddy area before 2015
was obtained from the Dongying Statistical Yearbook to
analyze the rice paddy area changes in the past decade.
The rice paddy cultivation dynamics in Dongying City
can be divided into three stages in the last 10 years: (1)
slight increase from 2011 to 2015, (2) rapid increase from
2015 to 2018, and (3) slight decrease but maintaining a
high level from 2018 to 2021 (see Fig. 7). Before 2015,
the rice cultivation area in the YRD was seen a slight
increasing trend. The rice areas in 2011 and 2015 were
3.86 ×10
3
ha and 6.78 ×10
3
ha, respectively, with an
average rate of increase of 0.73 ×10
3
ha/a. In 2016, rapid
expansion can be found, as rice cultivation increased by
5.40 ×10
3
ha, compared with the previous year. The peak
was reached in 2018, with a rice paddy area of
28.18 ×10
3
ha, which is nearly eight times as much as
2011. In the following 3 years, the rice planting still main-
tained a high level, with an area of no less than
26 ×10
3
ha each year.
At the county level, different counties showed different
trends over the period 2016–2021. Kenli County showed
a similar trend with that of Dongying City, since Kenli
County contributed the most to the overall expansion.
During 2016–2021, rice cultivation in Kenli County
accounted for more than 65% of the total rice area of
Dongying City. After 2018, the rice cultivation decreased
to some extent. Hekou County showed a similar rice cul-
tivation expansion trend to Kenli County, but with less
variation in magnitude. For Lijin County, however, rice
cultivation has continued to increase since 2011, and
reached the peak area of 5.23 ×10
3
ha in 2021.
Inter-comparison between rice paddies in
this study and official datasets
We summed the rice paddy areas by county from the rice
paddy maps from 2016 to 2020 and compared them with
the rice paddy planting area from the Statistical Year-
books at the county level (see Fig. 8A). We also compared
the rice paddy area for 2019 between our study and the
third national land survey of China (see Fig. 8B). The
areas of rice paddies were highly correlated with those
from the Statistical Yearbooks and the third national land
survey of China, with correlation coefficients over 0.95.
The high relationship indicated our mapping results were
close to the Statistical Yearbooks of Dongying Bureau of
Statistics and the third national land survey of China.
Discussion
Improvements in the rice paddy mapping
algorithm
In this study, we proposed an enhanced metrics-based
method which combines statistical and phenological
Figure 7. Variations in rice paddy area in counties of Dongying City from 2011 to 2021. The rice paddy area from 2016 to 2021 was estimated
from this study combining Sentinel-1/2 data. The rice paddy area from 2011 to 2015 was from the Dongying Statistical Yearbook due to unavail-
ability of Sentinel-2 data.
ª2022 The Authors. Remote Sensing in Ecology and Conservation published by John Wiley & Sons Ltd on behalf of Zoological Society of London. 9
C. Huang & C. Zhang Rice Paddy Expansion Impact on YRD Conservation
metrics to improve the accuracy of rice paddy mapping
in rice-wetland coexistent areas. Compared with the pre-
vious studies, we improved the phenology-based algo-
rithm in two aspects: (1) extending the time-series
analysis from the transplanting phase to the entire grow-
ing season, and (2) employing five specific phenological
metrics to provide a more comprehensive insight into the
differences in growth characteristics of rice paddies and
wetlands. The previous studies have typically used statisti-
cal relationships among different VIs (e.g., NDVI, LSWI,
and EVI) of the transplanting phase to identify rice pad-
dies (Dong et al., 2016; Xiao et al., 2006; Xiao, Boles,
et al., 2005; Zhang et al., 2015). However, such relation-
ships only represent the flooding characteristics in the
transplanting period. Although they are useful to discrim-
inate rice paddies from most land covers, it is difficult to
apply them to the rice-wetland coexistent area since both
the rice paddies and wetlands have similar flood signals
in the same phase due to water supplement (Zhou
et al., 2016). Therefore, when statistical metrics of the
transplanting phase were used for rice mapping, some
studies have required an additional thematic wetland map
to mask the wetland area (Jin et al., 2016). In this study,
phenology-based analysis was extended from the trans-
planting phase to the full growing season, thus, more
phenological features can be extracted to further discrimi-
nate between the rice paddies and the wetlands. Pheno-
logical features of the growing season have been
recognized and extensively applied to crop classification
such as sugarcane, cotton, soybean, and corn (Al-
Shammari et al., 2020; Lhermitte et al., 2011; Sakamoto
et al., 2005; Wang et al., 2020; Zhong et al., 2016). In this
study, we demonstrated that the phenological metrics in
the growing season, together with statistical indices in the
transplanting phase are capable of enhancing the
separability between rice paddies and wetlands in the
rice-wetland coexistent area. The resultant high-accuracy
(PA/UA/OA >90%) rice paddy maps demonstrate that
the temporal behaviors in different growing phases can be
well utilized for time-varying category discrimination,
thus facilitating the improvement of rice paddy mapping
in rice-wetland coexistent areas.
The reasonably high accuracy of the classification result
showed the effectiveness of our method to map paddy
rice in the YRD, however, it may be easily applied to
other temperate regions where the paddies are cultivated
as single cropping system such as Northeastern China,
Japan, and the Korea Peninsula (Carrasco et al., 2022;Jo
et al., 2020; Zhang et al., 2015). Moreover, since we used
a set of robust strategies in the stratified metrics-based
algorithms, it may have the potential to be applied to
some subtropical areas with winter wheat (rapeseed) –
summer rice double cropping system (Wang et al., 2015).
In the case, we used the algorithms to generate a double-
season vegetation map as the baseline map for the
phenology-based mapping of rice paddies. It should be
noted that although our method is simple and robust for
rice paddy mapping in temperate or subtropical regions
where the crop phenology, planting, and management
activities for rice cultivation are relatively consistent per
calendar year, this method may not be directly transfer-
able to the tropical regions like South and Southeast Asia
where the cropping systems are complex (Jain
et al., 2013; Minasny et al., 2019) and rice can be culti-
vated at any time in the wet season, or even in the dry
season with the support of irrigation systems (Huang
et al., 2020). For these regions where paddy rice can be
cultivated more than once (double- or triple cropping),
the shape-based similarity measures are more encouraging
(Guan et al., 2016; Minasny et al., 2019).
Figure 8. A comparison of the estimated areas of rice paddy at the county level from this study with Statistical Yearbook of Dongying Bureau of
Statistics (A, Left) and with the third national land survey of China (B, Right).
10 ª2022 The Authors. Remote Sensing in Ecology and Conservation published by John Wiley & Sons Ltd on behalf of Zoological Society of London.
Rice Paddy Expansion Impact on YRD Conservation C. Huang & C. Zhang
Intensified water conflicts caused by rice
paddy expansion in the YRD
This study showed that rice cultivation has increased
sharply in recent years in the YRD, although it is not a
traditional rice paddy planting region. The expansion of
rice cultivation can reduce food security pressure, but the
resulting water conflicts cannot be ignored (Zou
et al., 2018). Local surface freshwater resources are scarce
in the delta due to low precipitation and high vaporiza-
tion. At the same time, the high mineralization of shallow
groundwater hampers its utilization. As a result, Dongy-
ing City is heavily dependent on the Yellow River water
diversion. According to the Yellow River water resources
allocation scheme approved by the State Council in 1987,
the quota of water diverted from the Yellow River main
channel for Dongying City is 780 million m
3
/a, while in
2020, the total water consumption in Dongying City was
990 million m
3
, of which agriculture accounted for
60.5%. Due to the shortage of total water resources, there
is a serious conflict of water consumption among differ-
ent sectors.
Compared to dryland crops, rice is a high-water con-
sumption crop due to its special biophysical characteris-
tics and the very high water demand during the
transplanting and growing season (Dong & Xiao, 2016).
Moreover, most rice cultivation in the YRD utilizes land
with high salinity. As a medium salt-sensitive crop, rice
cultivation on saline land requires large amounts of fresh-
water to drench the soil salt (Bian et al., 2021). Referring
to the local standard in Dongying City, the irrigation
water amount for paddy fields is 330 m
3
/mu
(~0.0667 ha), which is much higher than that for irrigated
dryland (145 m
3
/mu). The rapid expansion of rice culti-
vation means more water resource consumption is needed
than before. For example, as the area of rice cultivation
increased from 4353 ha in 2001 to 24 367 ha in 2020, the
agricultural water consumption increased by approxi-
mately 0.56 billion m
3
.
The excessive water consumption in agriculture will
undoubtedly compete with other sectors, among which the
ecological water use is most likely to be misappropriated.
Studies have shown that in many developing countries,
despite increasing awareness of ecological conservation
among governments and residents, ecological water use
often gives way to industrial, agricultural, or domestic
water use when insufficient water resources create conflicts
between economic development and ecological conserva-
tion (Nikouei et al., 2012; Ouyang et al., 2020). In the
YRD, water for both agricultural irrigation and wetland
restoration comes from the Yellow River diversion. In gen-
eral, the highest volume occurs during the rainy season
from July to October, when 60% of the annual volume of
the river flows. Conversely, huge irrigation demand for
wetland vegetation growth and rice seeding occurs in
April–June when the Yellow River has a low volume (Li,
Huang, et al., 2020). With the expansion of rice paddy cul-
tivation, the agricultural irrigation water quota is far from
meeting the demand. As a result, ecological water has a
large chance to be inappropriately taken up, which
adversely affects the ecological restoration of the delta wet-
lands. How to balance the intensified conflict between rice
paddy water consumption and ecological requirements in
the YRD is worthy of further study.
The role of rice paddies for sustainable
conservation in the YRD
The study showed that rice cultivation in the YRD has
spread from both sides of the Yellow River Channel to the
surrounding areas (see Fig. 6). Although the rapid expan-
sion of rice paddy cultivation has caused increasing water
conflicts, the extensively distributed rice paddies over the
delta can be a critical complement to the degraded natural
wetlands, both in area and in ecological function.
The young wetland ecosystems in the delta have an
important status for biodiversity conservation in China
and the world. Since the late 1990 s, decreased sediment
loads to the delta, regulation of the river course to the
delta, and influences of agricultural development, have
led to a trend of rapidly decreasing natural wetlands and
significant shrinkage of habitats in the delta. Today, the
ecologically important habitats are mainly limited to pro-
tected areas, and most small wetland patches outside the
protected area have almost lost their ecological function
for biodiversity conservation. However, the large-scale
expansion of rice cultivation in the delta in recent years
has provided more habitat options for birds. Many stud-
ies have shown that, working as alternative wetlands, rice
fields are productive feeding grounds for many birds
(Fujioka et al., 2001,2010) and a nesting and resting
habitat for many migrating and wintering waterfowls
(Ib´a˜
nez et al., 2010; Tourenq et al., 2001), as well as pro-
viding suitable breeding conditions (Lane &
Fujioka, 1998). Especially after the decline in the quantity
and quality of natural wetlands, rice fields have become
more important as alternative habitats for many water-
birds (Ib´a˜
nez et al., 2010; Tourenq et al., 2001). During
our field surveys, many waterfowl were frequently found
resting or foraging in the rice fields under cultivation.
Although rice fields can provide ecological services sim-
ilar to those of natural wetlands (Natuhara, 2013), con-
sidering that the limited water resources in the YRD
region can hardly supply the water demand of large rice
fields, maintaining a certain scale of paddy fields and
optimizing their ecological value through proper
ª2022 The Authors. Remote Sensing in Ecology and Conservation published by John Wiley & Sons Ltd on behalf of Zoological Society of London. 11
C. Huang & C. Zhang Rice Paddy Expansion Impact on YRD Conservation
management is critical to sustainable conservation in the
context of water shortage in the YRD. Several studies
(Elphick, Baicich, et al., 2010; Elphick & Oring, 1998;
Elphick, Taft, & Lourenc
ßo, 2010) emphasized the impor-
tance of proper management for rice paddies as artificial
wetlands. Properly managed rice paddies can increase
landscape structure and improve landscape diversity in
rice-wetland coexistence areas (Maltchik et al., 2017). In
this study, we found the recent expansion of rice cultiva-
tion mainly resulted from industrial agriculture modes
characterized by large-scale, intensive production of rice
crops. The homogeneous pattern of rice paddies greatly
reduced the landscape diversity. Several studies have
demonstrated that constructing habitats with diverse
patches can greatly improve the ecological services pro-
vided by artificial rice paddies, especially in the absence
of natural wetlands (Pernollet et al., 2017; Ramachandran
et al., 2017; Strum et al., 2013). Taking the protected nat-
ure reserves as the core, and the paddy fields and other
small wetland patches outside the nature reserve as
important nodes, a composite agriculture-wetland net-
work can be constructed to improve the overall habitat
suitability and carrying capacity of the YRD. The close
proximity of nature reserves and the rice paddies favors
the constant movement of waterbirds between them and
the consequent interchange of material and energy in
both areas. Thus, maintaining a reasonable scale of rice
paddies can compensate for the impact of the loss of nat-
ural wetlands on ecological services and then fulfill the
overall sustainable conservation of the YRD.
Conclusions
Recent rapid expansion of rice paddies in the YRD has had
a significant impact in areas where rice paddies and wet-
lands coexist. This study proposed a stratified metrics-
based method which integrates statistical spectral indices
and phenological metrics to investigate rice cultivation
dynamics and then analyzed the impact of rice cultivation
expansion on ecological conservation in the YRD. We
found the rice paddy cultivation dynamics in the YRD can
be divided into three stages in the last 10 years: slight
increase from 2011 to 2015, rapid expansion from 2015 to
2018, and slight decrease but maintaining a high level from
2018 to 2021. The peak was reached in 2018, with the rice
paddy area of 28.18 ×10
3
ha, nearly eight times as much
as that in 2011. The rapid expansion of rice cultivation and
its high water consumption mean more freshwater resource
consumption, and the resulting intensified water conflict
will adversely affect wetland restoration and ecological con-
servation in the YRD. However, the large-scale expansion
of rice cultivation has also provided more alternative habi-
tats for many rare water birds, especially after the decline
in the quantity and quality of natural wetlands in the YRD.
Considering the limited water resources in the YRD region,
we suggest that maintaining a reasonable scale of rice pad-
dies and optimizing their distribution is a potential solu-
tion to fulfilling the overall achieving sustainable
conservation of the YRD in the context of water scarcity.
Acknowledgments
This work was supported by the Strategic Priority
Research Program of the CAS Earth Big Data Science
Project (XDA19060302). We are grateful to the anon-
ymous reviewers and the associate editor for providing
valuable comments and suggestions, which have greatly
improved this manuscript.
References
Aiello-Lammens, M.E., Chu-Agor, M.L., Convertino, M.,
Fischer, R.A., Linkov, I. & Resit Akc
ßakaya, H. (2011) The
impact of sea-level rise on Snowy Plovers in Florida:
integrating geomorphological, habitat, and metapopulation
models. Global Change Biology,17, 3644–3654.
Al-Shammari, D., Fuentes, I., Whelan, B.M., Filippi, P. &
Bishop, T. (2020) Mapping of cotton fields within-season
using phenology-based metrics derived from a time series of
landsat imagery. Remote Sensing,12, 3038.
Belgiu, M. & Csillik, O. (2018) Sentinel-2 cropland mapping
using pixel-based and object-based time-weighted dynamic
time warping analysis. Remote Sensing of Environment,204,
509–523.
Belgiu, M. & Dr˘agut
ß, L. (2016) Random forest in remote
sensing: a review of applications and future directions. ISPRS
Journal of Photogrammetry and Remote Sensing,114,24–31.
Bian, L., Wang, J., Liu, J. & Han, B. (2021) Spatiotemporal
changes of soil salinization in the yellow river delta of China
from 2015 to 2019. Sustainability,13, 822.
Borlaf-Mena, I., Badea, O. & Tanase, M.A. (2021) Assessing
the utility of Sentinel-1 coherence time series for temperate
and tropical forest mapping. Remote Sensing,13, 4814.
Boschetti, M., Busetto, L., Manfron, G., Laborte, A., Asilo, S.,
Pazhanivelan, S. et al. (2017) PhenoRice: a method for
automatic extraction of spatio-temporal information on rice
crops using satellite data time series. Remote Sensing of
Environment,194, 347–365.
Brisco, B., Li, K., Tedford, B., Charbonneau, F., Yun, S. &
Murnaghan, K. (2013) Compact polarimetry assessment for
rice and wetland mapping. International Journal of Remote
Sensing,34, 1949–1964.
Carrasco, L., Fujita, G., Kito, K. & Miyashita, T. (2022)
Historical mapping of rice fields in Japan using phenology
and temporally aggregated Landsat images in Google earth
engine. ISPRS Journal of Photogrammetry and Remote
Sensing,191, 277–289.
12 ª2022 The Authors. Remote Sensing in Ecology and Conservation published by John Wiley & Sons Ltd on behalf of Zoological Society of London.
Rice Paddy Expansion Impact on YRD Conservation C. Huang & C. Zhang
Clauss, K., Ottinger, M., Leinenkugel, P. & Kuenzer, C. (2018)
Estimating rice production in the Mekong Delta, Vietnam,
utilizing time series of Sentinel-1 SAR data. International
Journal of Applied Earth Observation and Geoinformation,73,
574–585.
Cui, B., Yang, Q., Yang, Z. & Zhang, K. (2009) Evaluating the
ecological performance of wetland restoration in the Yellow
River Delta, China. Ecological Engineering,35, 1090–1103.
da Cruz, R.P., Sperotto, R.A., Cargnelutti, D., Adamski, J.M.,
de FreitasTerra, T. & Fett, J.P. (2013) Avoiding damage and
achieving cold tolerance in rice plants. Food and Energy
Security,2,96–119.
Dong, J. & Xiao, X. (2016) Evolution of regional to global
paddy rice mapping methods: a review. ISPRS Journal of
Photogrammetry Remote Sensing,119, 214–227.
Dong,J.,Xiao,X.,Kou,W.,Qin,Y.,Zhang,G.,Li,L.etal.(2015)
Tracking the dynamics of paddy rice planting area in 1986–2010
through time series Landsat images and phenology-based
algorithms. Remote Sensing of Environment,160,99–113.
Dong, J., Xiao, X., Menarguez, M.A., Zhang, G., Qin, Y.,
Thau, D. et al. (2016) Mapping paddy rice planting area in
northeastern Asia with Landsat 8 images, phenology-based
algorithm and Google earth engine. Remote Sensing of
Environment,185, 142–154.
Elliott, J., Deryng, D., M¨
uller, C., Frieler, K., Konzmann, M.,
Gerten, D. et al. (2014) Constraints and potentials of future
irrigation water availability on agricultural production under
climate change. Proceedings of the National Academy of
Sciences of the United States of America,111, 3239–3244.
Elphick, C.S., Baicich, P., Parsons, K.C., Fasola, M. & Mugica,
L. (2010) The future for research on waterbirds in rice
fields. Waterbirds,33, 231–243.
Elphick, C.S. & Oring, L.W. (1998) Winter management of
Californian rice fields for waterbirds. Journal of Applied
Ecology,35,95–108.
Elphick, C.S., Taft, O. & Lourenc
ßo, P.M. (2010) Management
of rice fields for birds during the non-growing season.
Waterbirds,33, 181–192.
Farig, M., Shimizu, K., El-Kassar, G., Abou El Hassan, W.H.,
Fujimaki, H., Kita, I. et al. (2022) Influence of overplanting
paddy rice on irrigation water delivery performance: a case
study in the Dakalt branch canal, Nile Delta of Egypt. Paddy
and Water Environment,20,51–60.
Foody, G.M. (2002) Status of land cover classification accuracy
assessment. Remote Sensing of Environment,80, 185–201.
Fujioka, M., Lee, S.D., Kurechi, M. & Yoshida, H. (2010) Bird
use of rice fields in Korea and Japan. Waterbirds,33,8–29.
Fujioka, M., Yoshida, H. & Maeda, T. (2001) Value of fallow
farmlands as summer habitats for waterbirds in a Japanese
rural area. Ecological Research,16, 555–567.
G´omez, C., White, J.C. & Wulder, M.A. (2016) Optical
remotely sensed time series data for land cover classification:
a review. ISPRS Journal of Photogrammetry and Remote
Sensing,116,55–72.
Gorelick, N., Hancher, M., Dixon, M., Ilyushchenko, S., Thau,
D. & Moore, R. (2017) Google earth engine: planetary-scale
geospatial analysis for everyone. Remote Sensing of
Environment,202,18–27.
Griffiths, P., Nendel, C. & Hostert, P. (2019) Intra-annual
reflectance composites from Sentinel-2 and Landsat for
national-scale crop and land cover mapping. Remote Sensing
of Environment,220, 135–151.
Guan, X., Huang, C., Liu, G., Meng, X. & Liu, Q. (2016)
Mapping rice cropping systems in Vietnam using an NDVI-
based time-series similarity measurement based on DTW
distance. Remote Sensing,8, 19.
Gumma, M.K., Thenkabail, P.S., Maunahan, A., Islam, S. &
Nelson, A. (2014) Mapping seasonal rice cropland extent and
area in the high cropping intensity environment of Bangladesh
using MODIS 500 m data for the year 2010. ISPRS Journal of
Photogrammetry and Remote Sensing,91,98–113.
Hall, D.K. & Riggs, G.A. (2011) Normalized-difference snow
index (NDSI). In Encyclopedia of snow, ice and glaciers.
Dordrecht: Springer, pp. 779–780.
Huang, C., Zhang, C., He, Y., Liu, Q., Li, H., Su, F. et al.
(2020) Land cover mapping in cloud-prone tropical areas
using Sentinel-2 data: integrating spectral features with
NDVI temporal dynamics. Remote Sensing,12, 1163.
Huete, A., Didan, K., Miura, T., Rodriguez, E.P., Gao, X. &
Ferreira, L.G. (2002) Overview of the radiometric and
biophysical performance of the MODIS vegetation indices.
Remote Sensing of Environment,83, 195–213.
Huete, A., Liu, H., Batchily, K. & Van Leeuwen, W. (1997) A
comparison of vegetation indices over a global set of TM
images for EOS-MODIS. Remote Sensing of Environment,59,
440–451.
Ib´a˜
nez, C., Curc´o, A., Riera, X., Ripoll, I. & S´anchez, C. (2010)
Influence on birds of rice field management practices during
the growing season: a review and an experiment. Waterbirds,
33, 167–180.
Jain, M., Mondal, P., DeFries, R.S., Small, C. & Galford, G.L.
(2013) Mapping cropping intensity of smallholder farms: a
comparison of methods using multiple sensors. Remote
Sensing of Environment,134, 210–223.
Jeong, S., Kang, S., Jang, K., Lee, H., Hong, S. & Ko, D.
(2012) Development of variable threshold models for
detection of irrigated paddy rice fields and irrigation timing
in heterogeneous land cover. Agricultural Water
Management,115,83–91.
Jin, C., Xiao, X., Dong, J., Qin, Y. & Wang, Z. (2016)
Mapping paddy rice distribution using multi-temporal
Landsat imagery in the Sanjiang plain, Northeast China.
Frontiers of earth science,10,49–62.
Jo, H.-W., Lee, S., Park, E., Lim, C.-H., Song, C., Lee, H. et al.
(2020) Deep learning applications on multitemporal SAR
(Sentinel-1) image classification using confined labeled data:
the case of detecting rice paddy in South Korea. IEEE
Transactions on Geoscience and Remote Sensing,58,7589–7601.
ª2022 The Authors. Remote Sensing in Ecology and Conservation published by John Wiley & Sons Ltd on behalf of Zoological Society of London. 13
C. Huang & C. Zhang Rice Paddy Expansion Impact on YRD Conservation
Lane, S.J. & Fujioka, M. (1998) The impact of changes in
irrigation practices on the distribution of foraging egrets
and herons (Ardeidae) in the rice fields of Central Japan.
Biological Conservation,83, 221–230.
Le Toan, T., Ribbes, F., Wang, L.-F., Floury, N., Ding, K.-H.,
Kong, J.A. et al. (1997) Rice crop mapping and monitoring
using ERS-1 data based on experiment and modeling results.
IEEE Transactions on Geoscience Remote Sensing,35,41–56.
Lhermitte, S., Verbesselt, J., Verstraeten, W.W. & Coppin, P.
(2011) A comparison of time series similarity measures for
classification and change detection of ecosystem dynamics.
Remote Sensing of Environment,115, 3129–3152.
Li, H., Fu, D., Huang, C., Su, F., Liu, Q., Liu, G. et al. (2020)
An approach to high-resolution rice paddy mapping using
time-series Sentinel-1 SAR data in the Mun River basin,
Thailand. Remote Sensing,12, 3959.
Li, H., Huang, C., Liu, Q. & Liu, G. (2020) Accretion–erosion
dynamics of the yellow river delta and the relationships with
runoff and sediment from 1976 to 2018. Water,12, 2992.
Li, Q., Zhang, H., Du, X., Wen, N. & Tao, Q. (2014) County-
level rice area estimation in southern China using remote
sensing data. Journal of Applied Remote Sensing,8, 083657.
Liaw, A. & Wiener, M. (2002) Classification and regression by
randomForest. R News,2,18–22.
Liu, L., Xiao, X., Qin, Y., Wang, J., Xu, X., Hu, Y. et al.
(2020) Mapping cropping intensity in China using time
series Landsat and Sentinel-2 images and Google earth
engine. Remote Sensing of Environment,239, 111624.
Maltchik, L., Stenert, C. & Batzer, D.P. (2017) Can rice field
management practices contribute to the conservation of
species from natural wetlands? Lessons from Brazil. Basic
and Applied Ecology,18,50–56.
Mao, D., Luo, L., Wang, Z., Wilson, M.C., Zeng, Y., Wu, B.
et al. (2018) Conversions between natural wetlands and
farmland in China: a multiscale geospatial analysis. Science
of the Total Environment,634, 550–560.
Minasny, B., Shah, R.M., Che Soh, N., Arif, C. & Indra
Setiawan, B. (2019) Automated near-real-time mapping and
monitoring of rice extent, cropping patterns, and growth
stages in Southeast Asia using Sentinel-1 time series on a
Google earth engine platform. Remote Sensing,11, 1666.
Murray, N.J., Phinn, S.R., DeWitt, M., Ferrari, R., Johnston,
R., Lyons, M.B. et al. (2019) The global distribution and
trajectory of tidal flats. Nature,565, 222–225.
Natuhara, Y. (2013) Ecosystem services by paddy fields as
substitutes of natural wetlands in Japan. Ecological
Engineering,56,97–106.
Ni, R., Tian, J., Li, X., Yin, D., Li, J., Gong, H. et al. (2021)
An enhanced pixel-based phenological feature for accurate
paddy rice mapping with Sentinel-2 imagery in Google earth
engine. ISPRS Journal of Photogrammetry and Remote
Sensing,178, 282–296.
Nikouei, A., Zibaei, M. & Ward, F.A. (2012) Incentives to
adopt irrigation water saving measures for wetlands
preservation: an integrated basin scale analysis. Journal of
Hydrology,464, 216–232.
Ouyang, Z., Wang, H., Lai, J., Wang, C., Liu, Z., Sun, Z. et al.
(2020) New approach of high-quality agricultural
development in the Yellow River Delta. Bulletin of the
Chinese Academy of Sciences,35, 145–153.
Pernollet, C.A., Cavallo, F., Simpson, D., Gauthier-Clerc, M. &
Guillemain, M. (2017) Seed density and waterfowl use of
rice fields in Camargue, France. The Journal of Wildlife
Management,81,96–111.
Petitjean, F., Inglada, J. & Ganc
ßarski, P. (2012) Satellite image
time series analysis under time warping. IEEE Transactions
on Geoscience and Remote Sensing,50, 3081–3095.
Qin, Y., Xiao, X., Dong, J., Zhang, G., Shimada, M., Liu, J.
et al. (2015) Forest cover maps of China in 2010 from
multiple approaches and data sources: PALSAR, Landsat,
MODIS, FRA, and NFI. ISPRS Journal of Photogrammetry
and Remote Sensing,109,1–16.
Ramachandran, R., Kumar, A., Gopi Sundar, K.S. & Bhalla,
R.S. (2017) Hunting or habitat? Drivers of waterbird
abundance and community structure in agricultural
wetlands of southern India. Ambio,46, 613–620.
Roberts, D., Numata, I., Holmes, K., Batista, G., Krug, T.,
Monteiro, A. et al. (2002) Large area mapping of land-cover
change in Rondˆonia using multitemporal spectral mixture
analysis and decision tree classifiers. Journal of Geophysical
Research: Atmospheres,107, LBA 40-1–LBA 40-18.
Sakamoto, T., Van Phung, C., Kotera, A., Nguyen, K.D. &
Yokozawa, M. (2009) Analysis of rapid expansion of inland
aquaculture and triple rice-cropping areas in a coastal area
of the Vietnamese Mekong Delta using MODIS time-series
imagery. Landscape and Urban Planning,92,34–46.
Sakamoto, T., Yokozawa, M., Toritani, H., Shibayama, M.,
Ishitsuka, N. & Ohno, H. (2005) A crop phenology
detection method using time-series MODIS data. Remote
Sensing of Environment,96, 366–374.
Shi, J. & Huang, J. (2015) Monitoring spatio-temporal
distribution of rice planting area in the Yangtze River Delta
region using MODIS images. Remote Sensing,7, 8883–8905.
Singha, M., Dong, J., Zhang, G. & Xiao, X. (2019) High resolution
paddy rice maps in cloud-prone Bangladesh and Northeast
India using Sentinel-1 data. Scientific Data,6,1–10.
Son, N.-T., Chen, C.-F., Chen, C.-R., Duc, H.-N. & Chang, L.-
Y. (2013) A phenology-based classification of time-series
MODIS data for rice crop monitoring in Mekong Delta,
Vietnam. Remote Sensing,6, 135–156.
Strum, K.M., Reiter, M.E., Hartman, C.A., Iglecia, M.N.,
Kelsey, T.R. & Hickey, C.M. (2013) Winter management of
California’s rice fields to maximize waterbird habitat and
minimize water use. Agriculture, Ecosystems & Environment,
179, 116–124.
Torres, R., Snoeij, P., Geudtner, D., Bibby, D., Davidson, M.,
Attema, E. et al. (2012) GMES Sentinel-1 mission. Remote
Sensing of Environment,120,9–24.
14 ª2022 The Authors. Remote Sensing in Ecology and Conservation published by John Wiley & Sons Ltd on behalf of Zoological Society of London.
Rice Paddy Expansion Impact on YRD Conservation C. Huang & C. Zhang
Tourenq, C., Bennetts, R.E., Kowalski, H., Vialet, E., Lucchesi, J.-
L., Kayser, Y. et al. (2001) Are ricefields a good alternative to
natural marshes for waterbird communities in the Camargue,
southern France? Biological Conservation,100,335–343.
Tucker, C.J. (1979) Red and photographic infrared linear
combinations for monitoring vegetation. Remote Sensing of
Environment,8, 127–150.
Walker, J., De Beurs, K. & Wynne, R. (2014) Dryland
vegetation phenology across an elevation gradient in
Arizona, USA, investigated with fused MODIS and Landsat
data. Remote Sensing of Environment,144,85–97.
Wang, H., Wu, X., Bi, N., Li, S., Yuan, P., Wang, A. et al. (2017)
Impacts of the dam-orientated water-sediment regulation
scheme on the lower reaches and delta of the Yellow River
(Huanghe): a review. Global and Planetary Change,157,93–113.
Wang, J., Xiao, X., Liu, L., Wu, X., Qin, Y., Steiner, J.L. et al.
(2020) Mapping sugarcane plantation dynamics in Guangxi,
China, by time series Sentinel-1, Sentinel-2 and Landsat
images. Remote Sensing of Environment,247, 111951.
Wang, J., Xiao, X., Qin, Y., Dong, J., Zhang, G., Kou, W. et al.
(2015) Mapping paddy rice planting area in wheat-rice
double-cropped areas through integration of Landsat-8 OLI,
MODIS and PALSAR images. Scientific Reports,5,1–11.
Wang, M., Qi, S. & Zhang, X. (2012) Wetland loss and
degradation in the Yellow River Delta, Shandong Province
of China. Environmental Earth Sciences,67, 185–188.
Wang, X., Lian, Y., Huang, C., Wang, X., Wang, R., Shan, K.
et al. (2012) Environmental flows and its evaluation of
restoration effect based on LEDESS model in Yellow River
Delta wetlands. Mitigation and Adaptation Strategies for
Global Change,17, 357–367.
Wei, J., Cui, Y., Luo, W. & Luo, Y. (2022) Mapping paddy
rice distribution and cropping intensity in China from 2014
to 2019 with Landsat images, effective flood signals, and
Google earth engine. Remote Sensing,14, 759.
Xiao, X., Boles, S., Frolking, S., Li, C., Babu, J.Y., Salas, W.
et al. (2006) Mapping paddy rice agriculture in south and
Southeast Asia using multi-temporal MODIS images.
Remote Sensing of Environment,100,95–113.
Xiao, X., Boles, S., Liu, J., Zhuang, D., Frolking, S., Li, C. et al.
(2005) Mapping paddy rice agriculture in southern China
using multi-temporal MODIS images. Remote Sensing of
Environment,95, 480–492.
Xiao, X., Hollinger, D., Aber, J., Goltz, M., Davidson, E.A.,
Zhang, Q. et al. (2004) Satellite-based modeling of gross
primary production in an evergreen needleleaf forest.
Remote Sensing of Environment,89, 519–534.
Xiao, X., Zhang, Q., Hollinger, D., Aber, J. & Moore, B., III.
(2005) Modeling gross primary production of an evergreen
needleleaf forest using MODIS and climate data. Ecological
Applications,15, 954–969.
Xie, T., Liu, X. & Sun, T. (2011) The effects of groundwater
table and flood irrigation strategies on soil water and salt
dynamics and reed water use in the Yellow River Delta,
China. Ecological Modelling,222, 241–252.
Xu, H., Tian, Z., He, X., Wang, J., Sun, L., Fischer, G. et al.
(2019) Future increases in irrigation water requirement
challenge the water-food nexus in the northeast farming
region of China. Agricultural Water Management,213, 594–
604.
Yang, X., Xiao, X., Qin, Y., Wang, J. & Neal, K. (2021)
Mapping forest in the southern Great Plains with ALOS-2
PALSAR-2 and Landsat 7/8 data. International Journal of
Applied Earth Observation and Geoinformation,104, 102578.
Zhang, G., Xiao, X., Dong, J., Kou, W., Jin, C., Qin, Y. et al.
(2015) Mapping paddy rice planting areas through time
series analysis of MODIS land surface temperature and
vegetation index data. ISPRS Journal of Photogrammetry and
Remote Sensing,106, 157–171.
Zhang, G., Xiao, X., Dong, J., Xin, F., Zhang, Y., Qin, Y. et al.
(2020) Fingerprint of rice paddies in spatial–temporal
dynamics of atmospheric methane concentration in
monsoon Asia. Nature Communications,11,1–11.
Zhang, H., Chen, X. & Luo, Y. (2016) An overview of
ecohydrology of the Yellow River delta wetland.
Ecohydrology & Hydrobiology,16,39–44.
Zhao, S., Liu, X., Ding, C., Liu, S., Wu, C. & Wu, L. (2020)
Mapping rice paddies in complex landscapes with
convolutional neural networks and phenological metrics.
GIScience & Remote Sensing,57,37–48.
Zhong, L., Hu, L., Yu, L., Gong, P. & Biging, G.S. (2016)
Automated mapping of soybean and corn using phenology.
ISPRS Journal of Photogrammetry and Remote Sensing,119,
151–164.
Zhou, Y., Xiao, X., Qin, Y., Dong, J., Zhang, G., Kou, W. et al.
(2016) Mapping paddy rice planting area in rice-wetland
coexistent areas through analysis of Landsat 8 OLI and
MODIS images. International Journal of Applied Earth
Observation and Geoinformation,46,1–12.
Zou, Y., Duan, X., Xue, Z., Mingju, E., Sun, M., Lu, X. et al.
(2018) Water use conflict between wetland and agriculture.
Journal of Environmental Management,224, 140–146.
ª2022 The Authors. Remote Sensing in Ecology and Conservation published by John Wiley & Sons Ltd on behalf of Zoological Society of London. 15
C. Huang & C. Zhang Rice Paddy Expansion Impact on YRD Conservation
