Agricultural Water Management 212 (2019) 162–171

Contents lists available at ScienceDirect

Agricultural Water Management

journal homepage: www.elsevier.com/locate/agwat

Assessment of the relationship between adoption of a knowledge-intensive T

water-saving technique and irrigation conditions in the Mekong Delta of
Vietnam
⁎
Takayoshi Yamaguchi , Luu Minh Tuan, Kazunori Minamikawa, Shigeki Yokoyama
National Agriculture and Food, Research Organization, Japan

A R T I C LE I N FO A B S T R A C T

Keywords: Alternate wetting and drying (AWD) is an irrigation technique developed to save water in paddy rice farming.
Alternate wetting and drying (AWD) We investigated how local paddy irrigation conditions affected the adoption of AWD by farmers in An Giang
Continuous flooding province of the Mekong River Delta region, Vietnam. To this end, we used community-level statistics collected in
Paddy rice farming the dry seasons of 2009–2010, 2011–2012, 2013–2014, and 2014–2015. We focused on paddy elevation, density
of channel networks, and degree of triple-cropping as indicators of irrigation conditions. We performed a partial
rank order correlation coefficient analysis between AWD adoption rates and these irrigation conditions and
several other related factors. Variations in irrigation conditions caused by paddy elevation and infrastructure
status were identified as affecting AWD implementation. Although AWD is a knowledge-intensive practice for
optimizing watering amounts and timing in paddy rice farming, and requires no extra infrastructure or ma-
chinery, it does require well-arranged irrigation conditions. Moreover, the significance of the channel network in
AWD adoption was emphasized in a year with low rainfall, indicating that farmers prefer continuous flooding to
AWD as a hedge against water shortages. In AWD, continuous flooding is considered a waste of water from the
viewpoint of rice plant physiology; however, where irrigation conditions are less than ideal, continuous flooding
includes the role of water reservoirs. Therefore, the prerequisite conditions for diffusion of AWD need to be
reconsidered.

1. Introduction
As the global population continues to steadily increase, staple crop
yields must also increase to meet the demand for food (FAO, 2011).
Water is an essential resource for stable agricultural production, but as
competition for water resources between domestic, industrial, and
agricultural sectors increases, the agricultural sector must increase food
production with limited water resources, and improving agricultural
water productivity has become an urgent issue (Hundertmark and
Facon, 2003). Paddy rice farming, in particular, requires more water
than other staple crops such as wheat and maize because irrigated rice
paddies are generally flooded from the transplantation (or sowing)
stage to harvest (Tuong and Bouman, 2003). Thus, the establishment of
water-saving techniques is particularly vital in rice farming.
The International Rice Research Institute (IRRI) has found that
paddy rice only needs to be flooded during the rooting and flowering
stages (Van der Hoek et al., 2001). Consequently, IRRI developed a rice-
growing procedure called “Alternate Wetting and Drying” (AWD),
whereby paddy fields are only intermittently irrigated outside of these
critical stages. A “safe-AWD” procedure for this intermittent irrigation
stage can be described as follows: (1) a perforated plastic pipe is set into
the ground to precisely monitor the belowground water level; (2)
farmers supply no more than 5 cm of water to their paddies and then do
not irrigate again until the water level naturally declines to 10–15 cm
below the soil surface; (3) this process is repeated several times after
the rooting stage up until the flowering stage. When farmers follow this
procedure, total water usage can be reduced by 10%–40% relative to
continuous flooding with no significant negative impact on yield, im-
proving agricultural water productivity (e.g. Lampayan et al., 2015).
Hence, IRRI has attempted to spread the AWD practice in various Asian
rice-growing regions (e.g., Sibayan et al., 2010).
In addition to saving water, AWD can also reduce farmers’ pumping
costs and irrigation charges, thereby directly increasing their net in-
come. Moreover, because AWD is a knowledge-intensive technique (e.g.
Aakkula et al., 2005; Pingali et al., 1998) for optimizing irrigation
amounts and timing in paddies based on an understanding of soil water

⁎
Corresponding author.
E-mail address: pastoralism@gmail.com (T. Yamaguchi).

https://doi.org/10.1016/j.agwat.2018.08.041
Received 28 February 2018; Received in revised form 22 August 2018; Accepted 27 August 2018
0378-3774/ © 2018 Published by Elsevier B.V.
T. Yamaguchi et al.
Fig. 1. Map of An Giang Province.
conditions by means of a simple instrument, it can be adopted by local
farmers at low cost (Kurschner et al., 2010). Therefore, studies on AWD
diffusion have emphasized how the proper understanding and practice
of the technique are imparted to the farmers (Kurschner et al., 2010;
Rejesus et al., 2013).
However, based on empirical studies in China, Li (2001) and Loeve
et al. (2001) raised concerns that the successful implementation of
AWD requires not only a proper understanding and practice of the
technique but also reliable and flexible irrigation conditions. A similar
situation was also shown in a qualitative study in the Mekong Delta of
Vietnam (Yamaguchi et al., 2017), where farmers hesitated to adopt
AWD in paddies that did not directly connect with water channels or
were located at higher elevations, because such paddies tended to have
unstable water conditions (Yamaguchi et al., 2017). Those studies in-
dicated that farmers would be prepared to adopt AWD when they were
confident that their irrigation conditions were reliable and timely en-
ough to allow irrigation to be carried out according to schedule
(Mushtaq et al., 2006), and conversely that AWD might not be easily
accepted in areas that have suffered from water scarcity, despite the
fact that AWD was developed as a water saving technique. Moreover,
one of the AWD’s merits is its low cost for farmers to implement;
however, if implementation of AWD requires a considerable depen-
dence on irrigation conditions, then factors other than just introductory
costs must also be examined. These factors are likely to be of significant
concern when considering AWD diffusion. However, the relationship
between adoption of AWD and irrigation conditions has not been ver-
ified with quantitative data. Hence, this study aimed to investigate the
linkage between AWD diffusion and water supply conditions by using
quantitative data from the Mekong Delta, Vietnam.
The delta regions of South and Southeast Asian countries are good
candidates for AWD implementation (Kurschner et al., 2010; Quynh
163
Agricultural Water Management 212 (2019) 162–171
and Sander, 2015). Delta basins such as the Mekong, Irrawaddy, Chao
Phraya, and Bengal have long been primary rice production bases
(Barker et al., 1985), supporting the growing Asian population. How-
ever, the current sea level rise due to global climate change has caused
saline water intrusion into these regions, creating a shortage of water
for rice farming (Wassmann et al., 2004; IBRD and World Bank, 2010;
Driel and Nauta, 2013; Rahman and Bhattacharya, 2014). Moreover,
groundwater overdraft and decreasing river flows have also contributed
to deteriorating agricultural water conditions (MacDonald et al., 2015;
Thilakarathne and Sridhar, 2017). AWD has the potential to alleviate
the negative impacts of these issues on rice farming in these regions.
The Mekong Delta region accounts for more than half of the do-
mestic rice production in Vietnam, which is a leading rice exporter.
Nevertheless, for the reasons outlined above, it is increasingly difficult
for rice farmers in this region to obtain the required amount of water.
Therefore, water-saving technologies are particularly crucial for rice
farming in this region.
The Mekong Delta is a wide floodplain; therefore, slight variations
in elevation have a substantial impact on drainage, access to irrigation
water, and susceptibility to flooding, and these factors significantly
influence land cover and rice farm management in this region (Kaida,
1974; Thuy and Anh, 2015). We hypothesized that certain paddy ele-
vations would hinder farmers’ adoption of AWD by impeding water
control. This study evaluates the irrigation conditions of the Mekong
Delta rice farming according to “paddy elevation” and “water channel
density”.
Mekong Delta farmers have endured water shortages during the dry
season. In response, channel networks have grown as rice farming has
intensified (Kono, 2001). In the areas where channel networks are
limited, farmers occasionally use plot-to-plot irrigation, which leads to
difficulties in controlling paddy water (Yamaguchi et al., 2017).
T. Yamaguchi et al. Agricultural Water Management 212 (2019) 162–171

Table 1
AWD diffusion in An Giang Province during the 2009–2010, 2011–2012,
2013–2014, and 2014–2015 dry seasons.
Source: AGPPSD (2016).
2009-2010 2011-2012 2013-2014 2014-2015

Paddy area
Total (ha) 234,247 235,941 236,987 238,314
With AWD (ha) 41,952 87,624 117,853 123,223
AWD rates (%) 18 37 50 52

Rice farmers
Total (HHs) 228,165 237,705 243,003 227,039
Adopted AWD (HHs) 39,889 107,152 134,677 121,718
AWD rates (%) 18 45 55 54

Note: HHs: households.

In 2005, the local government, in collaboration with IRRI, in-

troduced AWD as a measure to lower pumping costs (Meier, 2007). The
local government has proactively arranged workshops on new tech-
nologies, including AWD, for farmers, and has set up model paddies
showcasing advanced rice farming techniques all around An Giang
(Meier, 2007; Diangkinay-Quicho, 2013; Yamaguchi et al., 2016). AWD
has been steadily adopted. By the dry season of 2014–2015, the total
area of paddies where AWD had been implemented had increased to
Fig. 2. Correlation between annual climate and rice-farming calendar in An about 120,000 ha, accounting for 52% of the total paddy area, and the
Giang Province. Temperature data is the 2009–2014 average. number of rice farming households that had implemented AWD ex-
Source: Climate data based on SOAG (2012) and SOAG (2015). ceeded 120,000 (Table 1). Compared to the situation in the Philippines,
which has approximately 100,000 AWD farmers in the entire country
Therefore, the correlation between AWD adoption and water channel despite it being the country where AWD diffusion originated
development is also assessed. (Lampayan, 2013), it can be said that the diffusion of AWD in An Giang
has been successful. Nevertheless, for unknown reasons, there were
disparities in the rates of AWD adoption among the local communities
2. Study area: a brief history of rice farming and AWD
(Yamaguchi et al., 2016). The aim of the present study was to use a
geographic information system (GIS) and statistical methods to identify
This study analyses AWD dissemination and implementation based
the effects of the irrigation conditions on provincial-scale AWD dis-
on the case of An Giang province, Mekong River Delta region, Vietnam
semination and implementation.
(Fig. 1). An Giang is the second largest rice producer in the Mekong
Delta (MAFFJ, 2011). Rice paddies account for about 70% of the total
provincial area of 340,000 ha (SOAG, 2015). The monthly average
3. Data and methodology
temperature exceeds 25 °C throughout the year and is very suitable for
paddy rice farming. Precipitation, however, varies notably between the
3.1. AWD data
May–October rainy season and the December–March dry season
(Fig. 2).
The administrative units of Vietnam are divided into first, second,
Prior to the 1950s, rice farming in the Mekong Delta was severely
and third tiers: (1) provincial-level units, (2) district-level units, and (3)
limited by water shortages during the dry season. In addition, rainy
administrative units at the communal level (World Bank, 2011); this
season farming was disturbed by the flooding of the Mekong River, so
study refers to the third-tier units as “communes”. As of 2014, An Giang
the only viable option was single cropping with floating rice (Xuan,
had 156 communes (SOAG, 2015).
1975; Hieu, 2010).
The An Giang Plant Protection Sub-Department (AGPPSD), the
After the 1960s, early maturing rice cultivars and irrigation pumps
agency that assumes the role of introducing agricultural innovations to
were introduced. Embankments were constructed to protect paddies
farmers, has published AWD adoption statistics for each commune since
from inundation except for flooding during the mid rainy season
2009. We used the data collected for the 2009–2010, 2011–2012,
(September–November). Double rice cropping from the dry season to
2013–2014, and 2014–2015 dry seasons (Table 1).
the start of the rainy season became common practice. The rate of
Each commune has at least one AGPPSD extension agent; AWD
adoption of double cropping by area in An Giang exceeded 90% by the
adoption rates have been reported by extension agents through direct
early 1990s (Otsuka, 2014).
observation of irrigation management in each paddy. However, there
Since the end of the 1990s, embankments have been raised and
are no rigid criteria to confirm whether farmers adopted AWD.
large-capacity pumps have been installed to drain rainwater from
Therefore, the AWD statistics could include accidental errors and out-
within the paddies to outside the embankments (Hieu, 2010; Tran et al.,
liers. Thus, the present study used nonparametric methods that are
2018). The infrastructure consisting of raised embankments and drai-
robust to outliers (Xiao et al., 2016).
nage pumps is known as a full-dike system. It has enabled triple-crop-
We conducted unstructured-interviews on AWD implementation in
ping, which had an adoption rate of 66.2% in this area by 2014 (SOAG,
An Giang with AGPPSD provincial staff members and extension agents
2015).
in September 2014 (Yamaguchi et al., 2016) and May and September
Intensification of rice farming in An Giang increased rice production
2015 (Yamaguchi et al., 2017); we complementarily utilized these
in the region, but because of increased production costs, grower rev-
qualitative data in the interpretation of the results.
enues did not commensurately increase. Thus, the provincial govern-
ment introduced a set of new measures to enhance the efficiency of rice
farming (Chi et al., 2013). AWD was one of those measures.

164
T. Yamaguchi et al.
Fig. 3. Correlation between paddy areas estimated with polygons and those determined from AGPPSD commune statistics in An Giang Province. Statistics were
derived from the 2014–2015 dry season data. The coefficient of determination (R2) was 0.94.
Source: AGPPSD (2016).
3.2. Paddy polygons
All GIS processing in this study was performed with ArcGIS v. 9.3.1
(Environmental Systems Research Institute Inc., Redlands, CA, USA).
Paddy polygons within An Giang were based on a map (1:50,000) is-
sued by MNREV (2003), and the paddy area of each commune was
based on these polygons. We performed Pearson’s correlation coeffi-
cient analysis between the estimated paddy area based on the paddy
polygons and the paddy area obtained from the AGPPSD statistics for
each commune in the 2014–2015 dry season. The results were well
correlated (Fig. 3), but the base map was published more than 10 years
earlier. The annual rate of increase in the provincial paddy area was
very low (0.7% y−1) from 2005 to 2015 (SOAG, 2015; AGPPSD, 2016).
Therefore, we concluded that the paddy polygons derived from the map
published in 2003 corresponded sufficiently well to the current paddy
distribution.
3.3. Paddy classification by elevation
Based on a topographic map (1:50,000) issued by AGPPSD, we ex-
tracted about 4000 elevation data points in 0.1 m increments. The map
was missing data for two small areas, so we complemented the data
with elevation point data from the MNREV map (2003). By using the
elevation data included in the paddy polygons, we created a paddy area
elevation model (50 m resolution) by inverse distance-weighted inter-
polation. We classified paddy areas into six elevation categories
(≤1 m; > 1–2 m; > 2–3 m; > 3–4 m; > 4–5 m; and > 5 m), and calcu-
lated the proportions of paddy area for each category relative to the
total paddy area in each commune. We performed Spearman’s rank
order correlation coefficient (ROCC) analysis between the AWD adop-
tion rate and the proportion of paddy area for each elevation category
and within each commune. The ‘ppcor’ package in R (Kim, 2015) was
used for the statistical analysis. Through this analysis, we estimated the
paddy elevations suitable for AWD implementation.
3.4. Irrigation networks
We traced the irrigation networks using channel and river polylines
derived from the MNREV map (2003). AGPPSD staff members stated
165
Agricultural Water Management 212 (2019) 162–171
that most of the major channels were constructed before the 1990s
when double-cropping was widespread; a number of narrow channels,
however, were constructed after 2000. Implementation of full-dike
systems created an opportunity to construct new channels to improve
drainage. Measurement of the extent of full-dike system construction
would provide a more accurate representation of current channel net-
work development, but there are no published statistics for paddy areas
enclosed by full-dike systems. Because triple-cropping is only feasible in
paddies having a full-dike system (Hieu, 2010; Tran et al., 2018), we
substituted the data of triple-cropping paddy area for each commune.
Information on triple-cropping paddy areas has been published only
for each district as a whole; therefore, we estimated commune-level
data. Previous studies (Sakamoto et al., 2009; Son et al., 2014) have
distinguished between double- and triple-cropping paddies in the Me-
kong Delta region by using the Extended Vegetation Index (EVI) from
MODIS satellite images. Three EVI peaks appear in triple-cropping
paddies, but EVI values are lower in double-cropping paddies during
the mid-rainy season (September–November) (Fig. 4). Therefore, we
were able to estimate the area of triple-cropping paddies by focusing on
the EVI values for the mid-rainy season.
Using the MOD13Q1 data in MODIS (Didan et al., 2015), we com-
piled images from the first half of November 2009, 2011, 2013, and
2014 (see Table 4 in the Results). We classified the pixels within paddy
polygons as planted or non-planted based on the EVI thresholds. We
increased the EVI thresholds from 500 to 6000 in increments of 100,
and compared the coefficients of determination (R2) between the esti-
mated and statistics-based values for triple-cropping paddies by district.
Using this information, we selected the most consistent EVI threshold.
3.5. Partial rank order correlation coefficient analysis
This study aimed to verify the effects of irrigation conditions on
AWD implementation. We used paddy elevation, density of channel
networks, and degree of triple-cropping as indicators of irrigation
conditions of respective paddies, and we added five other variables that
might also be correlated with AWD diffusion (these eight factors are
summarized in Table 2). We then performed partial rank order corre-
lation coefficient (PROCC) analysis on the correlations of these para-
meters with the AWD adoption rate for each commune by using the
T. Yamaguchi et al. Agricultural Water Management 212 (2019) 162–171

Fig. 4. Seasonal EVI transitions in double- and triple-cropping paddies in An Giang Province. Data were acquired in 2014 and derived from MOD13Q1 in MODIS.
Double- and triple-cropping paddies were identified by AGPPSD personnel during the 2015 field investigation.

‘ppcor’ package in R. PROCC measures the magnitude and direction of
the association between two ranked variables while it controls for the
effects of at least one of the other variables (Sheskin, 2000).
4. Results and discussion
4.1. Trends and regional differences in AWD diffusion in An Giang
By 2009–2010, the provincial capital Long Xuyen already had a
high AWD adoption rate (80%) (Fig. 5). Because AWD is a knowledge-
intensive technique, Diangkinay-Quicho (2013) and Chi et al. (2015)
reported that government efforts to transmit proper information to
farmers had been a significant factor in the diffusion of AWD in An
Giang. Long Xuyen was the first district where AGPPSD and IRRI col-
laborated on AWD diffusion in 2005 (Meier, 2007) and it is to be ex-
pected that such initial activities would promote AWD diffusion there.
In contrast, all of the other districts at that time had AWD adoption
rates of less than 30%.
In 2009–2010, there was also wide variation in the AWD adoption
rates among the communes. For example, in Tri Ton district where the
AWD adoption rate was only 20%, the Tan Tuyen and Ba Chuc com-
munes had adoption rates of 75% and 62%, respectively, whereas the
Ta Danh and Luong Phi communes had rates below 5%. From
2005–2009, approximately 500 farmers’ workshops on agricultural in-
novation were held in An Giang (AGPPSD, 2014). However, according
to AGPPSD staff members, the venues for such workshops tended to be

Table 2
Variables used in the partial rank order correlation coefficient (PROCC) analysis with AWD adoption rate.
Variable Calculation methods Referred Provincial By commune
year scale
Median Min. Max.

1. Proportion of paddies located at elevations suitable for AWD Suitable elevation paddies for AWD (ha) / total paddies 2003 75 94 2 100
(%) (ha)
2. Channel density Length of channels and rivers (km) / commune area (ha)a 2003 66 70 21 105
(m ha−1)
3. Proportion of triple-cropping paddies (%) Triple-cropping paddies (ha) 2009 44 45 0 100
/ total paddies (ha) 2011 63 77 1 100
2013 77 95 6 100
2014 71 85 5 100
4. Commune is in the provincial capital (binary dummy variable) Categorical variables –
(yes: 11 communes)
−1
5. Population density (HHs ha ) Total household numberb 2015 1.2 1.3 0.2 17.5
/ commune area (ha)a
6. Road density (m ha−1) Road length (km)c 2012 16 16 5 105
/ commune area (ha)a
7. Paddy management scale (ha HH−1) Total paddies (ha) 2009-2010 1.0 0.9 0.2 19.3
/ total rice farming householdsd 2011-2012 0.9 0.9 0.1 8.8
2013-2014 0.9 0.9 0.1 9.2
2014-2015 1.0 0.9 0.1 9.2
8. Proportion of paddies with acid sulfate soil (%) Paddies at acid sulfate soil (ha)e / total paddies (ha) 1984 5 0 0 44

HHs: households.
a
Based on the commune polygon data provided by AGPPSD.
b
Household data in 2015 (AGPPSD, 2016).
c
Based on the road polyline data provided by AGPPSD.
d
AGPPSD (2016).
e
Based on NIAPP (1990).

166
T. Yamaguchi et al. Agricultural Water Management 212 (2019) 162–171

Fig. 5. AWD adoption rates by commune in the dry seasons of 2009–2010 and 2014–2015. Each parcel is a commune boundary. White parcels represent communes
with missing data.
Source: AGPPSD (2016).

near communes that were convenient in terms of transportation, and

therefore such communes might show an anomalously high adoption
rate. However, the rates of AWD adoption Tan Tuyen and Ba Chuc
communes decreased sharply in 2014–2015 for unknown reasons. Be-
cause the AWD statistics were collected by extension agents without
any fixed process or criteria, there is the possibility that outliers were
generated.
The local government has held farmers’ workshops regularly after
2010 (AGPPSD, 2014), and such continuous efforts would have raised
the overall adoption rate of AWD in 2014–2015. Moreover, it was re-
ported that in An Giang, in addition to water saving, AWD had an effect
on reducing rice lodging (Ha et al., 2014; Lampayan et al., 2015), and
hence farmers often adopted AWD after first-hand observation of well-
grown rice in nearby AWD paddies (Yamaguchi et al., 2016). Such
spillover effects might promote AWD diffusion over the whole of An
Giang. However, differences in the adoption patterns were still pro-
nounced among the districts and the communes in 2014–2015.
Fig. 6. Paddy elevation distribution in An Giang Province.

4.2. Classification of paddies by elevation

Table 3
Spearman’s rank order correlation analysis between AWD adoption rate and
Fig. 6 shows the distribution of paddies by elevation. Approximately
proportion of paddies at each elevation within each commune.
half of the paddies were located at an elevation of > 1–2 m, and 30%
were ≤1 m in elevation. Fewer than 5% of the An Giang paddies were Paddy elevation range (m) Spearman’s rho
located at > 3 m elevation.
2009-2010 2011-2012 2013-2014 2014-2015
In 2009–2010, there was no significant correlation between the
AWD adoption rate and the proportions of paddy fields at any elevation >5 0.01 −0.15 *
−0.21** −0.28***
range (Table 3). In 2011–2012, the proportion of paddies at an eleva- > 4–5 −0.01 −0.11 −0.26*** −0.31***
tion of > 1–2 m was positively correlated with the AWD adoption rate, > 3–4 0.04 −0.07 −0.24*** −0.24***
> 2–3 −0.05 0.08 −0.02 −0.04
but the proportions at the ≤1 m and > 5 m elevations were negatively > 1–2 0.14 0.15* 0.42*** 0.42***
correlated with the AWD adoption rate. In 2013–2014, the proportion ≤1 −0.06 −0.22** −0.26*** −0.28***
of paddies at an elevation of > 1–2 m was positively correlated with the
*
AWD adoption rate and it remained so in 2014–2015. The proportions p < 0.1; ** p < 0.05; *** p < 0.01.
of paddies at the ≤1 m, > 3–4 m, > 4–5 m, and > 5 m elevations were
all negatively correlated with the AWD adoption rate in 2013–2014 and

167
T. Yamaguchi et al. Agricultural Water Management 212 (2019) 162–171

Table 4 Table 5
Estimation of triple-cropping paddy area using MODIS products. Partial rank order correlation (PROCC) analysis between AWD adoption rate
and eight variables by commune.
Time of MODIS data EVI threshold Estimated value Statistics-based R2
acquisition (ha) value (ha) Independent variable 2009-2010 2011-2012 2013-2014 2014-2015
n=135 n=137 n=137 n=136
1–16 Nov. 2009 4,900 76,995 86,586 0.99
1–16 Nov. 2011 1,700 140,485 133,723 0.96 1. Proportion of paddies −0.04 0.03 0.25*** 0.28***
1–16 Nov. 2013 700 183,419 163,181 0.95 located at elevations
1–16 Nov. 2014 2,500 155,383 157,231 0.95 suitable for AWD
2. Channel density 0.01 −0.11 0.11 0.15*
Note: The EVI threshold denotes the EVI separating planted paddies from non- 3. Proportion of triple- 0.14 −0.04 0.14 0.25***
planted paddies in the mid-rainy season. We calculated the cumulative con- cropping paddies
tribution ratio (R2) between the estimated and statistics-based EVIs for triple- 4. Communes in the 0.41*** 0.26*** 0.32*** 0.33***
cropping paddies by district. provincial capital
5. Population density −0.04 0.25*** 0.11 0.21**
6. Road density 0.01 0.03 −0.02 −0.09
even more so in 2014–2015. 7. Farm management scale 0.04 −0.01 0.08 0.02
Diangkinay-Quicho (2013) conducted a household survey and re- 8. Proportion of paddies on −0.08 0.02 0.04 0.04
ported that paddy elevation was not a significant factor affecting AWD acid sulfate soil
diffusion in An Giang; however, their study was conducted in 2009 and *
p < 0.1, **p < 0.05, ***p < 0.01.
thus is not inconsistent with the results of the current study. According
Note: Some communes were excluded from the analysis due to a lack of data, so
to interviews with AGPPSD extension agents (Yamaguchi et al., 2017), the number of communes analyzed varied from year to year.
lower paddies are not suitable for AWD due to drainage problems, but
higher paddies are also not suitable for AWD because those paddies network in AWD adoption (Yamaguchi et al., 2017), and channel
often have difficulties in irrigating when channel water levels are low. density was significantly positively correlated with the AWD adoption
A similar situation exists in the case of the Bengal Delta where it has rate in 2014–2015. Although this analysis only included major channels
been reported that lower paddies were not suitable for AWD (Kurschner constructed before 2003, this network was still having a significant
et al., 2010). In the case of An Giang, AWD implementation would be impact on the AWD adoption rate more than a decade later.
influenced by the vertical location of each paddy, with the elevation The proportion of triple-cropping paddies was more strongly posi-
of > 1–2 m being the most suitable for AWD implementation. tively correlated with the AWD adoption rate than was channel density.
The full-dike systems were originally constructed for mid-rainy season
4.3. Estimation of triple-cropping paddy area by commune rice farming and triple-cropping (Tran et al., 2018), but according to
AGPPSD personnel, the construction of full-dike systems also promoted
For the 2009 mid-rainy season, the triple-cropping paddy area with development of channel networks in order to provide effective drainage
the highest coefficient of determination had an estimated EVI threshold of rainwater from within the full-dike systems. Therefore, farmers
of 4900 (Table 4). We determined suitable EVI thresholds for 2011, cultivating rice in full-dike systems could easily adopt AWD.
2013, and 2014 based on the R2 for each year. The R2 values ranged There were consistently high positive correlations between the AWD
from 0.95 to 0.99, so the estimated triple-cropping paddy areas were adoption rate and communes within the provincial capital from 2009 to
highly correlated with the statistics-based values for each district. 2010 to 2014 to 2015. The relatively higher AWD adoption rate in the
Whereas Sakamoto et al. (2009) and Son et al. (2014) extracted paddy communes of Long Xuyen may be explained the area’s longer history of
areas based on only EVI-derived phenology and then classified the AWD promotion and dissemination (Meier, 2007).
pixels within these polygons into double- or triple-cropping, our clas- There were significant positive correlations between AWD adoption
sification was based on the precise paddy polygons extracted from the rates and population densities in 2011–2012 and 2014–2015. The
detailed map (1:50,000) issued by MNREV (2003); therefore, the ac- availability of household capital influences agricultural mechanization
curacy of our estimations is expected to be better than those of the (Mottaleb et al., 2016), and people living in urban areas of An Giang
previous studies. However, the accuracy of classification at the com- have more opportunities for off-farm employment and have higher in-
mune level still needs to be verified, which is a task for the future. comes than those in rural areas (SOAG, 2015). More farmers may be
The estimated and statistics-based values were well aligned in 2014 able to purchase power pumps for rice farming using their off-farm
but varied in the other years. In 2011 and 2013, the triple-cropping income in urban areas, thereby promoting AWD adoption.
paddy area was overestimated, whereas it was underestimated in 2009. Even though road network development is an enabling factor in
The influence of these discrepancies would be mitigated by using agricultural mechanization (Mottaleb et al., 2016), no significant cor-
nonparametric statistics in the PROCC analysis (Xiao et al., 2016). relation was detected between road density and AWD adoption rate.
There was no significant correlation between AWD adoption and
4.4. PROCC analysis of the AWD adoption rate area of paddy fields cultivated per household. Farmers managing large
paddy areas in An Giang generally use agricultural machinery
The PROCC analysis (Table 5) indicated that the proportion of (Tsukada, 2013), and the rate of power pump ownership is most likely
paddies at a suitable elevation (1–2 m) was positively correlated with higher among large-scale farmers than among small-scale farmers.
AWD adoption rates in 2013–2014 and 2014–2015 even after control- However, AGPPSD reported that paddy surface evenness is a factor
ling for confounding variables, but no such correlation was observed for determining AWD implementation by farmers (Yamaguchi et al., 2017),
2009–2010 or 2011–2012. In the Bengal Delta too, it was reported that and paddy surface irregularity generally increases with plot size (Jat
a 1 m difference in elevation made a difference in the suitability for et al., 2006). Therefore, the relationship between paddy cultivation
AWD implementation (Kurschner et al., 2010). Since delta regions are area per household and AWD adoption appears to be complicated by
flat, even a slight variation in elevation can influence paddy field hy- several other factors.
drology (Thuy and Anh, 2015), making a difference in AWD adoption There was no significant correlation between the AWD adoption
patterns. rate and the distribution of acid sulfate soil. Acid sulfate soil is widely
Takahashi (2013) reported that effective connections between distributed in the Mekong Delta and is detrimental to rice farming.
paddies and water channels were significant factors in AWD adoption. Moreover, aeration of acidic soil increases its deleterious effects on rice
Extension agents also mentioned the importance of the channel growth (Husson et al., 2000), thus paddies with acidic soil would not be

168
T. Yamaguchi et al.
conducive to AWD implementation. AGPPSD personnel, however,
mentioned that acidic soil liming was widely implemented by the local
government in about 2000, and the harmful effects of acid sulfate soil
may have been alleviated. In addition, the acidic soil distribution base
map (NIAPP, 1990) used in the PROCC analysis was drawn more than
30 years ago and might not reflect the current acidic soil distribution.
The PROCC analysis of 2009–2010 (Table 5) indicated that only the
provincial capital effect was significantly correlated with AWD adop-
tion. In 2011–2012, however, population density also was significantly
correlated. In 2014–2015, the number of variables significantly influ-
encing the AWD adoption rate had increased to five.
People who adopt an innovation in its early phase of dis-
semination—“innovators” or “early adopters”—tend to use various
media to actively gather information on the innovations, tend to be
highly educated, and tend to earn high incomes (Rogers, 2003). AWD
information was initially offered in the provincial capital (Meier, 2007),
which has lower poverty rates and higher education levels than other
regions in the study area (SOAG, 2015). Furthermore, the elevations of
most paddies (93%) in the provincial capital area are suitable for AWD,
and the channel density is higher (82 m ha−1) than that in other dis-
tricts (66 m ha−1). Therefore, conditions were such that AWD could
readily be accepted in the provincial capital as soon as it was in-
troduced, and in fact, the provincial capital has had a consistently
higher AWD adoption rate than the other districts.
AWD requires precise water management and is not feasible on
paddies with unfavorable irrigation conditions such as high or low
elevations or poor channel connections. Since the number of farmers
who adopted AWD has increased from 2009 to 2010 to 2014–2015,
these unfavorable irrigation conditions probably gradually became
more influential in the AWD adoption rate for each commune.
AGPPSD personnel stated that the amount of annual precipitation
also influenced AWD adoption (Yamaguchi et al., 2017). Fig. 7 shows
the monthly water level of the Mekong River in the 2013–2014 and
2014–2015 dry seasons. Due to heavy rainfall in December 2013, the
water levels were exceptionally high in the 2013–2014 dry season. In
contrast, the scant precipitation between August and November 2014
caused the water level to be unusually low in the 2014–2015 dry season
(MRC, 2015b). In December 2014, when the 2014–2015 dry season
farming started, the water level was about 50 cm lower than it was in
2014, and there was still a 20 cm difference in March when the dry
season ended. Acquiring irrigation water would have been more diffi-
cult in most paddies in the 2014–2015 dry season than it was in the
Fig. 7. Changes in the Mekong River water level during the 2013–2015 dry
seasons.
The Tan Chau and Chau Doc districts are located in the northern part of An
Giang Province.
Source: MRC (2015a).
169
Agricultural Water Management 212 (2019) 162–171
2013–2014 dry season. Therefore, irrigation conditions, such as the
channel density and degree of triple-cropping, would have played a
significant part in the implementation of AWD in 2014–2015.
5. Conclusion and recommendations
Although AWD is a technique for water saving in paddy rice
farming, it has been pointed out that paddy irrigation conditions would
be an essential factor influencing whether farmers make the decision to
adopt AWD or not (Li, 2001; Loeve et al., 2001). In this study, we use
quantitative data collected from An Giang located in the northern part
of the Mekong Delta, Vietnam, to identify significant relationships be-
tween AWD adoption rates and paddy irrigation conditions. Although
on the whole AWD has been smoothly disseminated in An Giang, there
are variations in the diffusion patterns at the district and community
levels. This study aimed to verify the relationships between AWD
adoption rates and paddy irrigation conditions at the commune level by
using ROCC and PROCC analyses with statistics and GIS data.
Our analysis revealed that paddies at higher (> 3 m) and lower
(≤1 m) elevations were not ideal for AWD implementation due to in-
convenient water conditions, and thus the proportion paddies at sui-
table elevation (1–2 m) positively influenced AWD adoption rate.
Moreover, channel density had a positive effect on AWD adoption rate
in 2015, which was a year with little precipitation. There was also
positive relation between the rate of triple-cropping and AWD adoption
rate in 2015. In An Giang, farmers are able to conduct triple-cropping
only under a full-dike systems; construction of full-dike systems pro-
motes channel network development, and therefore triple-cropping
farmers would more readily implement AWD.
These analyses indicated the significance of irrigation conditions in
the implementation of AWD, raising concerns about how widely AWD
can be disseminated. AWD is considered a knowledge-intensive tech-
nique that is possible to introduce at minimal cost to farmers (e.g.
Kurschner et al., 2010). However, efforts to transmit proper informa-
tion to farmers about AWD would not in themselves be sufficient to
ensure wide-area diffusion of AWD, and paddies suitable for AWD
should be carefully considered from the viewpoint of irrigation condi-
tions. Diffusion efforts should be conducted concurrently with con-
struction of irrigation facilities.
Moreover, the significance of the irrigation network in the adoption
of AWD was emphasized in 2015, a year with low precipitation, in-
dicating that in cases where water conditions are very severe, farmers
would prefer continuous flooding to AWD. Farmers in An Giang fear
drought in their paddies where water conditions are poor or in years
with low precipitation (Yamaguchi et al., 2017), and they would fear
cessation of water supply when their paddy soil was unflooded during
AWD implementation. The researchers who developed AWD considered
continuous flooding of paddies to be a waste of water from the view-
point of rice plant physiology, and thus suggested limiting flooding
periods such as in the AWD protocol (e.g. Humphreys et al., 2010).
However, where irrigation conditions are unreliable, farmers tended to
keep water in their paddies as a “reservoir” to hedge against the risk of
water shortages; from that perspective, continuous flooding has sig-
nificant meaning for farmers. When deciding paddies suitable for AWD
implementation, the meaning of continuous flooding in paddies should
be reconsidered from diverse perspectives.
Owing to the overall accuracy of statistics and GIS data in An Giang,
this study adopted nonparametric analysis methods to alleviate nega-
tive influences of accidental errors and outliers; thus, the quantitative
effects of irrigation conditions and other related factors on AWD
adoption is still unknown. Therefore, a parametric analysis needs to be
conducted to scrutinize data quality and to quantify the effects of each
respective factor.
The PROCC analysis showed that population density was also sig-
nificantly correlated with AWD adoption rates in 2011–2012 and
2014–2015, but the causal relationship between the two remains
T. Yamaguchi et al.
unknown. Moreover, the data available for the PROCC analysis was
limited in An Giang due to lack of sufficient statistics data at the
commune level. Hence the overall picture of the AWD diffusion process
in An Giang remains unexplained. It is expected that the whole process
of AWD implementation will be grasped though detailed household
surveys and though the integration of such micro-scale studies with
macro-scale analysis.
Acknowledgments
We thank Drs. Yagi Kazuyuki and Sudo Shigeto for their helpful
comments and useful discussions about this manuscript. We are also
deeply indebted to Dr. Nguyen Thi My Phung and the staff of the Sub-
Department of Plant Protection in An Giang. This study was in part
supported by the Ministry of Agriculture, Forestry, and Fisheries of
Japan through “Technology development for circulatory food produc-
tion systems responsive to climate change (development of a mitigation
option for greenhouse gas emissions from agricultural lands in Asia)”.
This study was also funded by the JIRCAS project “Development of
agricultural technologies based on sustainable management of the en-
vironment and natural resources in developing regions (FY
2011–2015).”
References
Aakkula, J., Jokinen, P., Lankoski, L., 2005. Knowledge-intensive technology in agri-
cultural and rural contexts: relationships with sustainability. Prog. Ind. Ecol. 2,
358–382.
An Giang Plant Protection Sub-Department (AGPPSD), 2014. Số liệu thực hiện AWD
2005-2010 tại An Giang (AWD workshops for farmers during 2005-2013). Internal
document. AGPPSD, An Giang (In Vietnamese).
AGPPSD, 2016. Bảng Thống Kê Áp Dụng Tiết Kiệm Nước Và 1 Phải 5 Giảm (Statistical
Table of AWD and 1 Must 5 Reduction). AGPPSD, An Giang Internal document (In
Vietnamese).
Barker, R., Herdt, R.B., Rose, B., 1985. The rice economy of Asia. Resources for the
Future. Johns Hopkins University Press, Washington. http://books.irri.org/
0915707152_content.pdf.
Chi, T.T.N., Anh, T.T.T., Tuyen, T.Q., Palis, F., Singleton, G., Toan, N.V., 2013.
Implementation of "one must and five reductions" in rice production, in an Giang
Province. Omonrice 19, 237–249.
Chi, T.T.N., Anh, T.T.T., Tuyen, T.Q., Palis, F., Diangkinay-Quicho, E., Singleton, G.,
2015. Impact of farmers’ adoption of “One Must and Five Reductions” on rice pro-
duction and water use in An Giang Province, Vietnam. Omonrice 20, 123–137.
Diangkinay-Quicho, E., 2013. Evaluation of the Adoption and Economic Impacts of
Alternate Wetting and Drying Technology in Irrigated Rice-growing Areas in An
Giang Province in the Mekong Delta, Southern Vietnam. Master thesis. Graduate
School of University of the Philippines, Los Baños.
Didan, K., Munoz, A.B., Solano, R., Huete, A., 2015. . MODIS Vegetation Index User’s
Guide (MOD13 Series); Version 3.0. University of Arizona, Tucson.
Driel, W.V., Nauta, T.A., 2013. Vulnerability and Resilience Assessment of the
Ayeyarwady Delta in Myanmar, Scoping Phase. Project report. Global Water
Partnership (GWP) and Delta Alliance, Wageningen.
FAO, 2011. The State of the World’s Land and Water Resources for Food and Agriculture–
Managing Systems at Risk. FAO, Rome and Earthscan, London.
Ha, T.T., Sanh, N.V., Rudek, J., Tin, H.Q., Tin, N.H., Tinh, T.K., Cui, T.Q., Pha, D.N., Kien,
H.T., Thanh, H.H., Ahuja, R., 2014. Summary report of the Vietnam Low Carbon Rice
Project (VLCRP), Primary achievement and results after 11 crop production in An
Giang And Kien Giang provinces, period of November 2012–December 2014. In:
VLCRP (Ed.), Dissemination and Regional Policy Dialogue Workshop on Low
Emissions and Sustainable Rice Cultivation. VLCRP, pp. 83–95. (accessed 10 January
2018. http://blogs.edf.org/climatetalks/files/2016/04/VLCRP-Summary-Report-
Final-English.pdf.
Hieu, T.V., 2010. Understanding Farmer Production Strategies in Context of Policies for
Adaptation to Floods in Vietnam: Case Study at Two Communes, an Giang Province,
Vietnam. Master Thesis. Faculty of Natural Resources and Agriculture Sciences of
Swedish University of Agricultural Sciences (Accessed 10 January 2018. http://stud.
epsilon.slu.se.
Humphreys, E., Kukal, S.S., Christen, E.W., Hira, G.S., Singh, Balwinder, Yadav, Sudhir,
Sharma, R.K., 2010. Halting the groundwater decline in north-west India—which
crop technologies will be Winners? Adv. Agron. 109, 155–217.
Hundertmark, W., Facon, T., 2003. Options for effective rice water management. Tran,
D.V., Duffy, R. (Eds.), Proceedings of the 20th Session of the International Rice
Commission 145–170.
Husson, O., Hanhart, K., Phung, M.T., Bouma, J., 2000. Water management for rice
cultivation on acid sulphate soils in the Plain of Reeds, Vietnam. Agric. Water
Manage. 46 (1), 91–109.
IBRD, World Bank, 2010. . Climate Risks and Adaptation in Asian Coastal Megacities.
Synthesis Report. World Bank, Washington.
170
Agricultural Water Management 212 (2019) 162–171
Jat, M.L., Chandna, P., Gupta, R., Sharma, S.K., Gill, M.A., 2006. Laser Land Leveling: A
Precursor Technology for Resource Conservation. Rice-Wheat Consortium Technical
Bulletin Series 7, Rice-Wheat. Consortium for the Indo-Gangetic Plains, New Delhi.
Kaida, Y., 1974. Hydrography of rice land in the Vietnamese part of the Mekong Delta.
Southeast Asian Stud. 12 (2), 143–156.
Kim, S., 2015. Package ‘ppcor’. (Accessed 10 January 2018. https://cran.r-project.org/
web/packages/ppcor/ppcor.pdf.
Kono, Y., 2001. Canal development and intensification of rice cultivation in the Mekong
Delta: a case study in Cantho Province, Vietnam. Southeast Asian Stud. 39 (1), 70–85.
Kurschner, E., Henschel, C., Hildebrandt, T., Julich, E., Leineweber, M., Paul, C., 2010.
Water Saving in Rice Production-dissemination, Adoption, and Short-term Impacts of
Alternate Wetting and Drying (AWD) in Bangladesh. SLE Publication Series S241.
Humboldt Universitat, Berlin.
Lampayan, R.M., 2013. Smart Water Technique for Rice. European Initiative for
Agricultural Research for Development (Accessed 10 August 2018). http://agritech.
tnau.ac.in/agriculture/pdf/csa_pdf/Smart_water_technique_for_rice.pdf.
Lampayan, R.M., Roderick, M.R., Singletona, R., Bouman, A.M., 2015. Adoption and
economics of alternate wetting and drying water management for irrigated lowland
rice. Field Crops Res. 170, 95–108.
Li, Y.H., 2001. Research and practice of water-saving irrigation for rice in China. Barker,
R., Loeve, R., Li, Y.H., Tuong, T.P. (Eds.), Water Saving Irrigation for Rice:
Proceedings of an International Workshop. 1–9.
Loeve, R., Dong, B., Zhoa, J.H., Zhang, S.J., Molden, D., 2001. Operation of Zhanghe
irrigation system. Barker, R., Loeve, R., Li, Y.H., Tuong, T.P. (Eds.), Water Saving
Irrigation for Rice: Proceedings of an International Workshop. 25–53.
MacDonald, A.M., Bonsor, H.C., Taylor, R., Shamsudduha, M., Burgess, W.G., Ahmed,
K.M., Mukherjee, A., Zahid, A., Lapworth, D., Gopal, K., Rao, M.S., Moench, M.,
Bricker, S.H., Yadav, S.K., Satyal, Y., Smith, L., Dixit, A., Bell, R., van Steenbergen, F.,
Basharat, M., Gohar, M.S., Tucker, J., Calow, R.C., Maurice, L., 2015. Groundwater
Resources in the Indo-gangetic Basin: Resilience to Climate Change and Abstraction.
British Geological Survey Open Report. Natural Environment Research Council,
Swindon.
Ministry of Agriculture, Forestry and Fishery, Japan (MAFFJ), 2011. Present situation of
agriculture and agrarian export policy in Vietnam. In: MAFF (Ed.), Research Analysis
on Foreign Agriculture in 2010: International Mutual Understanding Program.
Research Analysis Report of Foreign Agriculture Information (Asia). MAFF, Tokyo,
pp. 163–207 (in Japanese).
Meier, S., 2007. Water Saving Is More Than Saving Water: Farmers’ Adoption of
“Alternate Wetting and Drying” in a High Intensive Rice Production System: the Case
Study in an Giang Province, S.R. Viet Nam. Internal Report. AGPPSD, An Giang.
Ministry of Natural Resources and Environment (MNREV), 2003. Map of Vietnam.
Vietnam Publishing House of Natural Resources, Environment and Cartography,
Hanoi.
Mottaleb, K.A., Krupnik, T.J., Erenstein, O., 2016. Factors associated with small-scale
agricultural machinery adoption in Bangladesh: census findings. J. Rural Sci. 46,
155–168.
Mekong River Commission (MRC), 2015a. River Monitoring: Daily Observation and
Forecast. MRC (Accessed 10 January 2018. http://portal.mrcmekong.org/river-
monitoring.
Mekong River Commission (MRC), 2015b. Mekong River Commission 2014 Annual
Report. MRC (Accessed 10 January 2018). http://www.mrcmekong.org/assets/
Publications/governance/MRC-Annual-Report-2014.pdf.
Mushtaq, S., Dawe, D., Lin, H., Moya, P., 2006. An assessment of the role of ponds in the
adoption of water-saving irrigation practices in the Zhanghe Irrigation System, China.
Agric. Water Manage. 83 (1-2), 100–110.
National Institute of Agricultural Planning and Projection (NIAPP), 1990. Soil Map of the
Mekong Delta. NIAPP, Hanoi.
Otsuka, N., 2014. The development of rice seeds production in an Giang Province,
Mekong Delta. J. Int. Relat. Asia Univ. 23, 53–75 (in Japanese with English sum-
mary).
Pingali, P.L., Hossain, M., Pandey, S., Price, L.L., 1998. Economics of nutrient manage-
ment in Asian rice systems: towards increasing knowledge intensity. Field Crops Res.
56, 157–176.
Quynh, V.D., Sander, B.O., 2015. Applying and scaling up alternate wetting and drying
technology for paddy rice in the Mekong River Delta. CCAFS Workshop Report,
Climate Change, Agriculture and Food Security (CCAFS). (Accessed 10 January
2018. https://cgspace.cgiar.org/handle/10568/67068.
Rahman, M., Bhattacharya, A.K., 2014. Saline water intrusion in coastal aquifers: a case
study from Bangladesh. IOSR J. Eng. 4 (1), 7–13.
Rejesus, R.M., Martin, A.M., Gypmantasiri, P., 2013. Meta-impact Assessment of the
Irrigated Rice Research Consortium. Special IRRI Report. IRRI, Los Banos.
Rogers, E.M., 2003. Diffusion of Innovation, 5th edition. Free Press, New York.
Sakamoto, T., Cao, P.V., Kotera, A., Nguyen, K.D., Yokozawa, M., 2009. Detection of
yearly change in farming systems in the vietnamese Mekong Delta from MODIS time-
series imagery. Japan Agric. Res. Quart. 43 (3), 173–185.
Sheskin, D., 2000. Handbook of Parametric and Nonparametric Statistical Procedures,
Second Edition. Chapman and Hall/CRC, London.
Sibayan, E.B., de Dios, J.L., Florague, M.L., Javier, L.C., Espiritu, A.S., Lampayan, R.M.,
Nangel, A.S., 2010. Controlled irrigation adoption for efficient water management at
the system level for increasing and sustaining water productivity. In: Palis, F.G.,
Singleton, G.R., Casimero, M.C., Hardy, B. (Eds.), Research to Impact: Case Studies
for Natural Resource Management for Irrigated Rice in Asia. IRRI, Los Baños, pp.
135–150.
Statistics Office of An Giang Province (SOAG), 2012. 2011 Statistical Yearbook- An Giang
Province. An Giang Province service of Information and Communications, An Giang.
SOAG, 2015. 2014 Statistical Yearbook- An Giang Province. An Giang Province service of
T. Yamaguchi et al.
Information and Communications, An Giang.
Son, N.T., Chen, C.F., Chen, C.R., Duc, H.N., Chang, L.Y., 2014. A phenology-based
classification of time-series MODIS data for rice crop monitoring in Mekong Delta,
Vietnam. Remote Sens. 6, 135–156.
Takahashi, K., 2013. The roles of risk and ambiguity in the adoption of the system of rice
intensification (SRI): evidence from Indonesia. Food Secur. 5, 513–524.
Thilakarathne, M., Sridhar, V., 2017. Characterization of future drought conditions in the
Lower Mekong River Basin. Weather Clim. Extrem. 17, 47–58.
Thuy, N.N., Anh, H.H., 2015. Vulnerability of rice production in Mekong River Delta
under impacts from floods, salinity and climate change. Int. J. Adv. Sci. Eng. Inf.
Technol. 5 (4), 272–279.
Tran, D.D., van Halsema, G., Hellegers, P., Hoang, L.P., Tran, T.Q., Kummu, M., Ludwig,
F., 2018. Assessing impacts of dike construction on the flood dynamics of the Mekong
Delta. Hydrol. Earth Syst. Sci. 22, 1875–1896.
Tsukada, K., 2013. Mechanicalization of rice farming in the Mekong Delta. In: Sakata, S.
(Ed.), Agriculture and Rural Development in the Era of Rapid Economic Growth in
Vietnam. Institute of Developing Economies, Japan External Trade Organization,
Tokyo, pp. 59–88 (in Japanese).
Tuong, T.P., Bouman, B.A.M., 2003. Rice production in water-scarce environment. In:
Kijne, J.W., Barker, R., Molden, D. (Eds.), Water Productivity in Agriculture: Limits
and Opportunities for Improvement. IRRI, Colombo, pp. 53–67.
171
Agricultural Water Management 212 (2019) 162–171
Van der Hoek, W., Sakthivadivel, R., Renshaw, M., Silver, J.B., Birley, M.H., Konradsen,
F., 2001. Alternate wet/dry Irrigation in Rice Cultivation: a Practical Way to Save
Water and Control Malaria and Japanese Encephalitis? Research Report 47.
International Water Management Institute, Colombo.
Wassmann, R., Hien, N.X., Hoanh, C.T., Tuong, T.P., 2004. Sea level rise affecting the
Vietnamese Mekong Delta: water elevation in the flood season and implications for
rice production. Clim. Change 66, 89–107.
World Bank, 2011. Vietnam urbanization review. Technical Assistance Report. World
Bank, Washington.
Xiao, C., Ye, J., Esteves, R.M., Rong, C., 2016. Using Spearman’s correlation coefficients
for exploratory data analysis on big dataset. Concurr. Comput. Pract. Exp. 28 (14),
3866–3878.
Xuan, V.T., 1975. Rice cultivation in the Mekong Delta: resent situation and potentials for
increased production. South East Asian Stud. 13 (1), 88–111.
Yamaguchi, T., Tuan, L.M., Minamikawa, K., Yokoyama, S., 2016. Alternate Wetting and
Drying (AWD) irrigation technology uptake in rice paddies of the Mekong Delta,
Vietnam. Asian Afr. Area Stud. 15 (2), 1–23.
Yamaguchi, T., Tuan, L.M., Minamikawa, K., Yokoyama, S., 2017. Compatibility of
Alternate Wetting and Drying irrigation with local agriculture in An Giang province,
Mekong Delta, Vietnam. Trop. Agric. Dev. 31 (3), 117–127.