Agricultural and Forest Meteorology 239 (2017) 108–117

Contents lists available at ScienceDirect

Agricultural and Forest Meteorology

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

Research paper

Estimating maize yield potential and yield gap with agro-climatic

zones in China—Distinguish irrigated and rainfed conditions
Baohua Liu a , Xinping Chen a,∗ , Qingfeng Meng b,∗ , Haishun Yang c , Justin van Wart c
a
Center for Resources, Environment and Food Security, China Agricultural University, Beijing 100193, China
b
College of Agronomy and Biotechnology, China Agricultural University, Beijing 100193, China
c
Department of Agronomy and Horticulture, University of Nebraska-Lincoln, Lincoln, NE 68583-0915, USA

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

Article history: Understanding yield potential (Yp) and yield gap (Yg) in current intensive maize (Zea mays L.) production
Received 21 September 2016 is essential to meet future food demand with the limited resources. In this study, we used the agro-climatic
Received in revised form 14 February 2017 zones (CZs) and the reference weather stations (RWS) buffer zones, together with the Hybird-Maize
Accepted 26 February 2017
model to estimate maize Yp in the four maize-growing-regions of China under both irrigated and rainfed
conditions. In irrigated maize areas, we got 70 RWS buffer zones, and total maize area in the RWS buffer
Keywords:
zones covered 67% of the whole irrigated maize area. In rainfed maize areas, we got 106 RWS buffer
Yield potential
zones, which covered 51% of the whole rainfed maize area. As a result, the average Yp was 14.2 t ha−1
Yield gap
Irrigated
and farmers have achieved 58% of Yp. The average water-limited yield potential (Yw) was 10.7 t ha−1
Rianfed and farmers have achieved 65% of Yw. Further analysis for four maize-growing-regions showed that
Agro-climatic zone precipitation was a limiting factor for Yw to fully achieve Yp except in Southwest China (SW), whereas
Buffer zone the average precipitation was more than 653 mm during maize growing season. The ratio between Yw
and Yp (Yw/Yp) was 51% in Northwest China (NW), and around 80% in both Northeast China (NE) and
North China Plain (NCP). The comparison of Yp in different regions showed the low Yp in NE was due
to low temperature while Yp in both NCP and SW were limited by low solar radiation. In conclusion,
our findings highlight the efficiency and importance to estimate Yp, Yw and Yg by the upscaling method
with CZs and RWS buffer zones. Meanwhile, the comparison of Yp, Yw and Yg in different regions was
important to improve maize production in future in China.
© 2017 Elsevier B.V. All rights reserved.

1. Introduction 1995 (Meng et al., 2013). However, high-yielding experiments have

Maize has become the largest cereal food crop in China since
2013, and the maize production was 215 Mt in 2014, which
accounted for more than one-third of China’s cereal production and
was responsible for 21% of the global maize output (FAO, 2016).
With the economic growth and changing diet, demand for maize
in China by 2030 is estimated to be 47% higher than now (Chen
et al., 2014). Until the middle 1990s, China’s maize yield increased
in a near-linear fashion, but has stagnated at around 5.0 t ha−1 since
Abbreviations: Yp, yield potential; Yw, water-limited yield potential; Yg, yield
gap; Ya, actual farmers’ yield; YgI , the difference between Yp and Ya; YgR , the differ-
ence between Yw and Ya; CZs, agro-climatic zones; RWS, reference weather station;
NE, Northeast China; NW, Northwest China; NCP, North China Plain; SW, South-
west China; GYGA-ED, Global Yield Gap Atlas Extrapolation Domain; GDD, growing
degree days.
∗ Corresponding authors.
E-mail addresses: chenxp@cau.edu.cn (X. Chen), mengqf@cau.edu.cn (Q. Meng).
showed that the maize yield was higher than 15 t ha−1 at 159 sites in
China from 2006 to 2010 (Chen et al., 2012). Hence, understanding
the yield potential (Yp) and yield gap (Yg) in the current intensive
maize production is essential to meet the future food demand.
Yield potential is defined as the yield of an adapted crop variety
when grown with optimal water and nutrients management and
without yield losses due to biotic and abiotic stresses (Evans, 1993;
van Ittersum and Rabbinge, 1997; Fischer, 2015). Yg is the differ-
ence between Yp and actual farmers’ yield (Ya) (Lobell et al., 2009).
Estimating Yp and Yg can help assess the status of current farm-
ers’ yield relative to Yp and the possible space for yield gain in the
future (Lobell et al., 2009; Hochman et al., 2013). In addition, the
analysis of spatial distribution of maize yield gap can help reveal
major yield limiting factors and thus effective efforts will be made
to increase yield efficiently (Aggarwal and Kalra, 1994; Naab et al.,
2004; Bhatia et al., 2008; Mueller et al., 2012).
Recently water scarcity has already been a critical issue in
the world (Rijsberman, 2006; McLaughlin and Kinzelbach, 2015).

http://dx.doi.org/10.1016/j.agrformet.2017.02.035
0168-1923/© 2017 Elsevier B.V. All rights reserved.
 B. Liu et al. / Agricultural and Forest Meteorology 239 (2017) 108–117 109

Fig. 1. Selected buffer zones on the harvest areas in the three irrigated maize-growing-regions in China: Northeast China (NE), Northwest China (NW), North China Plain
(NCP). (a) Distribution of weather stations with 30 years of weather data since 1985 and harvested areas (hectare) of irrigated maize, agro-climatic zones (CZs) delineated
based on the GYGA protocol. (b) Buffer zones with a 100 km radius surrounding a weather station. (c) Overlap the buffer zones with the agro-climatic zones in order to ensure
the homogeneity of the agricultural climate in each buffer zones (d) Reference weather station (RWS) buffer zones which selected from the buffer zones according to their
covered maize harvest areas from big to small.

Estimating yield potential separately for irrigated and rainfed con-
ditions is important for evaluating the impact of water on food
production. Crop yield obtained with no other manageable limi-
tation apart from water supply is the water-limited yield potential
(Yw) (Lobell et al., 2009; Fischer, 2015). Irrigation is the essential
measure to increase Yw to fully achieve Yp. However, it is often
difficult to distinguish irrigated crop areas with the rainfed ones in
a large spatial scale. The crop distribution map of the global major
crops harvested area is often used in global studies (Mueller et al.,
2012), but often at the cost of precision. At the region scale, typi-
cal locations are often used to represent both irrigated and rainfed
areas (Grassini et al., 2009; Liu et al., 2016). Therefore, estimat-
ing crop Yp and Yw separately at a high precision raster map is
necessary.
Yield potential for several crops has been estimated in previ-
ous studies in various scales including global, regional and farm.
A grid-based approach is generally used in the global studies with
datasets on climate, soil, agricultural land use and general crop cal-
endars (van Ittersum et al., 2013). The advantage for this global
method is that it provides a framework for upscaling. However,
many details are ignored because of the large scale. For example, it
doesn’t distinguish irrigated and rainfed crops (Licker et al., 2010;
Foley et al., 2011), or explicitly describe the management informa-
tion (cropping systems, planting date, cultivar maturity, planting
density, etc.) (Neumann et al., 2010; Mueller et al., 2012), or ensure
representation of the weather data with model or the yield ceil-
ing with empirical approaches (Foley et al., 2011). In comparison,
regional scale studies have the advantage of location-specific envi-
ronmental conditions and management information, which results
in more locally relevant results. However, the region studies usu-
ally ignore the upscaling process. For example, Grassini et al. (2009)
applied 18 sites to estimate the Yp in Western Corn-Belt of US. Meng
et al. (2013) used 50 sites to estimate maize Yp of the whole China
with all sites being high-yielding fields from the published litera-
tures. The method to select the sites and their representativeness
for a region is worth of further discussion. In order to improve the
representation for a region, appropriate upscaling methods should
be further considered on the region scale studies.
Two questions should be considered when upscaling locations
to a large spatial scale: (1) the homogeneity of the climate, and
(2) acquire observed location-specific data. However, the challenge
of using a bottom-up approach is the time, expense and access to
acquire observed data, e.g., weather data, soil data and crop man-
agement data. The Global Yield Gap Atlas Extrapolation Domain
(GYGA-ED) aims to estimate the yield gap for major food crops
based on locally observed data. The GYGA-ED approach is used in
110 B. Liu et al. / Agricultural and Forest Meteorology 239 (2017) 108–117

Fig. 2. Selected buffer zones on the harvest areas in the four rainfed maize-growing-regions in China: Northeast China (NE), Northwest China (NW), North China Plain (NCP),
Southwest China (SW). (a) Distribution of weather stations with 30 years of weather data since 1985 and harvested areas (hectare) of rainfed maize, agro-climatic zones
(CZs) delineated based on the GYGA protocol. (b) Buffer zones with a 100 km radius surrounding a weather station. (c) Overlap the buffer zones with the agro-climatic zones
in order to ensure the homogeneity of the agricultural climate in each buffer zones. (d) Reference weather station (RWS) buffer zones which selected from the buffer zones
according to their covered maize harvest areas from big to small.

Fig. 3. Irrigated yield potential (Yp, a) and water-limited yield potential (Yw, b) in the four maize-growing-regions in China: Northeast China (NE), Northwest China (NW),
North China Plain (NCP), and Southwest China (SW). n represents the number of the selected reference weather station (RWS) buffer zones. Each point is a 30-year average
for a given location from 1985 to 2014. Solid and dashed lines in the boxes indicate medians and means, respectively. Upper and lower box boundaries indicate upper and
lower quartiles. Whisker caps indicate 95th and 5th percentiles. Circles indicate outliers.
 B. Liu et al. / Agricultural and Forest Meteorology 239 (2017) 108–117
this study, because with this method using of agro-climatic zones
(CZs) and reference weather station (RWS) buffer zones can get a
good balance to minimize the data volume with the goal of maxi-
mum appropriate representativeness of the specific locations. The
CZs is constructed from a matrix of three categorical variables:
(1) GDD with base temperature of 0 ◦ C, (2) temperature season-
ality (quantified as the standard deviation of monthly average
temperatures), and (3) an aridity index (annual total precipitation
divided by annual total potential evapotranspiration) (van Wart
et al., 2013). A buffer zone circle with 100 km radius centered
on a weather station was used in the GYGA-ED approach, which
was considered as the appropriate coverage area (van Bussel et al.,
2015).
The objectives of this study were to (1) use the CZs and RWS
buffer zones to estimate Yp, Yw and Yg in four maize-growing-
regions of China under both irrigated and rainfed conditions, (2)
use the GYGA-ED approach to upscale the results from locations
to the region, and (3) use 30 years weather data including solar
radiation, temperature and precipitation to analyze the variation
for Yp among four maize-growing-regions.
2. Materials and methods
2.1. Hybrid-Maize model
The Hybrid-Maize Model was used to estimate Yp and Yw. It is
a process-oriented model that simulates maize development and
growth on a daily time step under non-limiting or water-limited
conditions (Yang et al., 2006). It features temperature-driven
maize development, vertical canopy integration of photosynthe-
sis, organ-specific growth respiration, and temperature-sensitive
maintenance respiration (Yang et al., 2004). The model has been
tested and widely used in the USA (Grassini et al., 2009; Setiyono
et al., 2011; van Wart et al., 2013) and China (Bai et al., 2010; Chen
et al., 2011; Meng et al., 2013). Model input includes daily weather
data (i.e., solar radiation, maximum and minimum temperatures,
precipitation, wind speed, and relative humidity), hybrid maturity,
sowing date, and planting density. When estimating Yw, major soil
properties were required, including maximum rooting depth, soil
texture, and bulk density.
2.2. Data sources
Data of maize harvested area was obtained from China’s
National Bureau of Statistics (NBS). The data were at the county
level in 2009. We used the SPAM2005 (http://mapspam.info) to
distinguish irrigated and rainfed areas. Daily weather data for
the period from 1985 to 2014 for a total of 756 stations were
obtained from China meteorological Administration (http://data.
cma.cn) (Fig. 1a). Average maize growing degree days (GDD) was
calculated from the weather data for the maize growing season
from 1985 to 2014. Soil texture data, including proportion of
clay and silt, were retrieved from ISRIC-World soil information,
WISE international soil profile dataset (http://www.isric.org). Crop
management information including sowing and maturity date,
and planting density were provided by local agronomic experts
(Table 1).
Ya was obtained from farm surveys conducted during
2007–2008. The farm surveys included face-to-face interviews with
farmers and the questionary was designed to capture relevant data
(e.g., grain yield, planting density, harvest date, fertilizer and irriga-
tion management). For irrigated maize, a total of 1988 sets of data
were obtained with 693 in the NE, 214 in the NW, and 1081 in NCP,
which covered 33 RWS buffer zones in total. For the rainfed maize,
a total of 2049 sets of data were obtained with 736 in NE, 229 in
111
Table 1
Maize management information for the four maize-growing-regions: Northeast
China (NE), Northwest China (NW), North China Plain (NCP), Southwest China (SW).
Planting Date Maturity date Planting density
(1000 ha−1 )
NE Apr.25–May.5 Sep.15–Sep.26 75
NW Apr.20–May.1 Sep.20–Oct.1 75
NCP Jun.10–Jun.15 Sep.25–Oct.1 90
SW Feb.10–Apr.10 Jun.20–Aug.20 60–70
NW, 963 in NCP, and 121 in SW, which covered the 46 RWS buffer
zones in total.
2.3. Upscaling method
RWS buffer zones of irrigated maize in each region were selected
following the protocol described in Van Wart et al. (2013) and van
Bussel et al. (2015). Firstly, the distribution of irrigated maize area
and weather stations were drawn on the China map, and CZs were
superimposed on top of the map (Fig. 1a). Secondly, circles with
100 km radius surrounding all weather stations were drawn as the
buffer zones (Fig. 1b). Thirdly, overlap the buffer zones with the
CZs in order to ensure the homogeneity of the agricultural climate
in each buffer zones (Fig. 1c). Finally, selecting the reference buffer
zones (RWS) in each region according to their covered harvested
maize area from big to small until the total harvested area in buffer
zones became greater than 50% of each region (Fig. 1d). The selected
buffer zones were the RWS buffer zones. The similar process for
rainfed maize was shown in Fig. 2.
In total, we selected 70 RWS buffer zones for the irrigated maize
and 106 for the rainfed maize. China is a large country with a
wide distribution of maize area and great variation of climate. In
order to reduce the variation of climate effect on Yp and Yw, we
grouped the RWS buffer zones to four maize regions according to
the agro-ecological conditions: Northeast China (NE), North China
Plain (NCP), Northwest China (NW), and Southwest China (SW)
(Figs. 1a, 2a). While rainfed maize was grown in all four regions,
irrigated maize was grown in NE, NW, and NCP, but not in SW due
to relative adequate precipitation in this region.
Then we used crop areas to identify dominant soil types in each
RWS. Put the soil type, maize harvested area and RWS in a map.
Calculated the coverage maize harvested area of each soil type in
each RWS. Selected the dominant soil types according to their cov-
ered maize harvested area. The disciplines were: (1) More than 50%
coverage of maize harvested area in each RWS; (2) Selected the soil
type if maize harvested area in it was more than 10%; (3) No more
than three soil types were selected in each RWS.
For each irrigated RWS buffer zone, we run the Hybrid-Maize
model to estimate its Yp using 30 years weather data and maize
management data. For rainfed maize, Yw was simulated for each of
its dominant soil types, and then averaged using the harvest area
as the weight of each soil type. We then aggregated Yp or Yw in the
RWS buffers to the CZ level using the maize harvested area in each
RWS as its weight in the aggregation. Finally, we aggregated Yp or
Yw to the region level using the harvest area in each CZ as its weight
in the aggregation. Similarly, Ya was aggregated from each RWS to
the region level. For each region, the yield gap between Yp and Ya
(YgI ), and the yield gap between Yw and Ya (YgR ) were calculated.
3. Results
3.1. Selected RWS buffer zones and crop area coverage
The total maize harvested area in China was 27.15 Mha, with
10.26 Mha under irrigated conditions and 16.89 Mha under rainfed
112 B. Liu et al. / Agricultural and Forest Meteorology 239 (2017) 108–117

Fig. 4. Growing season climate date for irrigated maize in the three regions: Northeast China (NE), Northwest China (NW), North China Plain (NCP). (a) Minimum temperature.
(b) Maximum temperature. (c) Mean temperature, (d) Cumulative precipitation, (e) cumulative solar radiation. (f) Growing degree days (GDD). n represent the number of the
selected reference weather station (RWS) buffer zones. Each point is a 30-year average for a given location from 1985 to 2014. Solid and dashed lines in the boxes indicate
medians and means, respectively. Upper and lower box boundaries indicate upper and lower quartiles, while whisker caps indicate 95th and 5th percentiles, and circles
indicate outliers.

Table 2
The number of reference weather station (RWS) buffer zones maize areas in the entire China and the distribution in the four maize-growing-regions: Northeast China (NE),
Northwest China (NW), North China Plain (NCP), Southwest China (SW).

Total areas (106 ha) RWS Buffer zones number Areas in selected buffer zones (106 ha) Ratio (%)

Irrigated NE 2.47 23 1.61 65

NW 1.46 21 0.82 56
NCP 6.33 26 4.48 71

Rainfed NE 7.02 26 3.64 52

NW 2.05 31 1.05 51
NCP 3.57 23 1.85 52
SW 4.25 26 2.12 50
Sum 27.15 176 15.67 57

Ratio: the ratio between the area in the selected buffer zones of a region and the total area in the region.

conditions (Table 2). In total, we selected 70 RWS buffer zones for
irrigated maize, with 23 in NE, 21 in NW, and 26 in NCP. For irrigated
maize, the area covered by the RWS buffer zones accounted for 65%,
56% and 71% of the total harvest irrigated maize area in NE, NW,
and NCP, respectively. For the rainfed maize, we selected 106 RWS
buffer zones, with 26 in NE, 31 in NW, 23 in NCP, and 26 in SW. The
coverage of the harvested area by the RWS buffer zones accounted
for 52%, 51%, 52% and 57% of the total rainfed maize area in NE, NW,
NCP and SW, respectively.
3.2. Yield potential and yield gap
For the whole China, the average Yp and Yw was 14.2 t ha−1 and
10.7 t ha−1 , respectively (Table 3). The average Ya under irrigated
and rainfed conditions was 8.2 t ha−1 and 7.0 t ha−1 , respectively.
Therefore, the YgI was 6.0 t ha−1 for irrigated maize and YgR was
3.7 t ha−1 for rainfed maize. Farmers achieved 58% of Yp under irri-
gated conditions and 65% to Yw under rainfed conditions.
Among the three regions of irrigated maize, NW had the highest
average Yp (16.8 t ha−1 ) with the largest variation while NE and NCP
had a lower Yp along with smaller variation (Fig. 3a). The average Ya
in the three regions were similar and all were around 8.0–8.5 t ha−1 .
As a result, the highest average YgI was found in NW (8.4 t ha−1 ) and
Ya only achieved 50% of Yp. The average YgI in NE and NCP were
around 5–6 t ha−1 , and Ya achieved around 60% of Yp (Table 3).
Among the four regions of rainfed maize, the average Yw in NE
and NCP was at a similar level (11.7 t ha−1 and 11.1 t ha−1 ), which
was far above those in both NW and SW (Fig. 3b). For Ya, SW had
the lowest value of 5.1 t ha−1 while Ya in NE, NW, NCP was around
7–8 t ha−1 . The average YgR in NW was much lower than the other
three regions, and the Ya achieved 85% Yw (Table 3). In contrast,
the average YgR was highest in SW and Ya only achieved 53% Yw.
 B. Liu et al. / Agricultural and Forest Meteorology 239 (2017) 108–117 113

Fig. 5. Growing season climate data for rainfed maize in the four regions: Northeast China (NE), Northwest China (NW), North China Plain (NCP), and Southwest China
(SW). (a) Minimum temperature. (b) Maximum temperature. (c) Mean temperature, (d) Cumulative precipitation, (e) cumulative solar radiation. (f) Growing degree days
(GDD). n represent the number of selected reference weather station (RWS) buffer zones. Each point is the 30-year average for a given location from 1985 to 2014. Solid and
dashed lines in the boxes indicate medians and means, respectively. Upper and lower box boundaries indicate upper and lower quartiles, whisker caps indicate 95th and 5th
percentiles, and circles indicate outliers.

Table 3
Irrigated yield potential (Yp), water-limited yield potential (Yw), actual farmers’ yield (Ya), yield gap between Yp and Ya (YgI ), yield gap between Yw and Ya (YgR ), and the
Ya relative to Yp and Yw (Ya/Yp, Ya/YW ) in the whole China and the four maize-growing-regions: Northeast China (NE), Northwest China (NW), North China Plain (NCP),
Southwest China (SW).

NE NW NCP SW China
−1
Irrigated Yp (t ha ) 14.4 16.8 13.3 – 14.2
Ya (t ha−1 ) 8.5 8.4 8.1 – 8.2
YgI (t ha−1 ) 5.9 8.4 5.2 – 6.0
Ya/Yp (%) 59 50 61 – 58

Rainfed Yw (t ha−1 ) 11.7 8.5 11.1 9.6 10.7

Ya (t ha−1 ) 7.6 7.2 7.7 5.1 7.0
YgR (t ha−1 ) 4.1 1.3 3.4 4.5 3.7
Ya/Yw (%) 65 85 69 53 65

3.3. Analysis of climatic factors in different regions
For irrigated maize, growing season (i.e., from planting to phys-
iological maturity) temperature was low in NE with an average
minimum temperature (Tmin), maximum temperature (Tmax),
mean temperature (Tmean) of 14.3 ◦ C, 25.3 ◦ C, and 19.8 ◦ C, respec-
tively (Fig. 4). As a result, total seasonal available GDD was also low
in NE with an average of 1514. However, NE had relatively high
solar radiation 2658 MJ m−2 . NW also has a high solar radiation
(2860 MJ m−2 ). In comparison, solar radiation in NCP was relatively
low (1779 MJ m−2 ).
For rainfed maize across whole China, the precipitation during
maize season was higher than for the irrigated maize, while GDD,
temperature and solar radiation was similar in NE, NW and NCP
(Fig. 5). In addition, SW had plentiful precipitation and high temper-
ature. The average precipitation was as high as 653 mm, and Tmin,
Tmax, Tmean were averaged at 18.4 ◦ C, 26.8 ◦ C, 22.6 ◦ C, respec-
tively. Solar radiation in SW was relative low with 1853 MJ m−2 .
3.4. Relationship between climatic variable and yield potential
The relationship between cumulative precipitation during
maize growing season and Yp, and Yw was compared in each region
(Fig. 6). Among four regions, similar Yp and Yw was found only in
SW, which indicated water was not a limited factor for maize pro-
duction in this region (Fig. 6d). In NW (Fig. 6b), the total average
precipitation was 360 mm, and the difference between Yp and Yw
was 7.2 t ha−1 . In both NE (Fig. 6a) and NCP (Fig. 6c), the total aver-
age precipitation was around 450 mm and the difference between
Yp and Yw was 2.4 t ha−1 in NE and 2.2 t ha−1 in NCP.
In each region, Yp increased with the increase of total solar radi-
ation during maize season in four regions (Fig. 7a). Yp in NCP and
SW were relative low together with the low solar radiation. In com-
parison, Yp in NE and NW was higher with the high solar radiation.
114 B. Liu et al. / Agricultural and Forest Meteorology 239 (2017) 108–117

Fig. 6. The relationship between cumulative precipitation during maize growing season and yield potential in the four maize-growing-regions: Northeast China (NE, a),
Northwest China (NW, b), North China Plain (NCP, c), and Southwest China (SW, d). The green circles represent the irrigated yield potential while the red crosses represent
the rainfed yield potential. Each point is the 30-year average for a given location from 1985 to 2014. The black cross-shaped indicate the standard deviation. Asterisks indicate
correlation at *p < 0.05, **p < 0.01, and ***p < 0.001. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

The effect of mean temperature on Yp was different with solar radi-
ation (Fig. 7b). In NE and SW, the average mean temperature were
19.8 ◦ C and 22.6 ◦ C, and the Yp increased with the mean temper-
ature. However, there was no significant relationship between Yp
and mean temperature in NW and NCP, where the average mean
temperature was 21.4 ◦ C and 25.3 ◦ C, respectively.
4. Discussion
For the whole China, our study showed that Yp and Yw aver-
aged at 14.2 t ha−1 and 10.7 t ha−1 , YgI was 6.0 t ha−1 and YgR was
3.7 t ha−1 for irrigated and rainfed maize, respectively. Farmers
achieved 58% and 65% of Yp and Yw, respectively (Table 3). Yield
potential for maize has been estimated in many other countries.
As the biggest maize production country, USA also has the highest
Yp (Fischer et al., 2014; FAO, 2016). Van Wart et al. (2013) showed
that the Yp and Yw was 15.1 t ha−1 and 13.2 t ha−1 for USA maize
production, respectively. The high Yp in USA was mainly a benefit
from the well-irrigated management in a highly suitable temper-
ate region environment (Fischer et al., 2014). Moreover, the YgI
and YgR was lower compared with China, which was 3.4 t ha−1 and
3.5 t ha−1 under irrigated and rainfed conditions, respectively (Van
Wart et al., 2013). The low YgI and YgR in USA were the results
of excellent crop varieties and advanced management technolo-
gies (plant density, nitrogen management, precision planter, etc.)
(Fischer et al., 2014). In Kenya, Tittonell et al. (2008) showed that
the Yw was 5.4 t ha−1 while the YgR was 3.7 t ha−1 , and farmers
only achieved 31% of the Yw. In this area, the severely degraded
soil and drought led to the low Yw (Fischer et al., 2014; Heisey and
Edmeades, 1999). The high YgR in Kenya was mainly due to the
shortage resources, backward agronomic management technology
and smallholder production system (Tittonell et al., 2008). In addi-
tion, Brasil (the third maize production country) and Argentina (the
fourth maize production country) had the similar Yw and YgR as
China. Affholder et al. (2013) and Aramburu Merlos et al. (2015)
showed the Yw was 8.3 t ha−1 in Brail and 11.6 t ha−1 in Argentina
and the YgR was 3.7 t ha−1 and 4.8 t ha−1 , respectively.
 B. Liu et al. / Agricultural and Forest Meteorology 239 (2017) 108–117 115

Fig. 7. Relationship between yield potential (Yp) and cumulative solar radiation (a), and mean temperature (b) during maize growing season in the four maize-growing-
regions: Northeast China (NE, red triangles), Northwest China (NW, blue squares), North China Plain (NCP, yellow rhombuses), Southwest China (SW, green circles). Each point
is the 30-year average for a given location from 1985 to 2014. The black cross-shaped indicate the standard deviation. Asterisks indicate correlation at *p < 0.05, **p < 0.01,
and ***p < 0.001. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

In this study, we analyzed both Yp and Yw with distinguishing
irrigated maize and rainfed maize in the whole China. At the coun-
try level, the Yp was 33% (3.5 t ha−1 ) higher than the Yw (Fig. 3). It
indicated water has already become a big limiting factor for maize
production in China. However, we found varied results in different
regions. For example, there was nearly no restrictions of water for
maize production in SW with an average precipitation of 653 mm
during maize the growing season (Fig. 6d). However, when the
average precipitation dropped to 450 mm in NE and NCP, Yw was
80% of Yp (Fig. 6a, c). While the average precipitation dropped to
360 mm in NW, Yw was just 50% of Yp. Similarly, a recent study
showed that the critical precipitation level was 462 mm to close
the gap between Yw and Yp in the China maize belt (Meng et al.,
2016). In addition, water scarcity was already a critical problem
in China, which was expected to become more sever in the future
(Rijsberman, 2006; Jiang, 2009). China maize production may have
a big challenge under the most severe water scarcity. Meanwhile,
the sensitivity of maize production to water has also increased
under climate change (Meng et al., 2016), which will make the prob-
lem worse. We also found solar radiation and temperature were
also two important factors affecting Yp. Yp in NE was limited by
low temperature while Yp in NCP and SW were limited by the low
solar radiation. Parry et al. (2007) showed that the north-high lati-
tudes have experienced among faster warming trend in the world,
thus maize production in NE would benefit in future. However, the
observed significant decrease in solar radiation in NCP may have a
negative impact on Yp (Meng et al., 2016).
As the largest cereal crop in China, closing maize Yg is essential
to increase grain production and meet future food demand. Firstly,
reducing the 3.5 t ha−1 difference between the Yp and Yw through
the irrigation would be an effective measure. Meanwhile, Bu et al.
(2013) and Liu et al. (2010) showed that plastic film mulching
increased maize yield by 20%–80% because of the effective conser-
vation of available soil water and increase of soil temperatures in
the spring. Meanwhile, YgR between the Yw and Ya was still large,
especially in SW (4.5 t ha−1 ). This is because agricultural produc-
tion is dominated by smallholder farmers in SW and farm size is
an important factor to improve grain yield. In comparison, large-
scale farm is common in NE which is benefit to its high Ya and
low YgR . In additional, the hot and humid climate conditions in
SW increased the difficulty to control the biotic and abiotic stress
in maize production. It is also difficult for management under this
climate such as nutrient management with many rainy days. Gener-
ally, further yield boost would come from optimized management
choices, including variety, planting date, seeding rate, and fertil-
izer management (Chen et al., 2011). For example, the integrated
soil-crop system management was designed to make maximum
use of solar radiation and temperatures, and synchronized nutri-
ent supply and crop demand (Chen et al., 2011, 2014). Field results
showed the integrated soil-crop system management can increase
wheat grain yield by 35% compared with farmers’ practices (Chen
et al., 2014).
In this study, we used the GYGA-ED protocol to estimate Yp and
Yw for China maize production. The CZs scheme delineated har-
vest areas based on the growing degree days, aridity index, and
temperature seasonality seems to strive a good balance between
the zone size and the number of zones, and keeping a better cli-
matic homogeneity within the zones. The selection of the buffer
zones according to their clipped harvested crop areas from big to
small can ensure the representative of the RWS, while the mini-
mum 50% harvested area coverage of buffer zones to the region
can decrease the demand for data by a large margin. When the cov-
erage ratio reached 50% of region, the weighted averaged Yp tended
to be stabilized (Fig. 8). Similar result was also found in USA maize
production (van Bussel et al., 2015). In comparison, the estimation
of Yp and Yw in the past studies mainly relied on the gridded data or
even had no upscaling process. For example, Mueller et al. (2012)
and Neumann et al. (2010) estimated the Yp and Yw at a 5 arc-
minute by 5 arc-minute resolution at the global scale. Collecting
the weather data, agronomic data and soil data is time consuming
and expensive. Meanwhile, the quality of the gridded-interpolated
average weather data, the accuracy of the crop management data
and soil data is an issue on a large scale. In addition, the non-uniform
geospatial distribution of crop area within a grid was ignored. Some
other studies based on locally observed data have the advantage,
but often at the cost of spatial scale. Grassini et al. (2009) used 18
sites to estimate the maize Yp in Western Corn-Belt of USA. Lu and
Fan (2012) used 43 sites to estimate the wheat Yp in NCP. Liu et al.
(2016) used 55 sites to estimate the maize Yp in NE. In summary, the
bottom-up GYGA-ED protocol in this study obtains a good balance
116 B. Liu et al. / Agricultural and Forest Meteorology 239 (2017) 108–117
Fig. 8. Maize yield potential change with the coverage ratio of reference weather
stations buffer zones. The coverage ratio = area of reference weather stations buffer
zones/total maize area*100.
between RWS buffer zones and CZs to minimize the data volume
of weather, soils, and crop management with the goal of maximum
representativeness of the specific locations.
5. Conclusions
In this study, we used GYGA-ED protocol method together with
model simulations to estimate the maize yield potential and yield
gap in the four maize-growing-regions of China under both irri-
gated and rainfed conditions. We found Yp and Yw averaged at
14.2 t ha−1 and 10.7 t ha−1 , respectively. The corresponding YgI and
YgR was 6.0 t ha−1 and 3.7 t ha−1 , for irrigated and rainfed maize,
respectively. Meanwhile, the analysis of the climate factors showed
that precipitation was the most important limiting factor to maize
production except in SW. Yp was limited by low temperature in
NE, and by low solar radiation in both NCP and SW. Using GYGA-ED
protocol method to estimate the Yp, Yw and Yg when upscal-
ing location to the large spatial region maintains a good balance
between minimum data demand for weather, soils, and crop man-
agement and the goal of maximum appropriate representativeness
of the specific locations.
Acknowledgements
This work was supported by the National Basic Research Pro-
gram (973 Program) of China (No. 2015CB150402, 2015CB150401)
and the national key research and development program of China
(No. 2016YFD0300110).
References
Affholder, F., Poeydebat, C., Corbeels, M., Scopel, E., Tittonell, P., 2013. The yield gap
of major food crops in family agriculture in the tropics: assessment and
analysis through field surveys and modelling. Field Crop Res. 143, 106–118.
Aggarwal, P.K., Kalra, N., 1994. Analyzing the limitations set by climatic factors,
genotype, and water and nitrogen availability on productivity of wheat II.
Climatically potential yields and management strategies. Field Crop Res. 38,
93–103.
Aramburu Merlos, F.A., Monzon, J.P., Mercau, J.L., Taboada, M., Andrade, F.H., Hall,
A.J., Jobbagy, E., Cassman, K.G., Grassini, P., 2015. Potential for crop production
increase in Argentina through closure of existing yield gaps. Field Crop Res.
184, 145–154.
Bai, J., Chen, X., Dobermann, A., Yang, H., Cassman, K.G., Zhang, F., 2010. Evaluation
of NASA satellite-and model-derived weather data for simulation of maize
yield potential in China. Agron. J. 102, 9–16.
Bhatia, V.S., Singh, P., Wani, S.P., Chauhan, G.S., Rao, A.K., Mishra, A.K., Srinivas, K.,
2008. Analysis of potential yields and yield gaps of rainfed soybean in India
using CROPGRO-Soybean model. Agric. For. Meteorol. 148, 1252–1265.
Bu, L.D., Liu, J.L., Zhu, L., Luo, S.S., Chen, X.P., Li, S.Q., Hill, R.L., Zhao, Y., 2013. The
effects of mulching on maize growth, yield and water use in a semi-arid region.
Agric. Water Manage. 123, 71–78.
Chen, X.P., Cui, Z.L., Vitousek, P.M., Cassman, K.G., Matson, P.A., Bai, J.S., Meng, Q.F.,
Hou, P., Yue, S.C., Roemheld r, V., Zhang, F.S., 2011. Integrated soil–crop system
management for food security. Proc. Natl. Acad. Sci. U. S. A. 108, 6399–6404.
Chen, G.P., Gao, J.L., Zhao, M., Dong, S.T., Li, S.K., Yang, Q.F., Liu, Y.H., Wang, L.C.,
Xue, J.Q., Liu, J.G., Li, C.H., Wang, Y.H., Wang, Y.D., Song, H.X., Zhao, J.R., 2012.
Distribution, yield structure, and key cultural techniques of maize super-high
yield plots in recent years. Acta. Agron. Sin. 38, 80–85 (in Chinese with English
abstract).
Chen, X., Cui, Z., Fan, M., Vitousek, P., Zhao, M., Ma, W., Wang, Z., Zhang, W., Yan, X.,
Yang, J., Deng, X., Gao, Q., Zhang, Q., Guo, S., Ren, J., Li, S., Ye, Y., Wang, Z.,
Huang, J., Tang, Q., Sun, Y., Peng, X., Zhang, J., He, M., Zhu, Y., Xue, J., Wang, G.,
Wu, L., An, N., Wu, L.Q., Ma, L., Zhang, W., Zhang, F., 2014. Producing more grain
with lower environmental costs. Nature 514, 486–489.
Evans, L.T., 1993. Crop Evolution, Adaptation and Yield. Cambridge University
Press.
FAO, 2016. FAOSTAT—Agriculture Database, Available at: http://faostat.fao.org/.
Fischer, T., Byerlee, D., Edmeades, G., 2014. Crop Yields and Global Food Security.
Will Yield Increases Continue to Feed the World? Australian Centre for Int.
Agric. Res., Canberra.
Fischer, R.A., 2015. Definitions and determination of crop yield gaps, and of rates of
change. Field Crop Res. 182, 9–18.
Foley, J.A., Ramankutty, N., Brauman, K.A., Cassidy, E.S., Gerber, J.S., Johnston, M.,
Mueller, N.D., O’Connell, C., Ray, D.K., West, P.C., Balzer, C., Bennett, E.M.,
Carpenter, S.R., Hill, J., Monfreda, C., Polasky, S., Rockstrom, J., Sheehan, J.,
Siebert, S., Tilman, D., Zaka, D.P., 2011. Solutions for a cultivated planet. Nature
478, 337–342.
Grassini, P., Yang, H., Cassman, K.G., 2009. Limits to maize productivity in Western
Corn-Belt: a simulation analysis for fully irrigated and rainfed conditions.
Agric. For. Meteorol. 149, 1254–1265.
Heisey, P.W., Edmeades, G.O., 1999. Maize Production in Drought-Stressed
Environments: Technical Options and Research Resource Allocation. Part 1.
CIMMYT 1997/98 World Maize Facts and Trends. Centro Internacional de
Mejoramiento de Maíz y Trigo, Mexico D.F., Mexico.
Hochman, Z., Gobbett, D., Holzworth, D., McClelland, T., van Rees, H., Marinoni, O.,
Garcia, N.J., Horan, H., 2013. Reprint of Quantifying yield gaps in rainfed
cropping systems: a case study of wheat in Australia. Field Crop Res. 143,
65–75.
Jiang, Y., 2009. China’s water scarcity. J. Environ. Manage. 90, 3185–3196.
Licker, R., Johnston, M., Foley, J.A., Barford, C., Kucharik, C.J., Monfreda, C.,
Ramankutty, N., 2010. Mind the gap: how do climate and agricultural
management explain the ‘yield gap’of croplands around the world? Glob. Ecol.
Biogeogr. 19, 769–782.
Liu, Y., Li, S.Q., Chen, F., Yang, S.J., Chen, X.P., 2010. Soil water dynamics and water
use efficiency in spring maize (Zea mays L.) fields subjected to different water
management practices on the Loess Plateau, China. Agric. Water Manage. 97,
769–775.
Liu, Z., Yang, X., Lin, X., Hubbard, K.G., Lv, S., Wang, J., 2016. Maize yield gaps
caused by non-controllable, agronomic, and socioeconomic factors in a
changing climate of Northeast China. Sci. Total Environ. 541, 756–764.
Lobell, D.B., Cassman, K.G., Field, C.B., 2009. Crop yield gaps: their importance,
magnitudes, and causes. Annu. Rev. Environ. Resour. 34, 179.
Lu, C.H., Fan, L., 2012. Winter wheat yield potentials and yield gaps in the North
China Plain. Field Crop Res. 143, 98–105.
McLaughlin, D., Kinzelbach, W., 2015. Food security and sustainable resource
management. Water Resour. Res. 51, 4966–4985.
Meng, Q., Hou, P., Wu, L., Chen, X., Cui, Z., Zhang, F., 2013. Understanding
production potentials and yield gaps in intensive maize production in China.
Field Crop Res. 143, 91–97.
Meng, Q., Chen, X., Lobell, D.B., Cui, Z., Zhang, Y., Yang, H., Zhang, F., 2016. Growing
sensitivity of maize to water scarcity under climate change. Sci. Rep. 6, 19605.
Mueller, N.D., Gerber, J.S., Johnston, M., Ray, D.K., Ramankutty, N., Foley, J.A., 2012.
Closing yield gaps through nutrient and water management. Nature 490,
254–257.
Naab, J.B., Singh, P., Boote, K.J., Jones, J.W., Marfo, K.O., 2004. Using the
CROPGRO-peanut model to quantify yield gaps of peanut in the Guinean
Savanna Zone of Ghana. Agron. J. 96, 1231–1242.
Neumann, K., Verburg, P.H., Stehfest, E., Müller, C., 2010. The yield gap of global
grain production: a spatial analysis. Agric. Syst. 103, 316–326.
Parry, M., Canziani, O., Palutikof, J., van der Linden, P., Hanson, C., IPCC, 2007.
Climate change 2007: impacts, adaptation and vulnerability. Contribution of
working group II to the fourth assessment report of the intergovernmental
panel on climate change. 1–976.
Rijsberman, F.R., 2006. Water scarcity: fact or fiction? Agric. Water Manage. 80,
5–22.
Setiyono, T.D., Yang, H., Walters, D.T., Dobermann, A., Ferguson, R.B., Roberts, D.F.,
Lyon, D.J., Clay, D.E., Cassman, K.G., 2011. Maize-N: a decision tool for nitrogen
management in maize. Agron. J. 103, 1276–1283.
Tittonell, P., Vanlauwe, B., Corbeels, M., Giller, K.E., 2008. Yield gaps, nutrient use
efficiencies and response to fertilisers by maize across heterogeneous
smallholder farms of western Kenya. Plant Soil 313, 19–37.
Van Wart, J., Kersebaum, K.C., Peng, S., Milner, M., Cassman, K.G., 2013. Estimating
crop yield potential at regional to national scales. Field Crops Res. 143, 34–43.
 B. Liu et al. / Agricultural and Forest Meteorology 239 (2017) 108–117
van Bussel, L.G., Grassini, P., Van Wart, J., Wolf, J., Claessens, L., Yang, H., Boogaard,
H., de Groot, H., Saito, K., Cassman, K.G., van Ittersum, M.K., 2015. From field to
atlas: upscaling of location-specific yield gap estimates. Field Crop Res. 177,
98–108.
van Ittersum, M.K., Rabbinge, R., 1997. Concepts in production ecology for analysis
and quantification of agricultural input-output combinations. Field Crops Res.
52, 197–208.
van Ittersum, M.K., Cassman, K.G., Grassini, P., Wolf, J., Tittonell, P., Hochman, Z.,
2013. Yield gap analysis with local to global relevance—a review. Field Crop
Res. 143, 4–17.
117
van Wart, J., van Bussel, L.G., Wolf, J., Licker, R., Grassini, P., Nelson, A., Boogaard, H.,
Gerber, J., Mueller, N.D., Claessens, L., van Ittersum, M.K., Cassman, K.G., 2013.
Use of agro-climatic zones to upscale simulated crop yield potential. Field Crop
Res. 143, 44–55.
Yang, H.S., Dobermann, A., Lindquist, J.L., Walters, D.T., Arkebauer, T.J., Cassman,
K.G., 2004. Hybrid-maize—a maize simulation model that combines two crop
modeling approaches. Field Crop Res. 87, 131–154.
Yang, H., Dobermann, A., Cassman, K.G., Walters, D.T., 2006. Features, applications,
and limitations of the Hybrid-Maize simulation model. Agron. J. 98, 737–748.