Agricultural Water Management 228 (2020) 105869

Contents lists available at ScienceDirect

Agricultural Water Management

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

Study on factors affecting corn yield based on the Cobb-Douglas production T

function
Qichen Zhanga,b, Weihong Dongb,c,*, Chuanlei Wena,b, Tong Lia,b
a
Key Laboratory of Groundwater Resources and Environments, Ministry of Education, Jilin University, Changchun, People’s Republic of China
b
Institute of Water Resources and Environment, Jilin University, Changchun, People’s Republic of China
c
Construction Engineering College of Jilin University, Changchun, People’s Republic of China

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

Keywords: This paper presents an analysis of the quantitative correlations between corn yield and its influencing factors in
Remote-sensing interpretation Daqing City, China by establishing a Cobb–Douglas production function model. The effective precipitation, corn
Influencing factors planting area and chemical fertilizer and pesticide application rates were selected as the influencing factors of
Effective precipitation corn yield. Using the Cobb–Douglas production function model, the output elasticity and degree of influence for
Thiessen polygon
each factor on increasing grain yield were calculated. The current fertilizer and pesticide application rates,
effective precipitation and planting area had positive effects on increasing corn yield, and the Daqing City area
has the potential to produce more corn. Among the four influencing factors, the amounts of pesticide applica-
tions had the greatest impact on corn yield, followed by planting area, amounts of chemical fertilizer applica-
tions and, finally, effective precipitation. In this study, we used remote-sensing images combined with me-
teorological station data to calculate the effective precipitation in corn fields in Daqing City. The accuracy of this
method was 0.01%–11.0% greater than that of the traditional effective precipitation calculation method. The
innovation was the use of Thiessen polygons to calculate regional precipitation by combining satellite images
with ground meteorological station data. The insufficient sensitivity of satellite inversion for precipitation (short
and heavy rainfalls cannot be detected) and insufficient temporal resolution were avoided by using precipitation
data from ground meteorological stations. Using satellite image interpretations, the weight coefficient of pre-
cipitation could be confirmed according to the location and size of the study area, improving the accuracy of
Thiessen polygons in calculating regional precipitation. However, this method still has limitations. When cal-
culating precipitation over a short time, it can be limited by the cloudiness of satellite images. When calculating
the long-term precipitation trend, it can be limited by incomplete precipitation data from surface meteorological
stations.

1. Introduction in the environmental conditions and planting modes, and the factors
affecting corn yield are also complex. When studying the possibility of
Food security is the most basic issue related to human development. increasing the corn yield in Ghana, Marloes found that although corn
The safety of food affects the stability and development of the world yield can be increased by improving varieties and increasing growth
economy. As the population grows, lifestyles and diets change, and food rates, the key to increasing corn yield depends on the interactions of
demand will increase by ∼70% by 2050. Countries worldwide are fa- household, soil and management factors (van Loon et al., 2019v). Bibi
cing pressure to maintain food supplies (Hoff, 2011). Corn is the most found that appropriate changes in the type and application of herbi-
widely grown crop in China, with a planting area that ranks first in the cides can increase corn yield (Bibi et al., 2019). Based on corn planting
world (Wu and Liu, 2018), and its output ranks second in the world data from 15 countries, Li found that corn yield is closely related to
(Yang et al., 2017). Exploring the influencing factors of corn yield is of yield per unit water usage, yield per unit nitrogen applied and average
great significance for increasing corn yield potential, corn yield, and temperature (Li et al., 2019b). Suat found that farming methods have
farmers’ incomes, as well as ensuring food security (Li et al., 2019a). an impact on crop yield, water use and economic benefits (Irmak et al.,
Because of the wide range of corn planting areas, there are differences 2019). Zhao used the Apsim corn model to analyze the effects of

⁎
Corresponding author at: Institute of Water Resources and Environment, Jilin University, No. 2519 Jiefang Road, Changchun City, Jilin Province of China,
People’s Republic of China.
E-mail address: dongweihong@jlu.edu.cn (W. Dong).

https://doi.org/10.1016/j.agwat.2019.105869
Received 4 July 2019; Received in revised form 15 October 2019; Accepted 15 October 2019
Available online 10 November 2019
0378-3774/ © 2019 Elsevier B.V. All rights reserved.
Q. Zhang, et al. Agricultural Water Management 228 (2020) 105869

precipitation, soil and technical management on corn yield. The influ- 2.2. Data collection
ence of various factors on corn yield varied in different corn planting
regions in China (Zhao et al., 2018). Even in the same province, corn The traditional agricultural resource management was extensive,
yield will vary owing to factors such as accumulated temperature, and its information acquisition method is based on the combination
planting scale and soil texture (Jiang et al., 2019). census, sample survey, key survey and comprehensive survey. It is re-
Daqing City is located in the first accumulated temperature zone in ported from the village level and finally approved by the National
Heilongjiang Province, China, an area suitable for corn planting and a Bureau of Statistics. However, owing to the disadvantage of agricultural
main corn growing region (Zhu et al., 2014). Corn yield increased ra- planting dispersion, spatial and temporal variability (Shi et al., 2015),
pidly from 2000 to 2013 but started to decline in 2014. In this study, and inconsistent statistical methods, the coherence of data not very
corn yield was used as a research target, and the factors affecting corn accurate. Remote sensing image is an important means to obtain ground
yield in 2001–2014 in Daqing City, a semi-arid region of Northeast information. High and medium resolution remote sensing images can be
China were collected. By constructing a Cobb–Douglas (C–D) produc- obtained by satellite, airplane and radar. Through image analysis,
tion function model, a quantitative analysis of the correlations between people can get the information they need. Agricultural remote sensing
corn yield and various influencing factors was performed and feasible promotes agricultural production, management and research from the
changes were proposed to promote corn yield in Daqing City. The traditional extensive stage to the precise stage (Zhao, 2014). Precision
proposed correction method is unique in the following respects: Pre- agriculture refers to the use of satellite data to investigate the current
cipitation was calculated using a Thiessen polygon. This is different situation of farmland area and distribution, and to monitor crop
from the previously used weight calculation of a Thiessen polygon area. growth, soil water content and other information. Agricultural quanti-
In this study, the spatial distribution structure of the corn fields was tative remote sensing, as one of the core of precision agriculture (Huang
obtained using a remote-sensing interpretation method, and the pre- et al., 2018; Tang, 2018), is widely used in agricultural resources
cipitation was calculated using the weight of the corn field area in a survey, crop yield estimation, agricultural disaster monitoring, etc.
Thiessen polygon, which improved the accuracy of the actual pre- (Chen et al., 2016). In recent years, with the emergence of various sa-
cipitation used in corn planting. tellites with high spatial, temporal and spectral resolution, such as
THEOS satellite (RADI, 2011a), EROS-B satellite (RADI, 2011b), Op-
tiSAR satellite constellation (Fox et al., 2017), the accuracy of crop
2. Study area and data collection remote sensing identification, classification, land use and cover change
monitoring has been constantly improved (Liu et al., 2018), and agri-
2.1. Study area cultural quantitative remote sensing technology has been further de-
veloped.
Daqing City is located in the west of Heilongjiang Province (Fig. 1), In addition to remote-sensing data, meteorological and agricultural
with a geographical position between 45°22′ and 47°28′ north latitude data availability was important in the selection of study time periods.
and 123°44′ to 125°48′ east longitude, and a total area of 21,000 square Meteorological and agricultural data are obtained through meteor-
kilometers. The study area belongs to the first accumulated temperature ological stations and yearbooks. When selecting remote-sensing data, it
zone in Heilongjiang Province (above 2700 °C), with an annual average is necessary to avoid changes in ground cover type caused by geolo-
temperature of 4.2 °C. From July to August, the average temperature is gical, meteorological disasters (Wang et al., 2016), anthropogenic ef-
23.3 °C. The sunshine duration during the crop growing season fectives, seasonal differentiations, etc. And the imaging season of re-
(May–September) is 1300–1350 h. The precipitation during the mote sensing image directly affects the quality of interpretation, so it is
growing period is generally above 350–480 mm, accounting for more necessary to select remote sensing images during the growing period of
than 85% of the annual precipitation. crops to interpret the image according to the attributes of the objects
As of 2015, the corn planting area in Daqing City accounted for themselves and the differences between different regions of the same
97.9% of the sown area of dryland crops in Daqing City, and the corn objects. The optimum time is August-September. At this time, corn
yield was about 4.5 million tons, accounting for 97.60% of the dryland crops are close to maturity, but there is no harvest, so it is convenient to
crop yield. According to field surveys and data collection, the land distinguish different crops (Yang and Liu, 2008). While avoiding cloud
cover types in the study area are mainly corn fields, paddy fields, urban cover, poor image quality and other negative factors. After screening,
areas, waters (rivers, lakes and reservoirs), other land(sand, bare land, two images were selected from 2001 and 2014. Therefore, the time
other dryland crops), saline-alkali soil and meadow/forest.

Fig. 1. Location of Daqing City.

2
Q. Zhang, et al. Agricultural Water Management 228 (2020) 105869

Fig. 2. Distribution of the research area and meteorological stations.

3
Q. Zhang, et al. Agricultural Water Management 228 (2020) 105869

period for this study was determined to be 2001–2014. Vector Machine Classification. The support vector machine classifier
has the characteristics of strong adaptability, fast learning speed and a
2.2.1. Remote-sensing data small requirement for training scale. It has a high classification accu-
The remote-sensing images of the crop growth period show the racy (Mountrakis et al., 2011; Zhang et al., 2013). Therefore, the sup-
types of ground objects more clearly. Therefore, this paper selected a port vector machine classifier was selected.
remote-sensing image of the plant growth period (May–September) for
interpretation. Remote-sensing images data set is provided by 3.1.4. Evaluation method selection
Geospatial Data Cloud site, Computer Network Information Center, The confusion matrix method is a widely used classification accu-
Chinese Academy of Sciences. (http://www.gscloud.cn). Because re- racy evaluation method for remote-sensing applications and can di-
mote-sensing images are interfered by the factors, such as resolution rectly summarize the main classification accuracy information (Smits
and cloudiness, only two images were screened from those of et al., 2010). Therefore, we chose the confusion matrix method to
2000–2014. They were taken on August 11, 2001 and September 24, evaluate the classification results.
2014, and had image cloud cover of less than 1%. The image of 2001
was selected from Landsat 7, and the image of 2014 was selected from 3.1.5. Evaluation method selection
Landsat 8. In the results of remote-sensing interpretation classification, some
misclassified pixels will inevitably occur, which need to be eliminated
2.2.2. Meteorological data or reclassified. Currently, commonly used methods are Majority/
Eight meteorological stations, Qian Gorlos, Qian’an, Harbin, Anda, Minority analysis, Clump and Sieve. The Clump will cause some smaller
Tailai, Mingshui, Qiqihar and Fuyu, near Daqing City were selected as regions of interest to be included in the larger region of interest, re-
the sources of meteorological data for this study (Fig. 2). sulting in some errors in the final classification results. The Sieve will
result in unclassified pixels in the study area, leading to a reduction in
2.2.3. Other data the overall area (Deng et al., 2014). So, we chose to perform a Majority/
Corn yield, fertilizer application volume, pesticide dosage, planting Minority analysis.
area and other data were taken from the China Statistical, Heilongjiang
Statistical and Daqing Statistical yearbooks. 3.2. Method for calculating effective precipitation

3. Methods The main source of water for dryland crop growth is precipitation,
but precipitation is a meteorological variable that is difficult to measure
3.1. Remote-sensing image interpretation accurately (Tapiador et al., 2012). The objective law of uneven spatial
and temporal distribution determines that it cannot be calculated by
3.1.1. Image preprocessing conventional linear methods. The meteorological station is the most
The remote-sensing images were downloaded according to the la- direct means to obtain precipitation data. However, the meteorological
titude and longitude of the study area. The image resolution was station network is also irregularly distributed in space, and it is difficult
30 m × 30 m, the projection mode was Universal Transverse Mercator, to truly reflect the spatial distribution of actual precipitation. Common
the ellipsoid was WGS84, and the resampling method selected was methods for calculating the amount of surface precipitation include the
Cubic Convolution. The area of interest was established on the down- calculation of the average method, the equal precipitation line method,
loaded image, and the image was sequentially subjected to band mer- the Thiessen Polygon method, the triangle method and the grid inter-
ging, radiometric calibration, atmospheric correction and image mo- polation method (Liu, 2018; Long and liao, 2012). Among them, the
saic. It was divided according to the study area vector graphics. Thiessen Polygon method is widely used in hydrology (Qin et al., 2017;
Different multilayer stack display can enhance different ground objects. Yoshikawa and Shiozawa, 2006), meteorology and other fields because
According to the summary of predecessors (ENVI-IDL, 2013), Landsat-7 of its simple principle, convenient calculation and objective results
and Landsat-8 both have suitable multilayer stacks for agricultural (Chen et al., 2019). The Thiessen Polygon method, also known as ver-
vegetation classification. The multilayer stack of Landsat-7 is Band 5 tical bisection or weighted averaging, is a method of calculating
[Near-Infrared (1.55–1.75 μm)], Band 4 [Near-Infrared (0.77–0.90 μm)] average precipitation. The method first divides the area into a large
and Band 1 [Visible (0.45–0.52 μm)]. And the multilayer stack of number of acute triangles by connecting adjacent precipitation stations,
Landsat-8 is Band 6 [Short-wave infrared 1 (1.57–1.65 μm)], Band 5 then makes vertical bisectors in each of the triangles. The intersecting
[Near-Infrared (0.85–0.88 μm)] and Band 2 [Blue (0.45–0.51 μm)]. The vertical bisectors around each station form a polygon, and the area of
images were stretched linearly by 2% (Fig. 3). the polygon is calculated. The ratio of the area to the basin is used as
Corn yield, fertilizer application volume, pesticide dosage, planting the precipitation weight coefficient of each polygon. Finally, the pre-
area and other data were taken from the China Statistical, Heilongjiang cipitation of each station is multiplied by the weight of the area occu-
Statistical and Daqing Statistical yearbooks. pied by the station.
However, the area that needs to be determined in this study is the
3.1.2. Remote-sensing image classification corn field area in Daqing City; therefore, two improvements were made
Supervised classifications were determined to extract feature in- to the calculation process.
formation from remote-sensing images. The field survey and visual in- First, if the calculation of the precipitation weight is made using the
terpretation were carried out in turn, and the land cover types in administrative area of Daqing City, then the correlation between pre-
Daqing City were divided into seven categories: corn fields, paddy cipitation and corn production cannot be accurately reflected.
fields, meadow/forest, urban areas, saline-alkali soil, other land (sand, Therefore, this study will use the ratio of the corn field area of each
bare land, other dryland crops) and waters. Training samples were Thiessen polygon in the study area obtained by remote sensing to the
delineated according to land type. total area of the corn field in the study area as the precipitation weight
coefficient on each Thiessen polygon. Second, only some of the total
3.1.3. Classifier selection precipitation in May–September can be used by crops; therefore, it is
Currently, the classifiers commonly used in the supervised classifi- necessary to calculate the effective precipitation for each Thiessen
cation of The Environment for Visualizing Images include polygon.
Parallelepiped, Minimum Distance, Mahalanobis Distance (Deng et al., A previous study (Liu et al., 2009) found that when using the USDA
2014), Likelihood Classification, neural Net Classification and Support Soil Conservation Service recommended method to calculate effective

4
Q. Zhang, et al. Agricultural Water Management 228 (2020) 105869

Fig. 3. Remote-sensing preprocessed images, (a) is the preprocessed image of year 2001, (b) is the preprocessed image of year 2014.

precipitation, the effective precipitation will be too small or too large 3.3. Quantitative relationship analysis method for corn yield and its
when calculating the effective precipitation in days and months. It is influencing factors
more appropriate to calculate the effective precipitation over 10 days.
Therefore, the effective precipitation in this study was calculated over 3.3.1. Method selection
10 days. For the agricultural industry, studying the impact of various pro-
The Thiessen polygon calculation formula is as follows: duction factors can not only save agricultural production costs, but can
also increase the overall production capacity. Studying the impact of
n n
f1 P1 + f2 P2 + …fn Pn 1 various factors of production on agricultural production is actually to
P =
f1 + f2 + …fn
=
F
∑ fi Pi = ∑ Ai Pi, calculate their contribution to agricultural production. At present, the
i=1 i=1 (1)
main methods used in academia are data envelopment (Yan, 2019),
where P represents the average precipitation in the study area, the unit C–D production function (Li and Li, 2018; Lunduka et al., 2019) and
is mm; Pn represents the precipitation of each Thiessen polygon, the Solow residual (Moghaddasi and Pour, 2016). Among them, the C–D
unit is mm; fi represents the area of corn field in each Thiessen polygon, production model is widely used because it is suitable for calculating
the unit is km2; n represents the number of meteorological stations long-term estimations and is easy to calculate(Zhu et al., 2018).
around the study area; F represents the total area of corn field in the The C–D production function model is shown below:
study area, the unit is km2; Ai represents the weight of the corn fields at n
each meteorological station. Y= A∏ Xibi
The effective precipitation calculation formula is as follows: i=1 (3)

Where Y represents crop yield; A represents an undetermined coeffi-

[Pd × (125 − 0.6 × Pd )]/125, Pd ≤ (250/3) mm
Pe = ⎧ , cient; i represents the ith factor affecting crop yield; n represents the
⎨
⎩ (125/3) + 0.1 × Pd, Pd > (250/3) mm (2) total number of factors affecting crop yield; Xi represents the ith impact
factor input; and bi represents the elastic coefficient of the ith influence
where Pe represents the effective precipitation over 10 d, the unit is factor.
mm; and Pd represents the amount of precipitation over 10 days, the
unit is mm.
3.3.2. Impact factor screening
There are various influencing factors on corn yield. The necessary
comprehensive selection of the indicators is difficult to achieve in

5
Q. Zhang, et al. Agricultural Water Management 228 (2020) 105869

practice. Therefore, it is necessary to select the dominant factors from 2014 was less than 1%. Consequently, the interpretation accuracy
many influencing factors. At the same time, to avoid collinearity, the meets the requirements, and the results are credible. From 2001–2014,
indicators selected in this paper were direct influencing factors, and the rate of corn change in each year is unknown., so in this study, it is
indirect factors were not considered. Therefore, based on the degree of assumed that the annual rate of change of the cornfield area in each
impact on corn production, the difficulty of data collection and the Thiessen polygon is constant. The proportional weights of the Thiessen
actual situation of agricultural production in Daqing City, referring to polygon on corn fields are shown in Table 4.
the indicators in the Statistical Yearbook, the influencing factors were According to the precipitation recorded at the meteorological sta-
determined to be the amounts of chemical fertilizers and pesticides tions and the weight of precipitation (the proportion of corn fields)
applied, the amount of effective precipitation and the size of the corn from 2001 to 2014, the annual effective precipitation of the study area
planting area. was obtained (Table 5).

4.3. Analysis of factors affecting corn yield

4. Results

According to the collected corn yield, corn planting area, the

4.1. Remote-sensing interpretation results
amounts of chemical fertilizers and pesticides applied, and the effective
precipitation required for corn planting, the input amounts of various
In 2001, most of the corn fields were distributed in the southern part
material elements, such as fertilizers, were obtained (Table 6)
of the study area, especially in the southeastern region. By 2014, except
The C–D production function model was used to estimate the main
for the corn fields distributed in the southeastern region, the northern
influencing factors of corn production in Daqing City. The basic formula
regions also had a wide distribution. From 2001 to 2014, there were
was as follows:
629.41 km2 of corn fields were converted into six other types of land
use. 460.35 km2 were converted into meadow/forest, 89.04 km2 into Y = A × C α × P β × R γ × Lδ (4)
other land, 53.59 km2 into paddy fields, 14.56 km2 into urban areas,
The model expression obtained by logarithm transformation was as
8.58 km2 into waters, and 2.99 km2 into saline-alkali soil. The conver-
follows:
sion area accounted for 33.73% of the corn field area in 2001. By 2014,
a total of 4368.89 km2 of land had been converted into corn fields, lnY = lnA + α lnC + β lnP + γ lnR + δ lnL (5)
including 147.39 km2 of urban areas, 15.19 km2 of saline-alkali soil, where Y represents corn yield as a dependent variable; A represents the
23.82 km2 of paddy fields, 3479.13 km2 of meadow/forest, 692.43 km2 comprehensive technical level of the base period, which is a constant
of other land and 9.93 km2 of waters. Corn field area converted from term; C represents the amount of chemical fertilizer applied; P re-
other types of land use accounted for 77.94% of the total area of corn presents the amount of pesticide applied; R represents the effective
fields in 2014. According to the classification results, the cornfield area precipitation; L represents the planting area; and α, β, γ, δ and ε re-
increased from 1866.03 km2 in 2001 to 5603.12 km2 in 2014 (Table 1). present their production elasticity levels, respectively.
Fig. 4 shows the land use classifications of Daqing City in 2001 and Substituting the 2001–2014 data into the C–D production model
2014 (Table 1). yielded the results. The model’s R2 value was 0.958, adjusted R2 value
The closer the overall accuracy of the classification and the Kappa was 0.940, F test value was 51.93 and Durbin–Watson value was 2.234.
coefficient are to 1, the more reliable the remote sensing result is (Deng The coefficients are shown in Table 7.
et al., 2014; Warrens, 2016). And the classification results were eval- According to the results in Table 7, the model’s expression was as
uated using a confusion matrix. The overall accuracy coefficients of follows:
2001 and 2014 were 96.40% and 99.67%, respectively, which were
lnY = −3.934 + 0.796lnC + 0.423lnP + 0.384lnR + 0.484lnL (6)
both greater than 95%. The Kappa coefficients of 2001 and 2014 were
both greater than 0.9 at 0.9500 and 0.9930, respectively. Therefore, the The model was tested using the statistical data in the above tables.
interpretation of the remote-sensing data in this study was reasonable The adjusted R2 value was 0.94, indicating that 94% of the corn yield
and reliable. increase effect was related to changes in fertilizer, pesticide, pre-
cipitation and planting area. The t-test value of each of the influencing
4.2. Effective precipitation calculation results factors, except the constant term, was not greater than its critical value
(t = 1.4, at a significance level of 0.1). For agricultural production,
The Thiessen polygons for 2001 and 2014 are shown in Fig. 5. The fertilizers, pesticides, land and precipitation are all indispensable.
ratio of corn field area in each Thiessen polygon to the total corn field Moreover, each variable had a significant effect on corn yield.
area in the study area was calculated as the precipitation weight of each According to the Durbin–Watson test, there was no autocorrelation
Thiessen polygon. among the influencing factors; therefore, the results were retained.
The seven types of land use present in 2001 and 2014 are shown in Model analysis:
Tables 2 and 3.
As seen in Tables 2 and 3, the total area error between 2001 and 1 The sum of the output elasticity of each input factor was greater

Table 1
Land-use type conversion from 2001 to 2014.
2001

2014 Urban areas Saline-alkali soil Paddy fields Meadow/ forest Other land Waters Corn fields

Urban areas 39.84 16.36 2.41 4.2 8.29 5.63 0.78

Saline-alkali soil 3.89 31.96 0.15 0.29 1.91 0.85 0.16
Paddy fields 0.85 0.46 56.66 3.92 4.33 2.24 2.89
Meadow/ forest 23.43 7.7 29.12 40.46 28.3 10.29 24.67
Other land 18.48 29.57 2.27 15.68 31.56 1.68 4.77
Waters 7.5 12.53 5.98 1.45 4.11 78.82 0.46
Corn fields 6.02 1.44 3.4 34 21.5 0.6 66.27

6
Q. Zhang, et al. Agricultural Water Management 228 (2020) 105869

Fig. 4. Supervised classification results of remote-sensing images, (a) is the classification results for year 2001, (b) is the classification results for year 2014.

than 1, and it was positive, indicating that the growth rate of corn Then, the weight of the polygon area of Thiessen polygon was replaced
production in Daqing City is greater than the speed of corn material by the weight of the corn field area to calculate the effective pre-
input. Thus, corn planting in Daqing City is in a sustainable growth cipitation of the farmland.
stage and has the potential for increased grain production. In order to understand the difference between the new method and
2 Although the t-test values for chemical fertilizers, pesticides, the traditional method in calculating the effective precipitation, this
planting area and precipitation were not significant, the coefficients study also used the traditional Thiessen polygon method to calculate
are positive, indicating that these four variables have positive im- the multi-year effective precipitation in Daqing City. The traditional
pacts on corn yield. Thiessen polygon weights are shown in Table 8. Then use the meteor-
3 According to standardization coefficients, the order of influence for ological station precipitation, traditional weights and formula 2 to
the various factors on corn yield is: Pesticide application calculate traditional effective precipitation.
amount > planting area > fertilizer application amount > Compared with the traditional calculation of precipitation using the
precipitation amount. weight of the Thiessen polygon, the accuracy of effective precipitation
calculated using the weight of the corn field increased by 0.01%–11%
5. Discussion in Daqing City. As is shown in Table 9.
Although the method of using remote-sensing interpretation plus
In the past, most studies on agricultural precipitation were con- meteorological station data to determine effective precipitation is more
ducted directly with meteorological station data, but the spatial dis- accurate than traditional effective precipitation, there are still some
tribution characteristics of farmland were neglected. There is also sa- problems. The first is the accuracy of the remote-sensing interpretation.
tellite remote sensing for the indirect detection of precipitation, but it is The interpretation of general farmland requires visual interpretation
not sensitive enough to distinguish between light and heavy pre- plus supervised classification. The visual interpretation depends largely
cipitation (Liu et al., 2011). on the degree of human recognition of remote-sensing features and the
Therefore, this study proposed a feasible and effective method that resolution of the images (e.g. only 30 m for multispectral Landsat 7 and
can be used as an estimation for precision agriculture and seasonal crop 9 images). Moreover, the location and number of samples selected will
area. First, the Thiessen polygons were established using the locations also affect the final supervised classification results. Second, meteor-
of the meteorological stations near the study area, and then the spatial ological stations are also affected by wind, evaporation, environmental
distribution of the farmland as determined by satellite remote sensing noise and the size and distribution of raindrops when recording pre-
was used to calculate the weight of the corn field area in each Thiessen cipitation (Michaelides et al., 2009), and the distribution and failure of
polygon (the ratio of the corn field area to the total corn field area). meteorological stations also affects the research.

7
Q. Zhang, et al. Agricultural Water Management 228 (2020) 105869

Fig. 5. Thiessen polygons, (a) is the Thiessen Polygon of Daqing City in 2001, (b) is the Thiessen Polygon of Daqing City in 2014.

From the results of the C–D production function, it can be under- The output elasticity of chemical fertilizers was 0.796. As the most
stood that the production elasticity of the pesticide application amount obvious factor influencing the increase in corn planting yield, the in-
was 0.423, indicating that for every 1% increase in the pesticide ap- creased rate of fertilizer application was obviously less than that of
plication rate, the corn output was increased by 0.423%. The pesticides planting area, indicating that the application efficiency of fertilizers has
used in Daqing City are mainly herbicides and insecticides. As seen in improved.
Table 6, the increase in the rate of pesticide application was greater The impact of precipitation on corn yield was small, mainly because
than the growth rate of the planting area. This indicates that the in- Daqing City is in a semi-arid area. Consequently, most of the planted
crease in corn production is effective. However, it also indicates that the corn varieties are drought-tolerant varieties, with low demands for
resistance of weeds and pests is increasing, forcing the increased ap- precipitation. However, precipitation is still an indispensable input
plication of pesticides. Whether the utilization efficiency of pesticides is factor for corn cultivation.
improving requires further research. Although this paper only analyzed the material influencing factors
The planting area had a positive effect on the output of corn. For of corn production, it provides significant guidance for agricultural
every 1% increase in planting area, the corn output increased by production in Daqing City. In the agricultural production process, the
0.484%. influencing factors included not only water, fertilizer and pesticides at

Table 2
Area and distribution of various types of land in 2001 (unit: km2).
Meteorological station Corn fields Paddy fields Waters Saline-alkali soil Meadow/ forest Other land Urban areas Total area

Qian Gorlos 238.38 284.22 198.11 43.27 928.14 245.11 193.81 2131.04
Qian an 0.67 0.10 4.25 0.26 13.82 9.31 0.74 29.15
Harbin 473.98 59.88 18.04 8.20 100.07 88.15 35.21 783.53
An da 1089.81 99.22 687.45 830.12 4201.41 1215.42 1613.90 9737.33
Tai lai 29.67 102.97 287.84 92.89 726.19 943.10 140.09 2322.75
Mingshui 30.38 67.73 76.86 5.74 2146.16 205.17 111.54 2643.59
Qiqihar 1.14 28.84 283.72 24.57 493.15 162.72 86.80 1080.94
Fu yu 2.01 57.61 98.39 50.07 1623.73 351.63 266.33 2449.78
Total area 1866.03 700.57 1654.66 1055.12 10232.67 3220.60 2448.42 21178.07

8
Q. Zhang, et al. Agricultural Water Management 228 (2020) 105869

Table 3
Area and distribution of various types of land in 2014 (unit: km2).
Meteorological station Corn fields Paddy fields Waters Saline-alkali soil Meadow/ forest Other land Urban areas Total area

Qian Gorlos 291.42 392.73 171.48 59.92 571.82 513.59 130.09 2131.04
Qian an 3.66 8.01 7.99 0.15 7.47 0.49 1.39 29.16
Harbin 418.21 61.18 28.78 3.04 206.79 27.00 38.51 783.51
An da 2689.82 87.61 744.19 416.56 2517.40 2048.11 1233.65 9737.34
Tai lai 510.40 351.05 430.60 47.65 339.54 535.92 107.64 2322.80
Mingshui 869.27 101.07 92.53 3.09 1240.69 162.93 173.96 2643.54
Qiqihar 196.92 21.92 340.02 5.33 361.81 101.67 53.23 1080.91
Fu yu 623.42 30.71 134.54 6.11 1296.19 128.53 230.29 2449.79
Total area 5603.12 1054.29 1950.12 541.86 6541.72 3518.24 1968.76 21178.10

Table 4
Changes in the proportions of corn fields.
Years Qian Gorlos Qian’an Harbin Anda Tailai Mingshui Qiqihar Fuyu

2001 12.77% 0.04% 25.40% 58.40% 1.59% 1.63% 0.06% 0.11%

2002 12.19% 0.04% 24.02% 57.60% 2.17% 2.70% 0.33% 0.96%
2003 11.61% 0.04% 22.64% 56.80% 2.75% 3.76% 0.59% 1.80%
2004 11.03% 0.04% 21.26% 56.00% 3.33% 4.83% 0.86% 2.65%
2005 10.44% 0.05% 19.88% 55.20% 3.90% 5.90% 1.12% 3.50%
2006 9.86% 0.05% 18.50% 54.41% 4.48% 6.97% 1.39% 4.35%
2007 9.28% 0.05% 17.12% 53.61% 5.06% 8.04% 1.65% 5.19%
2008 8.70% 0.05% 15.74% 52.81% 5.64% 9.10% 1.92% 6.04%
2009 8.11% 0.06% 14.36% 52.01% 6.22% 10.17% 2.18% 6.89%
2010 7.53% 0.06% 12.98% 51.21% 6.80% 11.24% 2.45% 7.74%
2011 6.95% 0.06% 11.60% 50.41% 7.37% 12.31% 2.71% 8.59%
2012 6.37% 0.06% 10.22% 49.61% 7.95% 13.37% 2.98% 9.43%
2013 5.78% 0.07% 8.84% 48.81% 8.53% 14.44% 3.24% 10.28%
2014 5.20% 0.07% 7.46% 48.01% 9.11% 15.51% 3.51% 11.13%

Table 5
Effective precipitation in corn fields in Daqing City.
Years 2001 2002 2003 2004 2005 2006 2007

effective precipitation, mm 212.79 286.99 329.12 244.68 347.23 308.29 256.86

Years 2008 2009 2010 2011 2012 2013 2014

effective precipitation, mm 293.32 299.22 291.31 320.26 350.62 367.17 351.91

Table 6 Table 7
Input and output of corn production in Daqing City. Coefficients.
Years Corn yield Planting Fertilizer Effective Pesticides Model Coefficients Std.Error T test Beta
(10,000 tons) area (10,000 tons) precipitation (tons)
(10,000 (mm) lnA −3.934 2.121 −1.855
hectares) lnL 0.484 0.449 1.079 0.307
lnC 0.796 0.860 0.925 0.304
2001 67.52 18.66 12.44 212.79 619.58 lnR 0.384 0.458 0.838 0.086
2002 112.11 17.81 13.78 286.99 734.09 lnP 0.423 0.389 1.086 0.328
2003 82.65 17.63 15.68 329.12 663.82
2004 123.15 25.63 16.30 244.68 774.19
2005 165.45 25.50 17.89 347.23 805.85
6. Conclusions
2006 177.57 35.61 17.27 308.29 1195.41
2007 160.67 34.58 19.34 256.86 1250.36
2008 242.80 29.97 20.71 293.32 1546.79 This study quantified the factors affecting corn production in
2009 310.56 45.07 23.36 299.22 1782.30 Daqing City. The innovation of this research was to combine satellite
2010 416.18 51.41 24.81 291.31 1859.39 remote sensing with meteorological station data to calculate the ef-
2011 458.68 52.41 25.70 320.26 2086.96
fective precipitation of farmland. Compared with traditional effective
2012 540.38 53.57 26.04 350.62 2452.47
2013 431.16 54.15 27.35 367.17 2619.93 precipitation calculation methods, the accuracy increased by
2014 451.17 56.03 27.73 351.91 2701.76 0.01%–11%. In the past, precipitation was calculated either by using
satellite remote sensing technology or by using ground meteorological
station data. Few studies used the combination of satellite remote
the input end, but also factors such as climate (Jagustović et al., 2019), sensing technology and ground meteorological station data to calculate
soil quality (Akhtar et al., 2018), labor costs, agricultural subsidies and precipitation. This paper introduces and demonstrates the feasibility of
food price fluctuations (Majumder et al., 2012). Therefore, further re- this method.
search is needed on the quantification of various non-material influence Using the C–D production function model, the water–energy–grain
factors of precision agriculture. feeding mechanism of Daqing City was quantitatively studied. The

9
Q. Zhang, et al. Agricultural Water Management 228 (2020) 105869

Table 8
Weights of traditional Thiessen polygons.
Station Qian Gorlos Qian’an Harbin Anda Tailai Mingshui Qiqihar Fuyu

Weight 10.06% 0.14% 3.70% 45.98% 10.97% 12.48% 5.10% 11.57%

Table 9 Declaration of Competing Interest

Two effective precipitation calculations and their comparison.
Years Traditional effective Effective precipitation in Relative
None.
precipitation(mm) corn fields(mm) accuracy
Acknowledgments
2001 198.08 212.79 7.43%
2002 267.93 286.99 7.11%
2003 338.85 329.12 −2.87%
We thank Lesley Benyon, PhD, from Liwen Bianji, Edanz Group
2004 220.43 244.68 11.00% China (www.liwenbianji.cn/ac), for editing the English text of a draft of
2005 353.69 347.23 −1.83% this manuscript. I would like to acknowledge the contributions of all the
2006 316.55 308.29 −2.61% reviewers, and thank them for their insightful comments on early drafts
2007 244.37 256.86 5.11%
of this article.
2008 294.71 293.32 −0.47%
2009 299.25 299.22 −0.01%
2010 280.79 291.31 3.75% References
2011 319.88 320.26 0.12%
2012 339.11 350.62 3.40% Akhtar, S., Li, G.-c., Ullah, R., Nazir, A., Iqbal, M.A., Raza, M.H., Iqbal, N., Faisal, M.,
2013 365.21 367.17 0.54% 2018. Factors influencing hybrid maize farmers’ risk attitudes and their perceptions
2014 353.48 351.91 −0.44% in Punjab Province, Pakistan. J. Integr. Agric. 17, 1454–1462. https://doi.org/10.
1016/s2095-3119(17)61796-9. Retrieved from.
Bibi, S., Khan, I.A., Hussain, Z., Zaheer, S., Shah, S.M.A., 2019. Effect of herbicides and
intercropping on weeds and yields of Maize and the associated intercrops. Pak. J. Bot.
standardization coefficients show that the amount of pesticide used
51https://doi.org/10.30848/pjb2019-3(45). Retrieved from.
played a decisive role in increasing corn production, followed by the Chen, C., Wang, T., Xu, G., 2019. Analysis of Surface Rainfall in Minhang Area Based on
area planted and the amount of fertilizer used. Although precipitation Thiessen Polygon Algorithms. Shanghai Water 35, 21–23. Retrieved from. https://
had the least impact on grain production, it is also indispensable. kns.cnki.net/KCMS/detail/detail.aspx?dbcode=CJFQ&dbname=CJFDLAST2019&
filename=SWSH201902012&uid=
According to the simulation results, appropriate increases in the ap- WEEvREcwSlJHSldRa1FhdkJkVG5heGJNNk1vNG5VVGI3RnJhaGN2ajUwST0=
plications of pesticides and fertilizers can still increase food yields, but $9A4hF_YAuvQ5obgVAqNKPCYcEjKensW4IQMovwHtwkF4VYPoHbKxJw!!&v=
science-based chemical fertilizer and pesticide application guidelines MTI3MDkrZHVGaURoVjcvTk5qcllackc0SDlqTXJZOUVa
b1I4ZVgxTHV4WVM3RGgxVDNxVHJXTTFGckNVUkxPZVo=.
are needed to avoid polluting the soil, groundwater and rivers. Daqing Chen, Z., Ren, J., Tang, H., Shi, Y., Leng, P., Wang, l., Wu, W., Yao, Y., Hasituya, 2016.
City is located in a semi-arid area with a low annual precipitation, but Progress and perspectives on agricultural remote sensing research and applications in
there are many rivers, lakes and reservoirs, which can increase the ir- China. J. Remot Sens. 20, 748–767. https://doi.org/10.11834/jrs.20166214.
Retrieved from.
rigated corn field area to increase corn yield (Marek et al., 2016). In Deng, S., Chen, Q., Du, H., Xu, E., 2014. ENVI Remote Sensing Image Processing Method,
future research, we plan to improve the C–D production function model second edition ed. Higher Education Press.
by collecting more meteorological and agricultural economic data to ENVI-IDL, 2013. Description of Different Band Combinations of Landsat8. Retrieved
from. http://blog.sina.com.cn/s/blog_764b1e9d01019urt.html.
provide guidance for the sustainable production of corn in Daqing City.
Fox, P., Tyc, G., Beckett, K., 2017. In: The UrtheCast SAR-XL Multi-Band, Multi-Aperture
There are also some difficulties encountered in conducting this re- Spaceborne SAR System 2017 IEEE Radar Conference (RadarConf) IEEE. Seattle, WA,
search: USA. https://doi.org/10.1109/RADAR.2017.7944492. Retrieved from.
Hoff, H., 2011. Understanding the Nexus. Background Paper for the Bonn 2011
Conference: the Water Energy and Food Security Nexus. Retrieved from. Stockholm
(1) Because the area of this study area is large, it is necessary to in- University: Stockholm Environment Institute. http://www.water-energy-food.org/
tegrate three remote sensing images to show the entire study area. en/whats_the_nexus/background.html.
Because there is cloud interference when merging the remote sen- Huang, Y., Chen, Z.-x., Yu, T., Huang, X.-z., Gu, X.-f., 2018. Agricultural remote sensing
big data: management and applications. J. Integr. Agric. 17, 1915–1931. https://doi.
sing images of the same day, only three remote sensing images with org/10.1016/s2095-3119(17)61859-8. Retrieved from.
similar dates can be screened during the corn maturity period (early Irmak, S., Kukal, M.S., Mohammed, A.T., Djaman, K., 2019. Disk-till vs. no-till maize
August to the end of September). For example, in the three remote evapotranspiration, microclimate, grain yield, production functions and water pro-
ductivity. Agric. Water Manage. 216, 177–195. https://doi.org/10.1016/j.agwat.
sensing images in 2014, two were September 24 and one was 2019.02.006. Retrieved from.
September 15. This affects the accuracy of remote sensing inter- Jagustović, R., Zougmoré, R.B., Kessler, A., Ritsema, C.J., Keesstra, S., Reynolds, M.,
pretation to a certain extent. 2019. Contribution of systems thinking and complex adaptive system attributes to
sustainable food production: example from a climate-smart village. Agric. Syst. 171,
(2) In the process of screening 2000–2015 images, only remote sensing 65–75. https://doi.org/10.1016/j.agsy.2018.12.008. Retrieved from.
images of 2001 and 2014 were obtained, because remote sensing Jiang, B., Jiang, H., Qian, C., Li, L., Gong, X., Li, S., Hu, Y., 2019. Advances in research on
images of other years have more or less problems. Therefore, when factors affecting corn yield efficiency and ways of improvement in China. Jiangsu
Agric. Sci. 47, 12–15. https://doi.org/10.15889/j.issn.1002-1302.2019.05.004.
calculating the weight of the corn field in each Thiessen polygon, it
Retrieved from.
can only be assumed that the change rate of the corn field area is Li, J., Li, Y., 2018. Influence measurement of rapid urbanization on agricultural pro-
constant, which affects the amount of effective precipitation in the duction factors based on provincial panel data. Socioecon. Plann. Sci. https://doi.
org/10.1016/j.seps.2018.09.004. Retrieved from.
final calendar year to some extent.
Li, M., Fu, Q., Singh, V.P., Ji, Y., Liu, D., Zhang, C., Li, T., 2019a. An optimal modelling
approach for managing agricultural water-energy-food nexus under uncertainty. Sci.
Total Environ. 651, 1416–1434. https://doi.org/10.1016/j.scitotenv.2018.09.291.
Funding Retrieved from. https://www.ncbi.nlm.nih.gov/pubmed/30360272.
Li, Y., Li, Z., Cui, S., Chang, S.X., Jia, C., Zhang, Q., 2019b. A global synthesis of the effect
of water and nitrogen input on maize (Zea mays) yield, water productivity and ni-
This work was supported by the National Key R&D Program of trogen use efficiency. Agric. For. Meteorol. 268, 136–145. https://doi.org/10.1016/j.
China (Grant No. 2017YFC0404603). agrformet.2019.01.018. Retrieved from.
Liu, H., You, J., Wang, F., Zhou, Y., Li, X., Huang, P., 2018. Application and
Enlightenment of Agricultural Remote Sensing in Eu Countries. Chin. J. Agric.
Resour. Reg. Plann. 39, 280–287. https://doi.org/10.7621/cjarrp.1005-9121.

10
Q. Zhang, et al. Agricultural Water Management 228 (2020) 105869

20180837. Retrieved from. Tang, H., 2018. Progress and prospects of agricultural remote sensing research. Agric. Sci.
Liu, P., 2018. A secondary development of regional rainfall estimation using ArcGIS based Eng. China 8, 167–171. https://doi.org/10.19518/j.cnki.cn11-2531/s.2018.0119.
on Thiessen polygon method. China Flood Drought Manage. 28, 48–49. https://doi. Retrieved from.
org/10.16867/j.issn.1673-9264.2017173. +66. Retrieved from. Tapiador, F.J., Turk, F.J., Petersen, W., Hou, A.Y., García-Ortega, E., Machado, L.A.T.,
Liu, Y., Wang, L., Ni, G., Cong, Z., 2009. Spatial distribution characteristics of irrigation Angelis, C.F., Salio, P., Kidd, C., Huffman, G.J., de Castro, M., 2012. Global pre-
water requirement for main crops in China. Trans. CSAE 25, 6–12. https://doi.org/ cipitation measurement: methods, datasets and applications. Atmos. Res. 104-105,
10.3969/j.issn.1002-6819.2009.12.002. Retrieved from. 70–97. https://doi.org/10.1016/j.atmosres.2011.10.021. Retrieved from.
Liu, Y., Fu, Q., Song, P., Zhao, X., Dou, C., 2011. Satellite retrieval of precipitation: an van Loon, M.P., Adjei-Nsiah, S., Descheemaeker, K., Akotsen-Mensah, C., van Dijk, M.,
overview. Int. J. Adv. Earth Sci. Eng. 26, 1162–1172. Retrieved from. http://kns. Morley, T., van Ittersum, M.K., Reidsma, P., 2019v. Can yield variability be ex-
cnki.net//KXReader/Detail?TIMESTAMP=637064901961435000&DBCODE=CJFD plained? Integrated assessment of maize yield gaps across smallholders in Ghana.
&TABLEName=CJFD2011&FileName=DXJZ201111008&RESULT=1&SIGN= Field Crops Res. 236, 132–144. https://doi.org/10.1016/j.fcr.2019.03.022.
ShiVHcGcdqu1MpAS3MfbJvOWTCo%3d. Retrieved from.
Long, Y., liao, Z., 2012. Surface precipitation calculation of Qingyang area based on Wang, J., Yu, Y., Dong, L., 2016. Calculation of irrigation benefit sharing coefficient by C-
Thiessen polygons method. J. Hebei Univ. Eng. 29, 64–67. Retrieved from. http:// D production function model. Water Resour. Hydropower Northeast China 34, 64–66.
kns.cnki.net//KXReader/Detail?TIMESTAMP=637063842017841250&DBCODE= https://doi.org/10.14124/j.cnki.dbslsd22-1097.2016.11.028. Retrieved from.
CJFQ&TABLEName=CJFD2012&FileName=HJXU201203017&RESULT=1& Warrens, M.J., 2016. Category kappas for agreement between fuzzy classifications.
SIGN=cGnf66XnECgLL3yyMRli7SdZRuI%3d. Neurocomputing 194, 385–388. https://doi.org/10.1016/j.neucom.2016.02.038.
Lunduka, R.W., Mateva, K.I., Magorokosho, C., Manjeru, P., 2019. Impact of adoption of Retrieved from.
drought-tolerant maize varieties on total maize production in south Eastern Wu, L., Liu, Z., 2018. Current research status of water and fertilizer requirement char-
Zimbabwe. Clim. Dev. 11, 35–46. https://doi.org/10.1080/17565529.2017. acteristics, yield and quality of maize. Shanxi Agric. Econ. 12, 50–51. https://doi.
1372269. Retrieved from. https://www.ncbi.nlm.nih.gov/pubmed/30881484. org/10.16675/j.cnki.cn14-1065/f.2018.12.038. +53. Retrieved from.
Majumder, A., Ray, R., Sinha, K., 2012. Calculating rural-urban food price differentials Yan, J., 2019. Spatiotemporal analysis for investment efficiency of China’s rural water
from unit values in household expenditure surveys: a comparison with existing conservancy based on DEA model and Malmquist productivity index model. Sustain.
methods and a new procedure. Am. J. Agric. Econ. 94, 1218–1235. https://doi.org/ Comput. Inform. Syst. 21, 56–71. https://doi.org/10.1016/j.suscom.2018.11.004.
10.1093/ajae/aas064. Retrieved from. Retrieved from.
Marek, G., Gowda, P., Marek, T., Auvermann, B., Evett, S., Colaizzi, P., Brauer, D., 2016. Yang, B., Liu, Y., 2008. Key points of visual interpretation technology for agricultural
Estimating preseason irrigation losses by characterizing evaporation of effective remote sensing images. Modern Agriculture. pp. 37–39. Retrieved from. http://kns.
precipitation under bare soil conditions using large weighing lysimeters. Agric. Water cnki.net//KXReader/Detail?TIMESTAMP=637063221669456250&DBCODE=CJFQ
Manage. 169, 115–128. https://doi.org/10.1016/j.agwat.2016.02.024. Retrieved &TABLEName=CJFD2008&FileName=XDHY200804027&RESULT=1&SIGN=
from. QxqL%2bScWzTRxmK%2b1RroQnj573mg%3d.
Michaelides, S., Levizzani, V., Anagnostou, E., Bauer, P., Kasparis, T., Lane, J.E., 2009. Yang, H., Han, X., Zheng, F., 2017. Comparison of Corn yield, trade, consumption and
Precipitation: measurement, remote sensing, climatology and modeling. Atmos. Res. stocks in major global producing countries (regions) - based on 1996/1997-2016/
94, 512–533. https://doi.org/10.1016/j.atmosres.2009.08.017. Retrieved from. 2017 season data. World Agric. 6, 28–35. https://doi.org/10.13856/j.cn11-1097/s.
Moghaddasi, R., Pour, A.A., 2016. Energy consumption and total factor productivity 2017.06.005. +236. Retrieved from.
growth in Iranian agriculture. Energy Rep. 2, 218–220. https://doi.org/10.1016/j. Yoshikawa, N., Shiozawa, S., 2006. Estimating variable acreage of cultivated paddy fields
egyr.2016.08.004. Retrieved from. from preceding precipitation in a tropical watershed utilizing Landsat TM/ETM.
Mountrakis, G., Im, J., Ogole, C., 2011. Support vector machines in remote sensing: a Agric. Water Manage. 85, 296–304. https://doi.org/10.1016/j.agwat.2006.02.013.
review. ISPRS J. Photogramm. Remote. Sens. 66, 247–259. https://doi.org/10.1016/ Retrieved from.
j.isprsjprs.2010.11.001. Retrieved from. Zhang, P., Shi, W., Wong, M., Chen, J., 2013. A reliability-based multi-algorithm fusion
Qin, J., Chen, Y., Wang, H., 2017. Precipitation interpolation of Liuxihe model for wa- technique in detecting changes in land cover. Remote Sens. (Basel) 5, 1134–1151.
tershed flood forecasting based on Theissen polygon method. China Rural Water https://doi.org/10.3390/rs5031134. Retrieved from.
Hydropower 88–93. Retrieved from. http://kns.cnki.net//KXReader/Detail? Zhao, C., 2014. Advances of research and application in remote sensing for agriculture.
TIMESTAMP=637064902719247500&DBCODE=CJFQ&TABLEName= Trans. Chin. Soc. Agric. Mach. 45, 277–293. https://doi.org/10.6041/j.issn.1000-
CJFDLAST2017&FileName=ZNSD201701022&RESULT=1&SIGN= 1298.2014.12.041. Retrieved from.
KkHXoWCGDfHuBs8cxPBnhsza4f0%3d. Zhao, J., Yang, X., Sun, S., 2018. Constraints on maize yield and yield stability in the main
RADI, 2011a. China Remote Sensing Data Network, THEOS Satellite. Retrieved from. cropping regions in China. Eur. J. Agron. 99, 106–115. https://doi.org/10.1016/j.eja.
http://rs.ceode.ac.cn/satelliteintroduce/introTHEOS.jsp. 2018.07.003. Retrieved from.
RADI, 2011b. China Remote Sensing Data Network, Start Agent Distribution of EROS-B Zhu, H., Lv, J., Li, X., Wang, P., Li, Y., Wang, M., Wang, Q., Jiang, L., 2014. Potential
Satellite Data. Retrieved from. http://rs.ceode.ac.cn/eros-2011-5-4.jsp. variation of accumulated temperature zone in Heilongjiang Province under SRES A2
Shi, Z., Liang, Z., Yang, Y., Guo, Y., 2015. Status and prospect of agricultural remote /B2 scenarios. Chin. J. Agrometeorol. 35, 485–491. https://doi.org/10.3969/j.issn.
sensing. Trans. Chin. Soc. Agric. Mach. 46, 247–260. https://doi.org/10.6041/j.issn. 1000-6362.2014.05.001. Retrieved from.
1000-1298.2015.02.037. Retrieved from. Zhu, L., Zhu, Q., Luan, J., 2018. The study on the influence factors of urban modern
Smits, P.C., Dellepiane, S.G., Schowengerdt, R.A., 2010. Quality assessment of image agriculture based on C-D production Function-a case study of Hefei City. J.
classification algorithms for land-cover mapping: a review and a proposal for a cost- Heilongjiang Bayi Agric. Univ. 30, 97–101. https://doi.org/10.3969/j.issn.1002-
based approach. Int. J. Remote Sens. 20, 1461–1486. https://doi.org/10.1080/ 2090.2018.06.016. Retrieved from.
014311699212560. Retrieved from.

11