African Journal of Agricultural Research Vol. 6(17), pp.

4025-4033, 5 September, 2011

Available online at http://www.academicjournals.org/AJAR
DOI: 10.5897/AJAR11.770
ISSN 1991-637X ©2011 Academic Journals

Full Length Research Paper

MODIS data based NDVI Seasonal dynamics in agro-

ecosystems of south bank Hangzhouwan bay
Zhao Hangbin1, Yang Xiaoping2* and Li Jialin2
1
College of Information, South China Agricultural University, Guangzhou 510642, Guangdong, China.
2
Coastal Resources and Environment Research Center, Ningbo University, Ningbo 315211, China.
Accepted 9 August, 2011

Normalized difference vegetation index (NDVI) is an important index characterizing the growth
dynamics of plants. During the growth process of plants, NDVI value gradually increases, reaching its
maximum at a certain stage of its growth period and then gradually dwindle. The variation curve of
NDVI varies for different plants. Due to its high time resolution, Moderate Resolution Imaging
Spectrometer–Normalized difference vegetation index (MODIS-NDVI) data can be applied to the study
on the NDVI’s variation of agro-ecosystem. The analysis on the NDVI’s variation characteristics in
different cropping systems of the agro-ecosystem of south bank Hangzhouwan Bay (Southeast
Shanghai, China with north latitude of 30°02′ to 30°24′ and east longitude of 121°02′ to 121°42′) was
conducted by employing the field investigation data and remote sensing data of 23 time sequences
synthesized by MODIS-NDVI 16d in 2002. The results indicate that NDVI curves of different cropping
systems are characterized by specific variation law. NDVI feature model extracted from different
cropping systems based on MODIS-NDVI data can be used for the analysis of the cropping system of
the agro-ecosystem, providing basic data for remote sensing yield estimation.

Key words: MODIS – NDVI, agro-ecosystem, cropping system, seasonal dynamics.

INTRODUCTION

Vegetation is an integral part of the terrestrial ecosystem, vegetation’s growth situation and phenological monitoring
with its significant impact on the climate, hydrology and (Pei and Yang, 2000; Lu et al., 2002; Zhang et al., 2004;
biochemical process of the entire earth system. Satellite Song and Zhao, 2004), the correlation analysis of NDVI
remote sensing, which is independent from the and climatic factors(Li and Shi, 2000; Li and Tao, 2000;
constraints of natural and social conditions, can acquire Tang and Chen, 2003; Mao et al., 2003 ) and the
the observed data of earth vegetation extensively, thus monitoring of natural disasters (Zhou, 1998), etc.
becoming an effective approach for studying earth Since agro-ecosystem is the basic food source for the
vegetation and its variation. Normalized difference sustenance of human being, the monitoring of the growth
vegetation index (NDVI) is an important index for the dynamics of the crops and the accurate estimation of the
characterization of earth surface vegetation (Shi et al., yield is of great significance for maintaining national food
2000), and its temporal variation is of great significance security and social stability. As a consequence, the
for revealing the evolution of earth system on regional macroscopic monitoring of the growth situation of crops
short scale (Li and Shi, 2000). Therefore, researches and yield estimation using NDVI and ground monitoring
related to NDVI have become the hot spot in the study on data have also undergone rapid progress. In 1974, US
earth vegetation. Existing researches are mainly focused department of agriculture, national oceanic and atmos-
on vegetation division on varying scales (Sheng et al., pheric administration, and the commercial department of
1995; Li and Shi, 1999). Analysis of NDVI characteristics national aeronautics and space administration (NASA)
of vegetation and their seasonal variation (Pan et al., together formulated the plan for LACIE (large area crop
2000; Gao and Liu, 2000; Sun et al., 2003), the inventory experiment), which turned out to be successful.
China’s remote sensing yield estimation of crops has also
entered the stage of practical operation (Jiang et al.,
1999; Huang et al., 2003; Yang et al., 2004). The
*Corresponding author. E-mail: yxpnb@126.com. prerequisite for remote sensing yield estimation of crops
4026 Afr. J. Agric. Res.

is the acquisition of information concerning the cropping with spectrum, have 36 spectral channels, distributed in the region
system of agro-ecosystem and the planting areas of each of electromagnetic spectrum of 0.4-14 m. The ground resolution of
crop, and the method mainly adopted is to obtain the this apparatus is 250, 500 and 1000m, with a scanning width of
2330 km, and it can acquire observation data of terrestrial synthetic
planting percentage of each crop based on ground information of 36 global spectral bands twice daily (Liu and Ge,
agricultural sampling. Then the total planting area of each 2000) except the equator. Compared with AHVRR, MODIS is only
crop is calculated from regional farmland area (Yang et able to acquire NDVI information from limited wave bands, thus
al., 2004; Wu, 2000). In this paper, MODIS-NDVI data excluding moisture absorption area of infrared wave band which is
with high temporal resolution is used to study the more sensitive to the absorption of chlorophyll (Wang et al., 2003).
In order to counteract the adverse influence of cloud, atmosphere
seasonal variation characteristics of vegetation index of and non-satellite zenith angle observation. Constraint view-
principle cropping systems of the agro-ecosystem of maximum value compositing (CV-MVC) for 16d day-by-day image
south bank Hangzhouwan Bay, with the attempt to offer was adopted in this paper in MODIS NDVI, to obtain NDVI value of
new idea for the planting area estimation of each crop in vegetation (Wang et al., 2003). With the synthetic data of MODIS-
remote sensing yield estimation of regional agro- NDVI 16d of the research area in 2002, the temporal resolution of
ecosystem. the data was 1 km × 1 km, with 23 time sequences included.
The field investigation of current land use was conducted in
March, June, August and November of 2002. Six principle planting
patterns, that is, vegetables and cotton, cotton, double-cropping
THE STUDY AREA rice, vegetable and late rice, vegetables and Pyrus pyrifolianakai of
agro-ecosystem of south bank Hangzhouwan Bay were selected,
South bank Hangzhouwan Bay (referred in particular to as shown in Table 1. For each planting pattern, 5 to 8 monitoring
points were chosen for the study of seasonal variation of NDVI, with
Cixi City of Ningbo in this paper) is located at the
the monitoring points distributed at areas with large planting areas
northeast of Zhejiang Province of the middle part of and little topographic undulation in all kinds of planting pattern for
China’s eastern coastal area, or on the southeast of better matching with MODIS data with intermediate resolution. In
Shanghai, and outside Qiantang Estuary. Lying between addition, field interviews of 50 different locations were carried out in
north latitude of 30°02′ to 30°24′ and east longitude of December 2002, for the purpose of collection information of the
121°02′ to 121°42′, it has a total land area of 1154 km 2, planting pattern of each location from the peasant households to
confirm the NDVI feature model of each cropping system. Jaguar
with a total population of 1008.4 thousand. The land area
12-frequency handheld GPS produced by GARMIN Company was
per capita is merely 0.114 hm 2, less than 1/6 of national used in the interviews at the monitoring points and field interview
average. points for geographic coordinate positioning.
The climate of south bank Hangzhouwan Bay is north Open MODIS-NDVI data of the research area in ENVI 4.0. Since
subtropical south rim monsoon climate. The hills, plains, the data had geographical coordinates themselves, NDVI values of
intertidal zones and oceans constitute a stepped 23 time 23 time sequences could then be read by inputting the
coordinates of the monitoring points. As NDVI value of each time
topography along the direction from north to south. The interval is the synthesized product during the 16 days, these values
zonal vegetation belongs to subtropical evergreen broad- can be transferred into monthly average NDVI values using the
leaved forest, and the original vegetation of hills of following equation:
Cuiping Mountain in the southeastern region which has
basically disappeared is substituted with secondary forest NDVI j a NDVI (i) b NDVI (i 16) c NDVI (i 32) (1)
and man-made forest. The natural vegetation of the
middle of Binhai sedimentary plain is predominantly
herbal, and the crops artificially cultivated are mainly Where NDVI j is the monthly average of NDVI, and
composed of rice, cotton, soybeans, rape, and j 1, 2, ,11,12 month; NDVI is the MODIS-NDVI 16d
vegetables. The muddy intertidal zones in the north have
a broad beach face, with the vegetation dominated by
product containing part of the days of jth month; i is the Number
Spartina Alterniflora, reed and Suaeda community. Within for MODIS-NDVI 16d products, and i 001, 017, ,305,321 ;
this region, the weather is cool in summer and warm in a , b and c is the ratio of the number of days in jth month to the
winter, and the annual sunshine duration and total solar total number of days in jth month of i th, i 16 th and i 32 th
2
radiation is 2038.4h and 112 kilocalorie/cm respectively product, respectively.
with an average precipitation of 1272.8 mm annually. The NDVI value of each monitoring point was calculated according to
temporal and spatial distribution of sunshine, heat and the above equation, and statistical analysis was then conducted on
water are relatively good, and the climate is warm and the average value of NDVI, standard deviation and coefficient of
humid, with high light-temperature efficiency, abundant variation of each cropping system (Table 2). Analysis showed that
coefficient of variation of monthly average NDVI of each cropping
rainfall and long period for crop growth. system were mostly below 10.0%, indicating that monthly NDVI
value of each monitoring point had relatively good
representativeness and that taking the monthly average of NDVI
MATERIALS AND METHODS value of each monitoring point as the average level of a specific
cropping system was feasible. Accordingly, variation curve of
MODIS-NDVI data came from MODIS sensor of TERRA satellite of annual NDVI of each cropping system was obtained, with the
America’s earth observation system (EOS). This sensor, as the results shown in Figure 1. Meanwhile, in order to further analyze
new-generation optical remote sensing apparatus combining image the variation characteristics of NDVI of each cropping system, NDVI
 Hangbin et al. 4027

Table 1. Cropping systems of agro-ecosystem of south bank Hangzhouwan bay.

Cropping system Types of land use Basic information

Vegetable and cotton The period for cotton planting is from middle and late April to middle and late November, and the remaining period
Dry land
pattern is for vegetable planting.

Middle and late April is the period for planting, and middle November is the period for harvest, with no crop planting
Cotton planting pattern Dry land
during the remaining period.

Double cropping rice Early rise is transplanted in the late April, and is harvested in late July; late rice is transplanted in the late July, and
Paddy field
pattern is harvest in the middle and late November.

Vegetable and late rice The period from late July to middle November is for planting late rice, and the remaining period is for planting
Paddy field
pattern vegetables.

Vegetable pattern Dry land and paddy field Vegetables are planted all the year round, and greenhouse is used in winter.

Pyrus pyrifolianakai
Dry land and paddy field Early March is the time for sprouting; early April is the time for blossoming; middle August is the time for ripening.
pattern

Table 2. Statistical characteristics of monthly average NDVI for variety cropping system.

Month
Cropping system
1 2 3 4 5 6 7 8 9 10 11 12
av. 38.71 41.42 45.23 47.86 40.98 50.10 61.69 64.01 67.91 52.57 38.69 41.67
Vegetable and cotton pattern Std. 2.41 3.12 3.01 3.61 4.35 4.30 3.71 3.20 3.46 2.13 2.12 3.04
cv. (%) 6.23 7.53 6.66 7.54 10.61 8.58 6.01 5.00 5.10 4.05 5.48 7.30

av. 30.92 29.02 31.58 32.33 38.62 48.10 59.19 66.88 70.01 50.57 35.81 31.24
Cotton planting pattern Std. 3.41 3.07 2.96 4.30 2.16 3.07 3.41 4.01 5.16 4.05 2.37 3.01
cv. (%) 11.03 10.58 9.37 13.3 5.59 6.38 5.76 6.00 7.37 8.01 6.62 9.64

av. 38.74 41.98 44.28 45.7 50.36 67.9 36.45 72.22 76.54 64.94 51.12 46.88
Double cropping rice pattern
Std. 3.41 3.24 2.45 3.36 3.64 3.18 4.31 5.01 4.94 3.45 3.21 3.11
cv. (%) 8.80 7.72 5.53 7.35 7.23 4.68 11.82 6.94 6.45 5.31 6.28 6.63

av. 41.10 42.00 43.03 45.02 47.30 48.1 40.21 74.02 74.54 66.94 49.72 49.24
Vegetable and late rice pattern
Std. 3.65 3.87 3.24 3.15 3.24 4.01 3.05 4.76 4.30 4.01 3.99 3.06
4028 Afr. J. Agric. Res.

Table 2. Contd.

cv. (%) 8.88 9.21 7.53 7.00 6.85 8.34 7.59 6.43 5.77 5.99 8.03 6.21

av. 40.04 42.84 42.68 41.83 48.78 46.15 45.96 43.84 45.14 46.64 45.00 42.01
Vegetable pattern Std. 3.46 3.45 3.22 3.55 2.16 1.43 3.65 3.76 3.14 4.41 2.67 3.15
cv. (%) 8.64 8.05 7.54 8.49 4.43 3.10 7.94 8.58 6.96 9.46 5.93 7.50

av. 38.87 35.33 36.67 45.69 47.51 49.05 60.97 66.3 63.48 57.15 44.28 40.09
Pyrus pyrifolianakai pattern Std. 2.78 2.96 2.97 2.56 3.45 3.45 3.24 3.64 3.57 3.04 3.21 2.34
cv. (%) 7.15 8.38 8.10 5.60 7.26 7.03 5.31 5.49 5.62 5.32 7.25 5.84

Table 3. Grey relational analysis for NDVI curve of ground survey sample and variety cropping system.

Cropping system 1 2 3 4 5 6 7 8 9
Vegetable and cotton pattern 0.923 0.897 0.937 0.879 0.946 0.967 0.456 0.903 - 0.864(0.922)
Cotton planting pattern 0.941 0.931 0.889 0.934 0.921 0.881 0.951 0.926 - 0.92175
Double cropping rice pattern 0.874 0.961 0.944 0.923 0.899 0.907 0.964 0.937 - 0.92613
Vegetable and late rice pattern 0.917 0.934 0.881 0.864 0.861 0.941 0.951 0.927 - 0.9095
Vegetable pattern 0.924 0.916 0.937 0.929 0.869 0.893 0.926 0.931 0.907 0.91467
Pyrus pyrifolianakai pattern 0.911 0.974 0.437 0.419 0.891 0.934 0.929 0.913 0.931 0.815(0.926)
The data in parenthesis are the correlation coefficient excluding the influence of three samples which exit some error.

of each cropping system in January was taken as the cotton planting, with the sowing area accounting and late April at south bank Hangzhouwan Bay,
reference point, and the variation monthly NDVI value was for 1/3 of Zhejiang Province, creating a record and it takes around 120 days to proceed from
calculate as shown in Figure 2.
high yield of lint cotton per area. Since 1980s, the sowing stage, seedling stage, budding stage to
cotton planting area at south bank Hangzhouwan blooming stage, and this period is called growth
Bay gradually decreased with the adjustment of period. Then it comes to boll opening stage for
RESULTS
agricultural planting structure, and only part of the cotton in late August, which lasts until middle
Variation characteristics of NDVI of each high-yield dry land was reserved for cotton November when the pick is over. During these five
cropping system planting. Cotton undergoes five stages with growth stages, NDVI which reflects the growth
different nature from seed germination and fully situation of cotton displays corresponding
Cotton planting pattern grown that is sowing stage, seedling stage, variation law. It can be known from Figure 1 that
budding stage, blooming stage and boll opening the variation curve of NDVI of agro-ecosystem
It has not been very long since south bank stage. The period from sowing to picking is called with only cotton planted is the unimodal curve in
Hangzhouwan Bay was reclaimed, an extensive the entire growth period which lasts for about 210 the whole year. From January to early April, no
range of reclamation land was once used for days. The cotton is generally planted in middle crops were sowed in the field generally, resulting
 Hangbin et al. 4029

80
70
60
50
NDVI

40
30 Veget abl e ＆
Vegetable andcot t on
cotton cot t on
Cotton
Double cropping rice Vegetable and late rice
Doubl e cr oppi ng r i ce Veget pyrifolianakai
abl e ＆ l at e r i ce
20 Vegetable
Veget abl e
Pyrus
Pyr us pyr i f ol i anakai
10
1 2 3 4 5 6 7 8 9 10 11 12
Month
mont h
Figure 1. Variation curve of NDVI of each cropping system.

40
30
The change of NDVI

20
10
0
- 10 cot t on Veget abl e ＆
Cotton Vegetable andcot t on
cotton
- 20 Doubl e crcropping
Double oppi ng rrice
i ce Veget abl e ＆
Vegetable andl at e rrice
late i ce
- 30 Veget abl e
Vegetable Pyrus
Pyr pyrofolianakai
us pyr i f ol i anakai
- 40
1 2 3 4 5 6 7 8 9 10 11 12
Month
mont h
Figure 2. Monthly change of NDVI for variety cropping system.

in a very small NDVI value which was usually 30. Starting the reflectance value of red waveband channel of the
from middle and late April when cotton sowing began, senor decreased, while that of infrared waveband
NDVI gradually increased. Seedling stage of the cotton gradually decreased, resulting in the reduction of NDVI
started from May, and NDVI value rapidly increased as value. In middle and late November when cotton was
the cotton seedling grew. June and July were the months harvested and the cotton straw was removed from the
for budding, and NDVI further increased. With the cotton field, no vegetation covered the cotton field at this
approach of blooming stage in August, NDVI value time. Therefore, the information received by MODIS was
reached its maximum from middle and late August to basically the spectral information of the soil, with NDVI
early September, being 70. Consistent with the increase plummeting to the low level at the beginning of spring. It
in NDVI value, the monthly variation of NDVI value from can be known from Figure 2 that from October to
May to September were all positive, with the increment of December, the monthly variation of NDVI value was all
July being the maximum of 11.09, as shown in Figure 2. negative, with the reduction of 19.4, 14.8 and 4.6,
Although NDVI value reached its maximum in respectively.
September, the monthly increment of NDVI value was the
minimum of 3.13 during this period. After the cotton
entered the boll opening stage, the leaves yellowed, Vegetable and cotton pattern
leave pore closed, and the absorption ability of
chlorophyll weakened correspondingly. Consequently, Vegetable and cotton pattern was mainly distributed in
4030 Afr. J. Agric. Res.

dry land with long time of soil formation and good fertility. late rice. Since no vegetation covered the field after the
During slack season without cotton planting, vegetables reaping of early rice and small biomass of late rice
were usually planted. The variation process of NDVI of seedling transplanted, an apparent low ebb of 36
this type of agro-ecosystem was basically consistent with occurred in July. With the tillering stage and jointing-
that of cotton pattern during the cotton planting period. booting stage of late rice, NDVI value of August kept on
Cotton seedling stage started from May, and all the way increasing, reaching the second peak value of around 76
through the budding stage in June and July, to boll at the tasseling and blooming stage in September. With
opening stage in August and then to early September, the ripening of rice ear, the leaves yellowed and NDVI
NDVI value constantly increased (Figure1). During this value decreased correspondingly. After the reaping of
period, the monthly variation of NDVI value was all rice at the end of November, NDVI plunged to the low
positive, reaching its maximum of 11.6 in July. In October ebb for the second time. Seen from the monthly variation
and November, NDVI value gradually decreased, with the of NDVI value, during the growth period of early rice,
monthly reduction of NDVI value reaching its maximum of monthly variation of NDVI value from April to June was
15.3 in October (Figure 2). While during the period from positive, increment reaching the maximum of 17.5 in
December to April of the following year, NDVI value of June. NDVI value of July significantly reduced, with the
this pattern was apparently higher that that of cotton reduction of 31.4. During the growth period of late rice in
pattern because of vegetable planting, being over 40 August and September, the monthly variation of NDVI
normally (Figure 1). The monthly variation of NDVI value was all positive, with the increment reaching the
during the period from December to April of the following maximum of 35.8 in August. The monthly variation of
year was positive, ranging between 2 and 4; that in May NDVI value from October to December was all negative,
was negative, with a reduction of 6.9 (Figure 2), as a reduction reaching its maximum of 13.8 in November.
result of the fact that cotton seedlings were just
transplanted after the harvest of vegetables in May,
which naturally caused the smaller NDVI value than that Vegetable and late rice pattern
in April before the vegetable harvest.
Under the influence of adjustment of planting structure as
well as supply and demand relation of rice market
Double cropping rice pattern (structural surplus of rice), the sowing area of food crops
tremendously reduced from 1980’s. From 1987 to 2000,
Double cropping rice pattern is the traditional cropping the proportion of the planting area of food crop in the
system for paddy fields at south bank Hangzhouwan Bay, entire sowing area of crops dropped from 44.34 to
and an extensive planting area is still retained at present. 36.78%, a reduction of 7.56% point in 13 years, while as
Early rice is usually transplanted in middle April, and it a major aspect of the adjustment of the internal structure
enters tillering stage in early May, jointing-booting stage of food crops, the proportion of planting area of grains
in late May, tasseling-blooming stage in middle June, and decreased from 34.09 to 16.12%, that of early rice
ripening-reaping stage in early and middle July. Late rice accounting for the greatest portion in the decrease of
is transplanted right after the reaping of early rice in late planting area of rice. Therefore, vegetable and late rice
July, and it enters tillering stage in middle August, pattern has taken over to become the principle planting
jointing-booting stage in middle and late September, pattern of ecosystem of paddy fields at south bank
tasseling and blooming in middle and late October, and Hangzhouwan Bay. The variation process of NDVI value
ripening-reaping in middle November. After the reaping of of vegetable and late rice pattern during the planting of
late rice, some fields are planted with green manure or late rice was similar to that of double cropping rice
vegetable, while other fields are left desolated until the pattern, with large NDVI value in August and September,
cultivation of early rice in April of the following year. and a first peak value of 74.6 was reached in September.
During the growth process of double cropping rice, NDVI After this, NDVI value gradually decreased, monthly
curve reflecting the growth situation of rice also displayed reduction reaching its maximum of 17.2 in November.
corresponding variation law. From December to July of the following year, due to the
It can be known from Figure 1 that the variation vegetable planting in the fields, its NDVI value began to
process of NDVI value of double cropping rice pattern climb slowly but steadily from 41, reaching 48.1 in June.
displayed apparent bimodal curve. From January to April, The harvest of vegetables and planting of late rice in July
NDVI value was maintained at around 40. With the led to the lowest value of 40.2 in the entire year, with the
transplant of early rice seedlings in middle April, NDVI monthly reduction of 7.9 (Figures 1 and 2).
value of paddy field gradually increased. As the early rice
entered tillering stage and jointing-booting stage in May,
NDVI value rose sharply, reaching the first peak value of Vegetable pattern
68 at tasseling-blooming stage in June. July was the
month for the reaping of early rice and the sowing of With the construction of Cixi national agricultural science
 Hangbin et al. 4031

and technology park, facility agriculture and export- positive during its growth period (March to August),
oriented agriculture at south bank Hangzhouwan Bay are reaching the peak value of 11.9 in July. The monthly
developing at a rapid speed, with export-oriented variation of NDVI value was negative from September to
vegetable planting being an integral aspect of Cixi February of the following year, with the maximum
national agricultural science and technology park. With reduction of 12.9 during its defoliation in November
Changhe Town where vegetable production base with an (Figure 2).
area of 8k m2 is located being the core area, the entire
demonstration area for vegetable planting extends nearly
70 km2. Large areas of dry land and paddy fields have Confirmation of variation characteristics of NDVI of
been converted to vegetable planting, main crop varieties each cropping system
include broad beans and green soybean and other In order to confirm the applicability of variation
species of vegetable legumes, black maize and waxy characteristics of NDVI in judging different cropping
corn and other species of vegetable corns, cherry tomato system, the variation curves of NDVI value at each point
and Japanese wrinkle melon and other species of in different cropping systems obtained via field interview
solanberries, kale and broccoli and other species of leaf were taken as samples, and discriminant analysis was
vegetables and fruit vegetables, radish and tuber mustard conducted using grey correlation analysis in grey system
and other species of root vegetables, as well as green theory, with the basic steps listed as follows:
Chinese onion and Welsh onion and other species of
shallots. As a result of the modification of planting (a) Range transformation of raw data. The objective of
technology, introduction of high-quality varieties and the conducting range transformation is to eliminate the
promotion of greenhouse, the vegetable planting pattern dimension of raw data so as to make it comparable. The
can basically meet the requirement of an evergreen field. mathematical formula for range transformation is as
Therefore, the yearly variation process of NDVI value follows:
was a rather smooth curve, with the average maintaining
between 40 and 48 (Figure 1). In terms of the monthly X ij X j min
variation of NDVI value, the increment of May when X ' ij (2)
vegetables grew most vigorously reaching as high as 8, X J max X j min
while those in other months were below 3 (Figure 2)
(b) Determine master sequence X 0 and sub-sequence
Pyrus pyrifolianakai pattern X i . Variation curve of NDVI of each cropping system is
Ningbo has long been famous as China’s hometown of taken as master sequence as in Figure 1 and the sub-
pyrus pyrifolianakai, which enjoys high reputation at sequence is determined with the following method: the
home and abroad for its good quality and is an integral point of each cropping system obtained through ground
component of characteristic fruit production of Cixi investigation were synthesized into NDVI data with
national agricultural science and technology park. With its corresponding 16d MODIS-NDVI data acquired. Then,
planting area already reaching 30 km2, high-quality pyrus monthly variation curve of NDVI value of each point
pyrifolianakai can be found mainly in the piedmont paddy calculated according to Formula (1) is then taken as the
fields and dry lands of towns such as Henghe, Zhouxiang sub-sequence of the variation curve of NDVI value of the
and Hangzhouwan Bay, etc. Pyrus pyrifolianakai corresponding cropping system.
generally begins to sprout in early March, blooms in early (c) Calculate the absolute value 0i (t j ) of the difference
April, fruits in middle August, and defoliates in winter.
between master sequence and sub-sequence, that is:
Since the park for pyrus pyrifolianakai was only built not
long ago and the seedlings were greatly spaced.
Vegetables are usually inter-planted between pyrus 0i (t j ) X 0 (t j ) X i (t j ) (3)
pyrifolianakai of relative young tree age. The variation
curve of NDVI value of pyrus pyrifolianakai planting (d) Take the maximum max and minimum from
min
pattern is unimodal curve, and during its defoliation, its
NDVI value was less than 40. After March, with the 0i (t j ) .
sprout of tree leaves, its NDVI value gradually increased, (e) Calculate the correlation coefficient between master
reaching the maximum of about 66 in August when the sequence X 0 and sub-sequence X i . The calculation
fruits ripened. From September on, NDVI value gradually
decreased with the decrease of chlorophyll content of the formula is as follows:
leaves; after November when leaves started to defoliate,
NDVI further dwindle, reaching the level of one year ago min max
in December (Figure 1). In terms of monthly variation of 0i (t j ) (4)
0i (t j ) max
NDVI value, the monthly variation of NDVI value was
4032 Afr. J. Agric. Res.

Where is resolution coefficient, with its value generally Conclusion

selected as 0.5; since the data in correlation analysis
intersect with each other after the data transformation, Since only a small number of monitoring points were
involved for each cropping system in this study and only
the value of min is generally selected as 0. the variation characteristics of NDVI’s monthly average
(f) Obtain the correlation degree from correlation was analyzed, the determination of cropping system
coefficient. The calculation formula is as follows: might have been affected to some extent. Meanwhile,
due to the limited number of field interview locations for
1 n confirmation, its precision remains for further testing. In
0i 0i (t j ) (5) addition, the issue of mixed pixel induced by resolution of
n i 1 NDVI data was not considered in this research, which
might have some impact on the coastal areas with
The results of grey correlation analysis showed that fragmented land use, thus it needs further discussion in
except that the correlation coefficient of Sample No. 7 of subsequent studies.
cotton planting pattern, Sample No.3 and Sample No. 4
of pyrus pyrifolianakai planting pattern was below 0.5, ACKNOWLEDGEMENTS
those of all the remaining samples were uniformly high,
and passed the significance test when α = 0.05 (Table 3). This study was supported by national natural science
In terms of correlation degree index, under the influence foundation of China (No. 41171073), Natural Science
of one and two samples with smaller correlation Foundation of Zhejiang Province, China (No.Y5110321),
coefficient on cotton planting pattern and pyrus science and technology planning project of Zhejiang
pyrifolianakai planting pattern, respectively, their Province, China (2010C33155), social science foundation
correlation degree were 0.864 and 0.815, respectively, of Zhejiang Province, China (10JDHY02YB), natural
with those of remaining planting pattern above 0.91. science foundation of Ningbo, China (No. 2010A610105)
Excluding the influence of these three samples, the and K. C. Wong Magna Fund in Ningbo University.
correlation degree of cotton planting pattern and pyrus
pyrifolianakai planting pattern was 0.922 and 0.926, REFERENCES
respectively. Therefore, it can be concluded that the
monthly variation curve of NDVI value of each cropping Shi PJ, Gong P, Li XB (2000). Land use/ Method and practice of study
on plants covering change. Beijing: Science Press, pp. 36-66.
system obtained in this study is of universal significance Li XB, Shi PJ (2000). Sensitivity analysis of NDVI dynamic and air
for south bank Hangzhouwan Bay, and this curve can be temperature, precipitation changes of Chinese vegetation types.
used to determine the corresponding planting pattern. China J. Plant Ecol., 24(3): 379-382.
Sheng YW, Chen WY, Xiao QG and Guo L (1995). Macro-classification
of Chinese vegetation by using meteorological satellites. J. Chinese
Sci. Bull., 40(1): 68-71.
DISCUSSION Li XB, Shi PJ (1999). Study on NDVI changes rule of Chinese main
vegetation types based on NOAA/ AVHRR. J. Chinese Bull. Bot., 41
(3):314-324.
The study on the seasonal variation of vegetation of Pan YZ, Li XB, He CY (2000). General classification study on Chinese
principle agro-ecosystem of south bank Hangzhouwan vegetation based on NOAA/AVHRR &s Holdridge PE. J. Quat. Res.,
Bay using MODIS-NDVI data indicates that yearly 20(3): 270-280.
Gao ZQ, Liu JY (2000). Drive-factor analysis and model study on
variation curve of NDVI value displays different Chinese vegetation index changes based on remote sensing and
characteristics for different cropping system, which can GIS. J. Clim. Environ. Res., 5(2): 155-164.
be reflected with variation curve of NDVI value. Grey Sun L, Guo QX, Wang XC, Zhou XF (2003). NDVI change analysis of
correlation analysis reveals that this specific curve is of vegetation types in the middle and southern tract of Eastern China
northern-southern. Chinese J. Appl. Environ. Biol., 9(5): 449-454.
high precision in determining the cropping system o Pei ZY, Yang BJ (2000). Extracting spatial-temporal features and
principle agro-ecosystem of south bank Hangzhouwan designing plants’ growth vigour model of MTVI-NDVI. Int. J. Agric.
Bay. Thus, it can provide new idea for the analysis of Biol. Eng., 16(5): 20-22.
cropping system of agro-ecosystem and the estimation of Lu L, Li X, Cheng GD (2002). Analysis seasonal features of the Heihe
river Basin by using NOAA/ AVHRR vegetation index data. J. Desert
the area of each crop in remote sensing yield estimation. Res., 22(2): 187-191.
However, due to the discrepancies in light and conditions Zhang F, Wu BF, Liu CL, Luo ZM (2004). Study on monitoring plant
of different regions and climates, certain differences phenological methods by using vegetation index in sequence. J.
inevitably exist in the growth process of crops. Therefore, Trans. Chinese Soc. Agric. Eng., 20(1): 155-159.
Song XN, Zhao YS (2004). Study on extraction of plants’ temperature &
the establishment of monthly variation curve of NDVI moisture composite index derived from MODIS data. J. Geogr. Geo-
value of a specific agro-ecosystem should be in Inf. Sci., 20(2): 13-17.
compliance with the actual conditions of corresponding Li BG, Tao S (2000). Correlation analysis of AVHRR-NDVI and climatic
regions, in order to obtain the pattern for determining the factors. J. Acta Ecol. Sin., 20(5): 898-902.
Tang HP, Chen YF (2003). Relation between seasonal change and
cropping system and to perform the estimation of planting climatic factors of NDVI in Chinese southern-eastern tract. J. Quat.
areas. Res., 23(3): 318-325.
 Hangbin et al. 4033

Mao XS, Zhang YQ, Shen YJ (2003). Pre-test on vegetation index Wu BF (2000). Functioning RS methods on national farming information
change & impact factor of winter wheat. Chinese J. Eco-Agric., 11(2): monitoring and assessing. J. Acta Geogr. Sin., 55(1):25-35.
35-36. Liu C, Ge CH (2000). Characteristics and application of RS data of
Zhou YM (1998). Application of NOAA/AVHRR data in drought EOS-MODIS of USA. J. Remote Sens. Inf., 3:45-48.
inspection. J. Quart. J. Appl. Meteorol., 9(4): 496-500. Wang ZX, Liu C, Huete A (2003). The progress of study on plants index:
Jiang D Wang NB, Yang XH (1999). Study on Chinese grain crops from AVHRR-NDVI to MODIS-EVI. J. Acta Ecol. Sin., 23(5):979-987.
assessment based on RS. Chinese J. Nature. 21(6):351-355.
Huang JF, Wang RC, Jiang HX,Yang ZE (2003).The best time-phase
choice of Zhejiang Province rice RS-assessment based on GIS. J.
Chinese J. Appl. Ecol., 13(23): 290-294.
Yang XH, Zhang XP, Jiang D (2004). Methods on multi-crops cultivated
area extracted by MODIS sequential NDVI feature. J. Resour. Sci.,
26(6): 17-22.