977
7. Beijing 2008
nsing of urban
370-384
Institute.
Kesinl.htm
ities. American
Sciences.
ogt.html
ote Sensing of
6 (2003) 370 -
irvious. surface
emote Sensing
EXTRACTING CROPPING INDEX VARIATIONS IN NORTHERN CHINA BASED ON
NDVI TIME-SERIES
ZHU Xiaolin, WU Jin, CHEN Jin *
College of Resources Science and Technology, Beijing Normal University, Beijing, 100875, China-(zhuxiaolin,
wujinl985, chenjin)@ires.cn
Commission VII, WG VII/5
KEY WORDS: Remote Sensing; Agriculture; Change Detection; Indicators; Multitemporal; Spatial
ABSTRACT:
Multiple cropping system, characterized by cropping index, is of significant importance to Chinese food production and security.
Owing to the changing nature conditions and human activity, cropping index could show remarkable inter-annual fluctuations,
which in turn reflects arable land use intensity and indicates climate change impacts on agriculture system. NDVI (Normalized
Difference Vegetation Index) time-series is an effective indicator of vegetation status at regional scale. We developed a new method
for extracting cropping index from NDVI time-series, and employed it to extract cropping index and its inter-annual variations of
northern China from 1982 to 2003. The results show: remotely sensed cropping index is high consistent with statistical data at
province scale (J? 2 =0.9213, /><0.001, slope= 1.0775), demonstrating that this method can extract cropping index effectively and
correctly. The average cropping index of northern China increased from 87.27 in 1982 to 115.98 in 2003, with an average change
rate 1.3275 per year(R"=0.7955, P<0.001). The areas displayed different changes of cropping index, with Huang-Huai-Hai drainage
area experiencing a clear cropping index increase and other regions relatively less cropping index change.
1. INTRODUCTION
Multiple cropping system, characterized by cropping index, is
important to Chinese food production, which feeds 22% of the
whole world population by using 8.6% arable land of the whole
world (Liu, 1992; Ma, 2003). Cropping index refers to the times
of sequential crop planting in the same arable land in one year,
usually defined as the ratio of the total seeding area to the
arable land area (Liu, 1993), which reflects the using efficiency
of soil, water, light and other natural resources. Owing to the
changing nature conditions and human activity, cropping index
could show remarkable inter-annual fluctuations, which in turn
reflects the using intensity of arable land and indicates the
impacts of climate change on agriculture system. It is desired to
extract the cropping index and its change information by
remotely sensed data for agriculture sustainable development
and assessing the impacts of climate change on agriculture
system.
Traditionally, cropping index is calculated by statistical data at
local administration unit, which is time-lagged, labor
consuming, poor in creditability, and lack of details of spatial
distribution. On the other hand, remote sensing technology has
been widely applied to agriculture and crop growing status
monitoring (Harris, 2003; Seelan et al., 2003). The
development of remote sensing technology makes it possible to
obtain actual cropping index information efficiently and
reliably. Peak of the NDVI (Normalized Difference Vegetation
Index) time series curve could reveal the time when above
ground biomass of crops reaches the maximum and tone of the
curve fluctuates with the crops growing processes such as
sowing, seeding, heading, ripeness, and harvesting within a year.
Therefore, the cropping index could be defined as the number
of peaks of NDVI time-series per year (Fan and Wu, 2004).
However, since NDVI data are easy to suffer from cloud and the
other poor atmospheric conditions, the curve of NDVI time
series may include much noise, turning out to be small peaks
and valleys in one cycle, which makes it more difficult to
extract the cropping index effectively. Some methods have been
developed to extract cropping index by reconstructing high
quality NDVI curve, such as Harmonic Analysis of Time Series
(HANTS) (Fan and Wu, 2004) and so on (Yan et al., 2005).
Nevertheless, these methods depend too much on prior
knowledge, and can not get rid of all the disturbing noise
effectively.
This study aimed to (1) develop a new method to eliminate the
atmospheric effects and other contaminations and improve
accuracy of cropping index extracting form NDVI time-series
and (2) extract the cropping index and the Cropping Index
Variation (VCI) of 17 provinces of northern China from 1982 to
2003 by using 8km 15-day Maximum Value Composite
NOAA/AVHRR GIMMS NDVI time-series data.
2. STUDY AREA AND DATA
The cropping index exaction was conducted on the 17 provinces
of northern China (Figure. 1), of which the arable land area was
identified according to Chinese vegetation type map.
The Global Inventory Monitoring and Modelling Study
(GIMMS) AVHRR 8 km resolution NDVI 15-day composite
dataset covering the period of 1982 to 2003 (Tucker et al., 2005.
Available at http://gimms.gsfc.nasa.gov/) was used to exact the
cropping index. This dataset was produced by Maximum Value
Composition (MVC) technique which selects the highest value
of NDVI during every 15-day period for each pixel to remove
* Corresponding author: Chen Jin, email: chenjin@ires.cn