LANDSCAPE’S CHANGE DETECTION
IN THE SOUTHERN SLOPE OF THE QINLING MOUNTAINS, CHINA
Xuehua LIU *, Qiang JIN, Jihong LI, Shuang ZHANG
Department of Environmental Science and Engineering, Tsinghua University, Beijing 100084, China —
xuehua-hjx@tsinghua.edu.cn; (jingiang99, lijh02, zhangshuang03)@mails.tsinghua.edu.en
PS ICWG III/IV
KEY WORDS: Change detection, Landscape, Land use / cover (LUC), NDVI, Human activities, The Qinling Mountains, Giant
panda nature reserves (NRs).
ABSTRACT:
The landscape on the southern slope of the Qinling Mountains, the home of many wildlife species including the giant pandas, golden
takin and golden monkey, has been logged and influenced for a long term. This caused the obvious decreasing of the wildlife’s
habitat and has produced strong impacts on their survival and distribution. It is necessary for managers to have a quick and accurate
method to detect the landscape changes, which can help them to be aware the changing situation. Consequently, managers could take
measures to keep controlling the changes. Change detection through remote sensing has now been applied widely in terms of its
quick processes and accurate results. However, it has not been well used in detecting changes of landscape in giant panda nature
reserves in China up to now. This article focuses on the landscape’s change detection in Changqing, Foping and Guanyinshan NRs,
containing LUC change and NDVI change through spatial analyzing TM images acquired on September 15 1988 and September 8
1997. Two LUC maps from two different years have overall accuracies of 73.74% and 75.19% and kappa values of 0.6829 and
0.6968. À LUC changing map was created and show that the exact areas of “conifer”, “deciduous broadleaf”, and “mixed conifer-
broadleaf” forests have greater changing, however, three small-area LUCs (i.e., *farmland and settlement", *water", *rock and bare-
land") have larger changing rates. NDVI analysis of two years? images shows an obvious change with higher NDVI in 1988 than in
1997 in total. Most of changes are located in the regions along or near the rivers and reserves! boundaries, which reflects
substantially that human being has strong impacts on the surrounding landscape.
1. INTRODUCTION
The southern slope of the Qinling Mountains is not only one of
the important habitats for giant pandas in China, but also the
home for many other rare and endangered species. The
completion and diversity of the entire ecosystem is essential for
conserving those rare and endangered species and maintaining
balance of the ecosystem in the Qinling Mountains. However,
the vegetations, which form the home of those rare and
endangered species, on the southern slope of the Qinling
Mountains has been experiencing various extends of logging
and land use converting for a long term. This caused the
obvious decreasing of the area of wildlife's habitat and has
produced strong influences on wildlife's survival and
distribution.
In order to better handle the conflicts between the conservation
and develop, control the illegal forest cutting and habitat
degrading activities, and follow a sustainable approach, it is
necessary for managers to grasper the dynamic information of
wildlife species, their habitat and human activities inside and
surrounding the NRs. The remote mountain area with dense
forest cover brings imaginable difficulties in collecting all kinds
of information, without mentioning in-time collecting, and
makes a completion of the ground survey a long period. The
traditional manual recording and handling the collected
information is a cost and time-consuming task and hard to in-
time update information. Furthermore, without spatial
" Corresponding author. Tel: -86-10-62785610 ext 14; Fax:
Email: xuehua-hjx@tsinghua.edu.cn
information and analysis, it is not easy to make clear
comparisons among nature reserves in terms of their wildlife,
habitats and human impacts. Consequently, it causes managers
difficulties to make decisions due to lacking of strong
supporting information and evidence (Gao and Wu 2000).
Change detection through remote sensing has now been applied
widely because of its quick analysis processes, accurate results
and visual spatial information (Zhang 2003). LUC classification
and mapping for various-period images can be used to detect
the LUC type change (van Lynden and Mantel 2001), while
NDVI calculation and mapping for various-period images can
be applied to detect the change of vegetation quality which has
been applied in vegetation coverage assessment, crop yield
estimation and crop identification (Tian and Min 1998). Due to
that such a research has never been done in the home of giant
pandas on the southern slope of the Qinling Mountains, China,
this article applied remote sensing approach to detect landscape
change (i.e. LUC type change and NDVI value change) in
three giant panda NRs based on two TM images acquired in
1988 and 1997. The aims are (1) to illustrate the quick and
accurate remote sensing approach to the reserve managers, (2)
to provide the landscape change information to the managers
and other groups of people.
+86-10-62785687;
Inte
3.1
This
rese
data
TM
199
sam
2
m“.
clou
and
the
wer
fron
3.2
In c
dete
TM:
(Zh:
simi
info
info
200!
rese
and
(Sy:
and
