1 2004
. From
ocated
1997-
d by 1
ited to
value
od are
pixels.
ost no
onifer
values
1997,
reatly
Tease.
VIOUS
| The
, and
Is of
values
d and
pe of
- and
AB--
rest”
river
> are
1ing,
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004
mushroom producing (Liu 2002), have created long-term
impacts on the surrounding forest environment. The
consequence of forest cutting is the simplification of forest
structure and reduction of species diversity. The preferred
plantation in the valleys and at the lower elevation region is the
deciduous broadleaf species including the economic tree crop
species. Due to that the forest covers most of the study area,
although it has a larger changing area, but the changing rate is
still smaller comparing with the small-proportion LUCs, such as
“farmland and settlement", *water", *rock and bare-land”. LUC
"water" has the largest changing rate because the rainfall in
September 1997 was much higher than the one in September
1988. More surface flow filled into the rivers which made
"water" easy detected. The changing rates of “rock and bare-
land" and "farmland and settlement" are listed the second. Their
changing areas are more concentrated in Guanyingshan NR
because it continues commercial forest cutting from 1986 to
1998. Some patches of “rock and bare-land” are caused by
forest clear-cutting. There are about 60 and 244 local people
living inside Foping and Guanyingshan NRs respectively which
causes the increasing of “farmland and settlement”.
NDVI spatial calculation is also proven as a good approach to
detect LUC quality change. The NDVI pattern of 1997-image
was found great different from the one of 1988-image, which
indicates the environment in the study area has changed not
only on the LUC types in some areas but also on the quality of
the same LUC. The mushroom production happened during
1995-1999 might be the cause of NDVI reduction of deciduous
broadleaf forest at the lower elevation areas. Another possible
reason for NDVI reduction could be the difference of weather
condition between September 1988 and 1997. But we believe it
is human activities are playing a major role on the change of
panda landscape. The important point obtained here is that the
curve of NDVI value and pixel number for different LUCs can
be used to explain which LUC has changed and how much LUC
quality has changed. :
Two methods, i.e. maximum likelihood classification (MLC)
and NDVI calculation, were applied to map the LUC change on
type as well as on quality. Each has its advantages and
disadvantages. MLC used more layers’ data to identify various
LUC types which can be used to detect the change of LUC
types. However, for some LUCS, although they have no change
on their types but they do have change on quality, which MLC
cannot detect this quality change. Another disadvantage of
MLC is that the classification of the past image lakes of ground
truthing data which makes the image treatment more time and
energy consuming (Liu et al. 1999, Wu and Wang 2003, Liu
and Zhang 2004, Fuller et al., 2003). While NDVI calculation
based on the spectral bands and is a fast and easy-handle
approach, it can be used to detect the quality change. This can
compensate MLC and explain more information. NDVI also has
its disadvantage, namely that the plant growing condition can
greatly influence NDVI values, consequently the background
information could be magnified (Tian and Min 1998, Gao et al.
1998). Two methods, if combined for change detection, can
play a very good role on deriving information of detected
changes.
6. CONCLUSION
Two methods successfully detected the landscape changes in
the southern slope of the Qinling Mountains. During the period
of 9 years from 1988 to 1997, the landscape (including LUC
types and also quality through NDVI values) does have a great
deal of changing. Most of changes were detected from the areas
along the river valleys and the southern boundaries of the
reserves where the local people are living or where are easy
accessed by local people. It indicates the human activities are
the main driving force of landscape change in the study area. In
conclusion, forest types have a larger changing area, while
“water”, “farmland and settlement” and “rock and bare-land” all
have very high changing rates. In Foping and Changqing NRs,
the major changes are the conversion among forest types and
the forest quality reflected by the reduced NDVI. In
Guanyinshan, the conversion from forest types to “farmland and
settlement", "rock and bare-land" needs to pay attention. In
Changqing and Guanyinshan NRs, the second succession is the
natural force to change the landscape to the better condition. It
is expected that the forest environment will be protected
effectively and success naturally after the law of “natural forest
protection program" and the functional shift from forest
production to nature conservation.
Remote sensing approach can be used as an effective tool for
detecting the landscape change and for long-term monitoring.
Multi-method is recommended to detect change in order to
improve accuracy of change detection and to explain more
change information because of their capability compensation
(Li 2003). The outputs of change detection will help the
managers to make the right decision, policy, planning and take
actions.
Acknowledgement
The research is supported by key project of National Natural
Science Foundation of China (30230080). We give our great
thanks to three nature reserves (Foping, Changqing and
Guanyingshan) for their strong cooperation and supports in our
fieldwork. We thank Mr Jin Xuelin to provide us survey data.
References
Chen, J., He, R., Guo, H., 2004. Application of maximum
likelihood classifier to the abstraction of vegetation
information. Hydrogeology and Engineering Geology (2),
pp. 94-96.
CNNRA (Changqing National Nature Reserve Authority), 2001.
Qinling Nature Reserve Network: administration plan for
Changqing National Nature Reserve, Shaanxi.
Ding, J., Tashpolat, T., Liu, C., 2001. Study of improving
accuracy for urban land cover mode recognition by using
structure information of satellite image. Arid Land
Geography 24(1), pp. 15-22.
FNNRA (Foping National Nature Reserve Authority), 2001.
Qinling Nature Reserve Network: administration plan for
Foping National Nature Reserve, Shaanxi.
Fuller, R. M., Smith, G. M., Devereux, B. J., 2003. The
characterisation and measurement of land cover change
through remote sensing: problems in operational
applications. International Journal of Applied Earth
Observation and Geoinformation. 4(3), pp. 243-253.
Gao, R., Wu, J., 2000. The application of 3S technology to
biodiversity study. Remote Sensing Technology and
Application 15(3), pp. 205-209.
Gao, Z., Wei, H., Ding, F., 1998. Methods for Subtracting
Vegetation Information Using Vegetation Index(VI) from
TM Images. Journal of Arid Land Resources and
Environment 12(3), pp. 98-104.
859
