|
|
|
IMAGE-BASED VERSATILE LU INFORMATION:
A MULTIDIMENSIONAL CLASSIFICATION SCHEME
TO SUPPORT LOCAL PLANNING IN INDONESIA
Projo Danoedoro'”, Geoffrey McDonald*”, Stuart Phinn“, and David Pullar?
! Department of Cartography and Remote Sensing, Gadjah Mada University, Indonesia
2 ~ ‘ . . . .
2 School of Geography, Planning and Architecture, The University of Queensland Australia
3 Sustainable Ecosystem, CSIRO, StLucia, Australia
KEYWORDS: Remote sensing, Land-cover, Land-use, Mapping, Classification, Planning, Decision support, High resolution
ABSTRACT
Land-cover (LC) and Land-use (LU) are recognised as important factors in environmental assessment and planning. However,
different applications normally require different LC/LU information contents. Various planning tasks might face difficulties when
two or more LC/LU maps with different classification schemes share the same area of interest. Consequently, redundant works on
LC/LU surveys of the area are carried out in order to make sure that the collected LC/LU maps to be used contain relevant
information. In order to overcome such problem, a versatile LU information system (VLUIS) is developed. The VLUIS is mainly
developed based on remotely sensed imagery. Its versatility is characterised by the following aspects: (a) multilevel categorisation
with respect to particular range of spatial resolution; (bymultiple attributes of LC and LU contained in a classification scheme,
represented by five LU dimensions, i.e. spectral, spatial, temporal, ecological, and socio-economic function; (c)layer stack data
storing model for those dimensions enabling flexible attribute retrieval through a spatial query for relevant applications. In this
paper, examples of extraction methods using remotely sensed imagery, i.e. Landsat TM/ETM ' (30 m) and Quickbird (2.4 m) are
given, particularly for the first dimension of the VLUIS. Semarang area in Central Java, Indonesia, was chosen due to its relatively
complex LU phenomena within a narrow strip of image coverage. The use of the VLUIS is put in the context of refinement of the
KDLD (Key Dataset for Local Development), which has been developed by various local governments in Indonesia. As compared
to Landsat TM/ETM--based image processing, the use of high-spatial resolution imagery such as Quickbird multispectral requires
more complex spatial analysis in order to derive versatile LU dataset.
1. INTRODUCTION LU maps of the same area that were produced with different
approaches may represent different information, which may
Land-cover (LC) and Land-use (LU) are recognised as lead to users’ confusion in planning.
important factors in environmental assessment and planning. :
Unfortunately, such information is still difficult to obtain when Problems and situation described in the aforementioned
quality, relevance, and newness are considered as major paragraphs exist in Indonesian planning program, particularly at
criteria. Fresco (1994) claimed that accurate data on LU and both provincial and local levels. The local planning, according
LU changes are not easily found, both in the global and to Suroso (2000) is mainly characterised by “more LU
continental scales as well as the national and regional ones. In oriented” programs such as LU zoning and allocations,
order to support planning, remotely sensed imagery has been conservation and measures parallel with coordination and
used as a major source of LC (LC) and LU (LU) information implementation of policies. On the other hand, during the past
worldwide (Stefanov ef al., 1999; Campbell, 2002; Tapiador fifteen years, many provincial and local authorities have
and Casanova, 2003). Regional and urban planning activities in developed a so-called Key Dataset for Local Development
many countries also make use of LU information, which is (KDLD) containing a set of maps for supporting a range of
frequently derived from remotely sensed data (Carlson and planning activities. It was found that the quality of maps stored
Sanchez-Azofeifa, 1999). in the KDLDs is not adequate to support planning tasks due to
their newness, accuracy, and relevance. Among others, LC/LU
Various techniques in remote sensing can derive LC and LU maps are recognised as important information with the lowest
information with a great diversity in contents. This diversity is reliability due to their low quality. As a consequence, each
also related to the fact that LC and LU are different concepts. institution tends to develop its own LC/LU information with
The differences had been discussed by various authors, e.g. van limited consultation to the others. Therefore, redundant works
Gils et al. (1990). However, most classification schemes on LC/LU surveys take place and incompatibilities between
intentionally exchange both concepts (e.g. Anderson et al, maps partly covering the same areas come up.
1976; Malingreau and Christiani, 1982; Sandy, 1982) In
addition, it should also be realised that LU is a Based on this current situation, problem associated with the
multidimensional concept, which may be viewed from various local planning in Indonesia can be viewed from the LU
perspectives ranging from spectral (related to LC), spatial, information perspective, i.e. there is a need for developing up to
temporal, ecological, socio-economic function and legal date, accurate and relevant information on LC/LU to support
aspects. Visual interpretation could directly derive both LC and local planning tasks in Indonesia. To solve such a problem,
LU information at particular levels or scales, but digital image efforts should be carried out in conjunction with the advances
processing that is not supported by GIS analyses can generally in remote sensing technology, which can deliver various spatial
derive LC classes only. Thus, under this circumstance, LC and thematic data required by various local planning tasks.
348
Intern
The (
multi
devel
local
schen
the fii
Land:
Java,
reseat
inforr
schen
spatia
Studi
have
emin«
(Andi
categ
widel
enviri
(1982
Van (
LU |
categ
two. |
under
Di €
(200:
Simil
the u
single
at th
classi
single
level.
anotk
use :
purpx
(1991
class
to so
Reso
By u
can :
1996
regra
et al
imag
Syste
map
throu
conte
lands
steep
