Nguyen Dinh, Duong
development of a new classification technique which allows fast and automated analysis of both single date and multi-
temporal data sets for land cover mapping. Recently, research into an automated classification of land cover was
initiated by the author in the framework of the NASDA Research Announcement for the ADEOS-II satellite, which will
carry a GLI sensor that has 6 spectral channels similar to those of LANDSAT TM. One of the issues of automated
classification of GLI data is to develop a system for defining land cover class in the image domain. This means a land
cover category should be described by values derived from image data. The author has discovered several image
invariants based on graphical analysis of the spectral reflectance curve of a pixel. The invariant found include
modulation of the spectral reflectance curve, total reflected radiance index (TRRI) and spectral angles. These invariants
seem to be quite stable for image data generated by the same sensor. In this paper the author reports on efforts to
develop a digital definition of land cover using image invariants for automated classification. The proposed legend has
been applied to a data set simulated for the future GLI sensor and the obtained classification result has proved the
chosen approach to be correct.
2 REMOTE SENSING BASED LAND COVER CLASSIFICATION SYSTEM
There are several land cover classification systems announced by IGBP-DIS, FAO, UNESCO, CORINE or
LCWG/AARS (Land Cover Working Group / Asian Association on Remote Sensing ) that were designed based on
different concepts. In general, all these systems meet both scientific needs (global change studies) and social needs
(global, continental and national land use planning). Most of these systems have well developed hierarchical structures
so that classes of the same level will have similar characteristics. However, each land cover class is described by a
terminology and descriptors which follow conventional land cover mapping concepts and they are mainly suitable for
integrated visual image interpretation and rather difficult for application in digital image processing, especially when
using multi-temporal remote sensing data. To allow automated classification of land cover, each land cover category
should be organised to have three components as in Table 1.
Land cover definition components Information sources
Static component Single date remote sensing image
Dynamic (seasonal change) component Multi-temporal remote sensing images
Information on broader biophysical and socio-economic | Auxiliary information (topography, soils, climate)
circumstances
Table 1. Components of land cover category definition
The static component describes current physical status of the cover at the moment of observation. This type of
information could be extracted from single date remote sensing data. Example of this type of data includes water,
vegetation of different coverage densities, soil types (sandy, muddy, dry, wet etc.).
The dynamic component (seasonal change or variation) is extracted from a multi-temporal remote sensing data set. This
type of information reflects change of leaf coverage, water level or dryness of certain land cover categories.
The information on broader biophysical and socio-economic circumstances can not be derived from remote sensing data
and it should be extracted from other information sources or database such as topography, soils or climate.
Considering the above idea, a flowchart for land cover mapping is proposed in Figure 1.
RS data attime 1 — [—3»| Land cover map at time 1
Final land cover
map
RS data attime 2 |—>] Land cover map at time 2
Auxiliary
information
layer
RS dataattimen |—®| Land cover map at time n
Figure 1. Flowchart of land cover mapping by multi-&emporal remote sensing data
986 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000.
Bas
ma]
lanc
by !
dyn
com
sho
