INTERPRETATION OF TROPICAL VEGETATION USING LANDSAT ETM+ 
IMAGERY 
M. M. Rahman * *, E. Csaplovics®, B. Koch”, M. Köhl“ 
? Department of Earth Sciences, Dresden University of Technology, Helmholz strasse 10, D-01069 Dresden, Germany — 
mahmud@frsws10.forst.tu-dresden.de, csaplovi@rcs.urz.tu-dresden.de 
® Dept. of Remote Sensing and Landscape Information System, Albert-Ludwigs University, Tennenbacher strasse 4, D- 
79106, Freiburg, Germany — Barbara. Koch@felis.uni-freiburg.de 
* Department of World Forestry, University of Hamburg, Leuschner strasse 91, 21031 Hamburg, Germany- 
m.koehl@holz.uni-hamburg.de 
KEY WORDS: Forestry, Vegetation, Interpretation, Monitoring, Landsat, Imagery 
ABSTRACT: 
Monitoring of tropical forest is important as it plays a vital role in Global biodiversity and terrestrial carbon dynamics. Remote 
sensing can be effectively used to monitor the tropical vegetation. However, an operational tool for interpretation and monitoring of 
tropical forest has not yet well developed. Therefore, the current study aimed to generate an interpretation key and a spectral library 
for the forests of south-castern Bangladesh. The study used Landsat ETM+ imagery. Geometrically corrected imagery was obtained 
from USGS and later was adjusted in the field using GPS. Atmospheric correction was done by modified dark object subtraction 
method. Eight different vegetation types could be separated from the satellite image. The study has developed an interpretation key, 
which can be effectively used for monitoring tropical vegetation. A discrete spectral library was also developed from Landsat ETM+ 
image bands. The result of the study will be quite useful for interpretation and monitoring of tropical forests in similar ecosystem. 
1. INTRODUCTION to the configuration or the general outline of objects as recorded 
on imagery. Man-made feature sometimes follow a regular 
1.1 Monitoring of Tropical Forest boundary, but the natural features usually follow natural 
boundary. Site refers to topographic or geographic location and 
Forests of the tropical zones, which constitute about half of the is a particular important aid in the identification of vegetation 
world’s forests and which are mostly occurring in developing types. Texture is the frequency of tonal change on the 
countries, are overexploited in some regions, under harvested in photographic image. It determines the overall visual 
others, and unmanaged in nearly all. Those forests consist of ‘smoothness’ or ‘coarseness’ of image features. Pattern relates 
World's largest biodiversity and play an important role in ^ (o the spatial arrangement of objects. Association refers to the 
global terrestrial carbon budget. Scientific management of ^ occurrence of certain features in relation to others (Lillesand 
tropical forests needs a large amount of reliable information that and Kiefer, 2000; Howard, 1991). 
can only be obtained in a time and cost effective way with the 
help of aerial photographs and other forms of remote sensing. In 1.3 Visual-interpretation Keys 
spite of importance and urgency, relative less progress has been 
made in recent years in the use of air and space-borne imagery 
of tropical forest vegetation than that of temperate and boreal An image-interpretation key is a set of guidelines used to assist 
vegetation (Lund 1997). Therefore, the objective of this study is interpreters in rapidly identifying features on a remote sensing 
to generate a visual-interpretation key and spectral library for ^ image. Depending on the method of presenting diagnostic 
monitoring vegetation of south-eastern Bangladesh, which are features, image-interpretation keys may be grouped into two 
currently not available. classes: selective keys and elimination keys. Selective keys are 
usually made up of typical illustration and descriptions of 
1.2 Elements of Visual Interpretation objects in a given category. In contrast, elimination keys require 
the user to follow a step-by-step procedure, working form the 
The six primary elements of visual interpretation are tone or general to the specific. One of the more common forms of 
colour, size, shape, texture, shadow and pattern (Estes and elimination keys is the dichotomous type. Here, the interpreter 
Simonett, 1975). In addition of those height, size and ^ qst continually select one of two contrasting alternatives until 
association may be added (Howard, 1970). Tone refers to the he or she progressively eliminates all but one item of the 
relative brightness or colour of objects on imagery. Size, shape category, which is correctly describing the feature under 
and position (site), are combined under the term contextual investigations (Avery and Berlin, 1992). 
information. Size of the object often help to identify it, which 
includes the recognition of old over-mature trees, maturing Following the above principles several interpretation keys have 
trees, saplings, seedlings and shrubs on an image. Shape relates been developed both for tropical and temperate forests based on 
  
* Corresponding author and permanent address: Bangladesh Space Research and Remote Sensing Organization (SPARRSO), 
Agargaon, Sher-E-Bangla Nagar, Dhaka-1207, Bangladesh 
157