004 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B-YF. Istanbul 2004 
  
Table 2. Description of ve 
e 
getation type and their appearance on a Landsat image (4 red, 5 — green and 3 > blue) 
  
Vegetation type 
Vegetation characteristic 
Interpretation remark 
  
Primary 
(tropical wet 
evergreen and 
semi-evergreen) 
forests 
Multi-storeyed forests with a number of 
matured trees in the upper canopy; shrubs and 
sometimes bamboo in lower canopy 
In some areas a gregarious occurrence of 
Dipterocarpus species is noticed 
Dark brown with sometimes reddish 
brown spots 
Rough texture 
Irregular boundary 
  
Secondary 
(young) forests 
Mixture of a number of species at pole-stage 
Cleared several years ago and left undisturbed 
Red to reddish brown 
Smoother texture than the mature 
forests 
  
  
  
Mixed Dominated by bamboo (Muli: Melocanna Yellow and yellowish brown 
vegetation baccifera, Mitinga: Bambusa tulda) with sometimes Texture varies with region 
dominated by scattered trees in the upper canopy Difficult to distinguish from natural 
bamboo Assoication varies with local topography, top forests when bamboo is intermixed 
and mid slope covered by small-sized bamboo 
(Muli) whereas foothill and valleys by larger 
bamboos (Mitinga) 
Shrubs Shrubs intermixed with seedling and sapling of Light yellow with green, smooth 
natural vegetation, bamboo and grasses texture 
Formerly covered by natural forests, repeated If canopy is closed appears light 
disturbance resulted scrubby or bush type vegetation yellow but in case of exposed soil it 
appears green 
Located nearby the habitation 
Acacia Plantation with Acacia auriculformis and A. Dark red with smooth texture, dark 
plantation 
mangium 
Excellent growth was noticed 
tone is sometimes identical 
Identification between two species is 
not possible without ground information 
Usually follows a regular boundary 
  
Mixed A variety of indigenous species: Dipterocarpus Reddish brown, smooth texture, 
plantation with turbinatus, Sysygium grande, Artorcarpus regular boundary 
indigenous Chaplasha, Gmelina arborea, Chikrassia tabularis 
species etc. 
  
Green with dark brown spots. Those 
spots might appear due to scattered 
remaining of natural vegetation inside this 
Scattered trees, |e 
teak plantation, 
teak coppice 
This class has a large variability, all the e 
mentioned class appears as the same spectral 
response on a Landsat image 
° Teak coppice appears if teak plantations were class 
removed e 
° Scattered trees appear due to the extreme 
human interference on natural forests 
Teak trees have large leaves with no 
undergrowth, scattered trees with scrubby 
surrounding appears as same. 
° No difference can be made with teak 
plantation or coppice 
  
  
  
  
  
  
Rubber e Introduced before one-two decades for the ° Green on Landsat image. This tone 
plantation production of rubber appears from bare soil because trees were 
e The spacing is variable leafless during the time of image 
» Rubber tree sheds leaves during winter. acquisition 
CY. Beginning of January the leaves turn to pale; by the 
of | end of the month trees are complete leafless 
ps 
ed 
ely reflectance due to the internal leaf structure varies with age, and 4. CONCLUSION 
rot young vegetation has a higher reflectance than the mature one 
in this spectral region though shadow might have an additional The study generated an interpretation key and a spectral library, 
influence on it. which can be used for interpretation and monitoring forest in 
ver | the tropics. However, there are several limitations of this study. 
in | For example, variation in crown characteristics within a specific 
of | 3.4 Selective Interpretation Key species can be a source of interpretation error leading to 
of | misclassification. Among the factors that can cause significant 
m | The above result can be summarized and presented in table 2 for ^ variation are tree age and stress. Trees growing on deep, fertile 
Icy interpretation purpose. To ease the interpretation, Landsat soils at low elevations tend to have different crown forms than 
) p- image and its corresponding field photos have also been their counterparts on shallow, rocky soils at high elevations. 
he included (Figure 4 and figure 5). Position of a tree crown on a satellite image can also change its 
ral appearance. Trees on shaded slopes will appear darker and have 
161