Andrade, Nilo Sergio de Olive 
  
allow crop fields and pasture establishment. Deforested land is used for some years and when soil looses its fertility the 
areas are sometimes abandoned, leading to forest regeneration. 
Human occupation in the study area has considerably increased during the last three decades, following the building of 
federal road BR-364 and the development of some of the oldest and largest settlement projects in the state of Rondönia. 
The deforestation process continues to be intense in the region (Alves et al. 1998a), arousing the interest in monitoring 
it. The BR-364 road bisects the area of study and served as the way of penetration in the region at the time of its 
colonization. To the present, the road constitutes the basic link by land between Rondönia and the more populated south 
of Brazil, serving as the major way to export the region’s production, and to bring goods in. The region includes several 
of the largest settlements projects organised in the state of Rondönia under the programs of Amazon colonization during 
the 1970’s, including: Projeto Integrado de Colonizacäo (PIC) Ouro Preto (created in 1970), Projeto de Assentamento 
Dirigido (PAD) Burareiro (1974), PIC Padre Adolpho Rohl (1975), PAD Marechal Dutra (1978) and PAD Machadinho 
(1982) (Instituto Nacional de Colonizacäo e Reforma Agrária (INCRA) 1996). These projects occupied more than 2 
million ha in the region, where more than 24,000 families were settled from 1970 till 1995 (INCRA 1996). The majority 
of the properties in these settlements is relatively small (25 to 100 ha, INCRA 1996) and is scattered along secondary 
roads that emerge from the major arteries, creating the typical Rondônia "fish-bone" pattern. The area also includes 
mid-size (1000 ha) and large (10,000 ha or more) properties. The native vegetation in the region consists of dense 
tropical forests with mild seasonal characteristics (RADAMBRASIL 1978), a 4 to 6-month dry season and elevations 
ranging from 100 to 600m. The region also presents several areas covered by non-forest vegetation, associated to the 
presence of rocky soils. 
3. DATA AND METHODS 
For this study, two TM images (Path/Row = 231/067), acquired in August 03, 1995 and July 07, 1997 were used. Both 
TM images were geocoded using coordinates acquired on the field by differential GPS equipment. After the geocoding 
the images had their radiometric characteristics enhanced. All these previous steps, as well as the classifications were 
performed using the ENVI (Environmental for Visualizing Images) software, version 3.2. 
When image data is available in digital form, spatially quantised into pixels and radiometrically quantised into discrete 
brightness levels, there are two approaches that may be adopted in endeavouring to extract information. One involves 
the use of a computer to examine each pixel in the image individually with a view to making judgements about pixels 
specifically based upon their attributes. This is referred to as quantitative analysis since pixels with like attributes are 
often counted to give area estimates. The other approach involves a human analyst/interpreter extracting information by 
visual inspection of an image composed from the image data. In this he or she notes generally large scale features and is 
often unaware of the spatial and radiometric digitisations of the data. This is referred to as photointerpretation or 
sometimes image interpretation; its success depends upon the analyst exploiting effectively the spatial, spectral and 
temporal elements present in the composed image product. Because of this, to perform a visual analysis, the 
analyst/interpreter needs to elaborate the interpretation keys. These interpretation keys consist of the description of 
elements that characterize a specific feature of the terrain. In this study the characteristics used for composing the 
interpretation keys were size, shape, color and texture. 
This technique consists of the acquisition of information about a given target on the surface, through the analysis of its 
response either in different individual channels or combined channels under the form of color compositions, considering 
the spectral and temporal aspect of these images, see Pereira et al. (1989), Donzeli et al. (1992), Watrin (1994) and 
Valério Filho and Pinto (1996). 
The visual analysis was then accomplished using the mentioned aspects (size, shape, color and texture), and the 
following classes were obtained: Mature forest: area dominated by the Open Tropical Forest, without the effect of 
anthropic activities. The mature tropical forests are formed by a great number of species and high variability, presenting 
in the image with a tonality of green dark. Secondary forest: resultant of the abandonment of areas previously occupied 
by the tropical forest (mature or secondary), allowing the regeneration of the vegetation, or secondary succession. In the 
image is represented by a clearer green coloration than the one of the mature forest. Regrowth: defines the areas of 
secondary succession with age below 5 years. These are characterized by a tonality of green clearer than the one of the 
secondary forest. Pasture: areas under agronomic and cattle-raising activities with various shapes and sizes presenting 
clear magenta coloration. Bare soil: areas characterized by various shapes and sizes with coloration dark magenta. 
Burned areas: areas without defined form and characterized by a dark purple coloration. 
The objective of the second part of this study was to compare the results obtained with the supervised and unsupervised 
classifiers with the visual analysis result and verify how suitable they are. 
Unsupervised classification is a means by which pixels in an image are assigned to spectral classes without the user 
having foreknowledge of the existence or names of those classes. It is performed most often using clustering methods. 
These procedures can be used to determine the number and location of the spectral classes into which the data falls and 
to determine the spectral class of each pixel. The analyst then identifies those classes a posteriori, by associating a 
sample of pixels in each class with available reference data, which could include maps and information from ground 
  
64 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000.