Bacchi, Gustavo Sanches
Comparative analysis of land cover maps obtained from vertical aerial photographs
and panchromatic Spot images.
Gustavo Sanches BACCHI; Marcos RODRIGUES; José Alberto QUINTANILHA; Diana Sarita HAMBURGER
University of Sáo Paulo, Brazil
Laboratório de Geoprocessamento, Escola Politécnica
gubacchi@usp.br
marcos@cartaconsultoria.com.br
jaquinta@usp.br
dianash@usp.br
Technical Commission TC VII-1
KEY WORDS: Land use/Land cover, Classification, Remote sensing, SPOT, Watersheds.
ABSTRACT
The land cover map is one of the most important spatial data in many geographical information systems applications.
Most of the environmental analyses use land cover and land use maps (CAHILL et al., 1995). This study focuses on the
integrated watershed land cover planning as a basis to water quality assurance to be used by local governments.
Thinking on one of the main problems suffered by local governments and small watershed associations, the resource
scarceness, this study made a comparison between two remote sensing sources in a low cost digital land cover
classification.
1 INTRODUCTION
The regional watershed planning approach of most of the Brazilian towns are characterized by shortage of the basic
data. Without any doubt, this data privation is one of the most important restrictions for the watershed planning
developments and land cover zoning implementation. Because of that, the proposal and the use of low cost
methodologies that could increase the viability of the basic data acquisition for the planners’ demand are important
tasks in the regional scenario.
The regional scenario could be described as the decision makers that work in small watershed planning and frequently
have scarceness of resources to develop their planning tasks. In this context, the optimization of the resources was the
main approach of the present study, and this philosophy must be remembered at the employed classification method
analysis. With this approach, the method adopted in this study was based in the panchromatic image classification and
aerial photographs, in order to evaluate the degree of difficulty to obtain the land cover map. Also because of this
approach, one of the main objectives of this study was to identify, between two sources of digital remote sensing data
sources, the most adequate for the regional scenario.
2 LAND COVER MAP GENERATION METHODOLOGY
2.1 The remote sensing data sources analyzed
The present study used a vertical aerial photograph which was obtained in a 1:25.000 scale. This photograph was
converted for the digital environment by a 600 dpi optical scanner device. The other source used was an HRV Spot
panchromatic satellite image (K716/J397) obtained ten months after the photogrammetric flight over the Ibaté-Mirim
watershed. This watershed is one of the main water sources for urban supply of the Aracariguama town situated in the
state of Sáo Paulo, southeast of Brazil.
Since both remote sensing data have different spatial resolutions, they were submitted to a compatibilization process to
enable the accuracy analysis described later on. Both of the digital remote sensing sources were imported in to the
Environment for Visualizing Images — ENVI (RESEARCH SYSTEMS INC., 1997) for digital processing
(RICHARDS, 1994).
98 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000.