Ss
ATMOSPHERIC CORRECTION OF SATELLITE IMAGES IN A TROPICAL REGION
Jurandir Zullo Junior
Scientific Researcher III, CEPAGRI/UNICAMP - Cidade Universitária "Zeferino Vaz" - 13081970 - Campinas - SP - Brasil
Tel.: 55-192-393669 - Fax: 55-192-394717 - Email: jurandir@cpa.unicamp.br
Commission VII, Working Group 1
KEY WORDS: Correction, Software, Agriculture, Classification, Vegetation, Atmospheric Correction
ABSTRACT
This paper describes our work on atmospheric correction of satellite images, started in 1990, focusing on the computer system
called SCORADIS. It gives also special attention to the acquisition of the atmospheric parameters used as input data to the System,
the importance of the atmospheric correction and the studies in development at UNICAMP. All experimental work is being done
under the conditions of tropical atmosphere, soil occupation and data disposable existent in Sáo Paulo state. Our main objective is
to contribute to a better use of satellite images in applications related to vegetation monitoring, like the biomass estimating, for
example. The most part of the atmospheric studies in remote sensing corresponds to the atmospheric conditions of the north
hemisphere that are very different of our conditions.
RESUMÉ
Ce papier decrit le travail sur la correction atmosphérique des images satellitaires qui est en developpement a l'Université de
Campinas depuis 1990. On present le logiciel de correction d'images satellitaires SCORADIS, les méthodes d'aquisition des
paramétres atmospheriques d'entrée au Systéme et l'importance de la correction atmosphérique. Le travail experimental est fait sur
les conditions atmosphériques, d'occupation du sol et de disponibilité de données de l'état de Sáo Paulo. L'objectif principal du
travail est d'améliorer la qualité des images satellitaires pour les applications agricoles.
1. INTRODUCTION AND MOTIVATION
The importance of using satellite images without atmospheric
effects in agricultural applications, like vegetation monitoring,
for example, motivated the beginning of this work. We are in a
very important agricultural region of Brazil that has great areas
of sugar cane, corn, wheat and soybean. The sugar cane, for
example, has a production system well developed and very
technical that admits a monitoring by satellites. A part of its
production is destined to the production of alcohol that is used
like fuel by cars.
We have a tropical atmosphere with some conditions that are
very particular and different of the most part of the research
work done around the world: the high water vapor content in
the most part of the year, the concentration and the type of the
aerosols.
The state of Sáo Paulo locates in the southeast region of Brazil,
with latitudes varying from 209S to 25°S and longitudes from
44°W to 53°W. The Capricorn Tropic crosses its territory.
Campinas (220954'S, 47905" W) is the city where the most part
of the experimental work takes place.
The first part of this work (1990-1994) was devoted to treat
images with homogeneous or constant atmospheric conditions
in the whole image. This is the case, for example, of the most
part of Landsat-TM and Spot-HRV images and also of small
pieces of Noaa-AVHRR images. Our main interest since the
acquisition of a Noaa-AVHRR antenna on December 1994 is in
correct images with spatial variability of atmospheric
parameters to use them in monitoring applications, like
biomass estimating, for example. Our main interesting in this
paper are the visible, near and middle infrared images although
some results are useful to treat thermal infrared images. This is
the case of the estimate of water vapor content explained in the
section 3.1.1
The following section (The Scoradis) describes the version 1.0
of the computer system that was produced during the first part
of our work (1990-1994) and the improvements that are being
done to generate the version 2.0 on June 1996. The section 3
(Input data) describes our efforts to have good input data to run
successfully the Scoradis. Section 4 is about the importance of
the atmospheric correction in two typical uses of satellite
images for agricultural purposes: calculation of vegetation
indexes and automatic pattern classification.
2. THE SCORADIS
The SCORADIS ("Sistema de COrrecáo RADiométrica de
Imagens de Satélite" or System of Radiometric Correction of
Satellite Images) is a computer system written in FORTRAN-
IV, based in the 5S Model (Tanré et allii, 1990), to treat
satellite images. It cans correct images from spectral bands
situated between 0.4pm and 2.5um, like the Landsat-TM,
Spot-HRV, Noaa-AVHRR1 and Noaa-AVHRR2, for example.
The type of our main applications of the satellite images
explains our preference for correcting images instead of some
pixels only. The possibility to parallel its calculation modules
is very interesting in monitoring applications where the
decisions have to be taken quickly.
831
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996
