International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
MAPA DE USO E
COBERTURA DA TERRA
Regiao Sinop. Claudia ltaûba (MT)
z
dageeda
1 S Aq ^
mn
e
iB
"
PP PRI
Figure 4. Land use/land cover thematic map of 2003 obtained
from ASTER/Terra image.
5. CONCLUSIONS
As for the analysis of spectral values, the spectral and spatial
resolution of ASTER images allowed a detailed discrimination
of land use/land cover classes, even considering the large
thematic complexity of the area under study (spectral
superposition and regenerative process).
Besides bands 2 (630-690 nm), 3 (760-860 nm) and 4 (1600-
1700 nm) which demonstrated its' utility for land use / land
cover classification, we recommend the inclusion of bands 6
(2185-2225 nm) and 8 (2295-2365 nm), in order to complement
the classification process.
The architecture of the neural network ART2 inserted in the
environment SGRNA, presents characteristics (such as
simplicity and easiness to enter and manipulate data) of an
adequate ASTER image classifier. Besides that, the
classification performance obtained by the neural network, as
related to the ground truth, can be considered as satisfactory,
taking into account the investigated thematic classes. The
vigilance parameter is an important variable in the classification
process done by neural network ART2, because it helps to
control the number of patterns to be classified and that this
number is not too large (e.g. for the case of values very close to
1). The training values, test and the population data vary
according to the size of the area and of computational resources
available, because for larger values there is an increased
demand of computer capacity.
126
This new procedure of ASTER image analysis shows
efficiency, indicating that ASTER products are of significant
importance for the inventory and thematic monitoring of the
Brazilian Amazon.
REFERENCES
Earth Remote Sensing Data Analysis Center (ERSDAC), 2001.
Crosstalk Correction Software — User's Guide. Tokyo, 16 p.
Hill, J; Sturn, B., 1991. Radiometric
multitemporal Thematic Mapper data for use in agricultural
land-cover classification and vegetation monitoring,
International Journal of Remote Sensing, 12 (7), pp. 1471-
1491.
Iwasaki, A.; Fujisada, H.; Akao, H.; Shindou, O.; Akagi, §.
2001. Enhancement of Spectral Separation Performance for
ASTER/SWIR. In: 47" Annual Meeting SPIE, San Diego, USA,
Vol. 4486, pp. 42-50.
Mendoza, E., 2004. Síntese Genética de Redes Neurais
Artificiais ART2 na Classificaçäo de Imagens ASTER para
Mapeamento de Uso e Cobertura da Terra na Regido Norte do
Mato Grosso. Dissertacào de Mestrado, Instituto Nacional de
Pesquisas Espaciais, INPE., Sáo José dos Campos, Brasil. 115p.
Moderate Resolution Imaging Spectroradiometer (MODIS),
2003. MODIS Atmosphere: Water Vapor product.
http://modis-atmos.gsfc.nasa.gov/MODOS5 L2/acquiring.hunl
(accessed 24 Ago. 2003)
Silva, N. C., 2003. Classificaçäo semi-automática de imagens
de sensoriamento remoto por meio de sintese genética de redes
neurais artificiais. Tese de Doutorado. Universidade de
Brasília-UNB, Distrito Federal, Brasil. 121p.
5.1 Acknowledgements
The authors acknowledge the support received from CNPq
(Grants 300677/91-0, 190012/02-1/PEC-PG), from FEMA-MT
(Mato Grosso State Foundation for the Environment) and to
company ELABORE (of the municipality of Sinop-MT for the
logistic support). The authors are also grateful to the Earth
Remote Sensing Data Analysis Center (ERSDAC-Japan), for
the delivery of ASTER images in the frame of the project
"Validating ASTER images for vegetation and land use
mapping in tropical forest area: the Brazilian Amazon"
(ASTER Announcement of Research Opportunity - Agreement
Nr. H140250).
This study is a contribution to the Program "Science and
Technology for the Management of Ecosystems”, from the
Brazilian Ministry for Science and Technology (MCT),
evaluating new technologies available to monitor environmental
issues in the Amazon.
correction of
USE O
"Agric
KEYWORD
ABSTRACT
This present
as an inputs t
Research Sta
analyzed for
indices were
(SD, Colorati
parameters v
relationship 1
to estimate s
followed by :
and PCI (84
multiple regr
sand also for
soil fertility v
It is well knc
even within :
is to fine 1
variability, a
of production
degradation c
1999). Hence
inputs requir
Conventional
samples take:
the limitation
and costly. ,
and laborator
an efficient,
spatial differ
offers the poi
soil propertit
strategies to c
soil propertie
matter, soil
content and r
behaviour (L
factor in dete
in quantities
According to
an increase ir
in a decrease
infra red bar
