@InProceedings{AmaralShimAher:1997:PABrAm,
author = "Amaral, Silvana and Shimabukuro, Yosio Edemir and Ahern, Frank J",
title = "Multitemporal Radarsat fine mode images to identify different
forest types at Tapajos National Forest - PA, Brazilian Amazon",
year = "1997",
organization = "Simp{\'o}sio Latino-Americano de Percepcion Remota, 8.",
keywords = "VEGETACAO, Floresta Nacional de Tapaj{\'o}s (PA), florestas,
Radarsat, imagens de radar, mapeador tem{\'a}tico (Landsat),
m{\'a}xima verossimilhanca, analise multitemporal.",
abstract = "This work describes the use of multitemporal Radarsat images to
identify different forest types at Tapajos National Forest, Para,
Brazilian Amazon. Color composition and principal components
derived from Radarsat fine mode images acquired in dry and wet
season were analyzed to evaluate the discrimination of forest
classes based on the phytoecological map. Landsat-TM image was
used as ancillary data.Image segmentation and classification
proceddure were applied to Radarsat data. Radar images from dry
season showed better results for discrimination of forest types
and land use than images from wet season. Radarsat principal
component images generated the best color composition when
combined with Landsat-TM principal components to discriminate both
forest types and deforestation. The segmentation procedure did not
present good result for individual Radarsat fine mode images.
Radarsat image classification, using Maximum Likelihood - ICM
algorithm, showed a potential to discriminate different forest
types, but it still requires some visual interpretation or
geographical manipulation in order to extract homogeneous forest
classes. The improvement of the information extraction from
Radarsat images will provide an useful tool for forest managemeent
and monitoring.",
conference-location = "Merida, VE",
conference-year = "2-7 nov. 1997",
copyholder = "SID/SCD",
label = "8409",
language = "en",
organisation = "SELPER",
targetfile = "INPE 7089.pdf",
urlaccessdate = "11 maio 2024"
}