@InProceedings{AmaralShimValeAher:1997:ReFoTr,
author = "Amaral, Silvana and Shimabukuro, Yosio Edemir and Valeriano,
Dalton de Morisson and Ahern, Frank J",
title = "Relation between forest and tropical and microwave remote sensing
parameters at Tapajos National Forest in Brazilian Amazon",
year = "1997",
organization = "Simp{\'o}sio Latino-Americano de Percepcion Remota, 8.",
keywords = "ESTUDOS INTEGRADOS DO MEIO AMBIENTE, FLORESTA NACIONAL DE TAPAJOS
(PA), FLORESTAS, PROJETO PRORADAR, RADARSAT, MAPEADOR TEMATICO
(LANDSAT), INDICE DE AREA FOLIAR, LANDSAT 5, INDICE DE VEGETACAO
DA DIFERENCA NORMALIZADA.",
abstract = "This paper presentes the relation between remote sensing and
forest parameters obtained for two forest types at Tapajos
National Forest, Par6 state, Brazil. It in part of the activities
developed in the PRORADAR project, a INPE/CCRS cooperation, to
evaluate RADARSAT capabilities for forest applications. The study
area comprises the Tapaj6s National Forest, located between the
coordinates of 55' 30' and 54' 36' west Longitude and 2' 30' and
40 18' south Latitude, south of Santarem city (Para state),
Brazilian Amazon. Despite of several ongoing research projects in
this study area, the National Forest is under a management
program(Brazilian Institute of Renewable Resources - IBAMA)to
explore its forest resources. Tropical Dense Forest, with abundant
economic wood species, dominates the two principal
geornorphological regions: Amazon Low Plateaus and Xingu and
Tapaj6s High Plateaus. Forest parameters such as height (H),
diameter at breast height (DBH)for emergent trees(trees with DBH
bigger than 40 cm), and the relative Leaf Area Index (LAI)for each
stand were measured. Radar Backscattering of RADARSAT fine mode
images and Normalized Difference Vegetation Index (NDVI)and
Reflectance values of LANDSAT 5 - TM constituted the remoting
sensing parameters. Height, DBH and LAI measurements were taken
from two different vegetation types at Tapajos National Forest, in
October 1996. For each vegetation type analyzed, two sample areas
were surveyed: one along 6 transects of 400 m and another along 3
transects of 200 m. LAI values were measured using LAI-2000
(LICOR). The forest measurements ewre positioned with a GPS
reference on the field and imported into a GIS. A sernivariogram
analysis was performed to evaluate the spatial relation between
LAI measurements. It was obtained an average distance of 50 m from
which the LAI measurements were not related to each other. The
LAI, DBH and height values were used to generate a regular grid of
the same spatial resolution as the remote sensing finest
resolution (12.5 m x 12.5 m)Frorn the grid values, the average
values of 50m x 50m were computed. LANDSAT 5 - TM image, acquired
on July 08, 1996, was orthocorrected to the cartographic
projection of the field work data and resampled to 12.5 m spatial
resolution. For each sampled area, LANDSAT 5 TM, channels 3, 4,
and 5 digital values were extracted, and NIDVI values were
computed from channels 3 and 4. Four RADARSAT fine mode images,
with different incidence angles (F2D and F5D), acquired in May and
October 1996, were calibrated and ortho-corrected. They were
resampled to the same pixel size as TM and Field data. The
blackscattering values (Gamma nought)related to each sampled area
were extracted and the average values (50m x 50m), 10 dB units
were computed. Remote sensing variables were generally more
correlated to height instead of LAI measurements, which can be
assigned to the dossel structure variation. Medium correlation was
obtained for RAIDARSAT fine mode (F2D)and channel 3 TM. The other
resoults from correlation and multivariate regression between
forest and remote sensing parameters were presented and
discussed.",
conference-location = "Merida, VE",
conference-year = "02-07 nov. 1997",
label = "8397",
organisation = "SELPER",
targetfile = "INPE 7077.pdf",
urlaccessdate = "28 abr. 2024"
}