@InProceedings{BecerraAlva:2006:DeTrSa,
author = "Becerra, Jorge Alberto Bustamante and Alval{\'a}, Regina
C{\'e}lia dos Santos",
affiliation = "{} and {Instituto Nacional de Pesquisas Espaciais (INPE)}",
title = "Detection of tropical Savannah (Cerrado) physiognomies in the
Legal Amazon by the application of the vegetation and moisture
indices with MODIS time series data",
booktitle = "Proceedings...",
year = "2006",
editor = "Vera, Carolina and Nobre, Carlos",
pages = "861--867",
organization = "International Conference on Southern Hemisphere Meteorology and
Oceanography, 8. (ICSHMO).",
publisher = "American Meteorological Society (AMS)",
address = "45 Beacon Hill Road, Boston, MA, USA",
keywords = "Land cover classification, Cerrado physiognomies, Legal Amazon,
vegetation and moisture index, MODIS image analysis.",
abstract = "Deforestation and agricultural expansion are the main
anthropogenic factors that determine the land cover changes in the
Legal Brazilian Amazon. These changes could alter the dynamic
equilibrium between regional climate and vegetation. The mapping
of the land cover usually clusters, as a single class, the
forested, savanna type and grassland Cerrado physiognomies. This
is caused by the difficulty to differentiate these physiognomies
among them, due to similar spectral responses. However, the
spectral behavior of these land cover classes may be identified
when they are analyzed along the annual cycle, including both the
dry and the rainy season. To validate this hypothesis the State of
Tocantins, which is characterized by presenting extensive savannah
areas, was chosen. The main data used were two MODIS Vegetation
Indices, the normalized difference vegetation index (NDVI) and the
enhanced vegetation index (EVI), both for 16-days composition
periods and 250 m resolution, and the near and medium infrared
bands. A moisture index, the land surface water index (LSWI), was
computed using these bands. To cover the area of the study, a
mosaic of two MODIS tiles (h13v09 and h13v10) was used. For the
seasonal analysis, four mosaics were used: for the rainy season,
from January 01 to 16, 2004, and from March 21 to April 05, 2004,
and for the dry season, from July 11 to 26, 2004 and September 29
to October 14, 2004. For each period, 3 mosaics were elaborated,
corresponding to NDVI, EVI and LSWI index images, totaling 12
mosaics. The land cover classes were determined through an
unsupervised classification technique of the index images. To
validate the results of the classification, the Global Land Cover
- GLC 2000 map (1 km resolution) was used. Finally, for the
determination of the seasonal patterns of each class evaluated, a
longitudinal profile was elaborated from a transect that includes
the main classes of vegetation cover. The results of the
classification show the occurrence of mixtures between classes of
land cover: deciduous forest with agriculture, during the rainy
season, and agriculture with savannah, during the dry season. They
are identified from the profiles of the 3 indices that show
similar tendency in the grouping of the land cover classes. Thus,
a way to identify the Cerrado physiognomies in the Amazon region
could be, first, the separation of the Cerrado class from the
other classes using the rainy season data and, thereafter, by the
classification of the Cerrado physiognomies, using the dry season
data.",
conference-location = "Foz do Igua{\c{c}}u",
conference-year = "24-28 Apr. 2006",
copyholder = "SID/SCD",
language = "en",
organisation = "American Meteorological Society (AMS)",
ibi = "cptec.inpe.br/adm_conf/2005/10.31.23.42",
url = "http://urlib.net/ibi/cptec.inpe.br/adm_conf/2005/10.31.23.42",
targetfile = "861-868.pdf",
type = "Impacts of land cover and land use changes",
urlaccessdate = "20 maio 2024"
}