@MastersThesis{Crusco:2006:SeReAn,
author = "Crusco, Nat{\'a}lia de Almeida",
title = "Sensoriamento remoto para an{\'a}lise multitemporal da
din{\^a}mica de {\'a}reas agr{\'{\i}}colas",
school = "Instituto Nacional de Pesquisas Espaciais (INPE)",
year = "2006",
address = "S{\~a}o Jos{\'e} dos Campos",
month = "2006-06-19",
keywords = "projeto geosafras, an{\'a}lise multitemporal, estimativa de
{\'a}rea agr{\'{\i}}cola, estat{\'{\i}}stica
agr{\'{\i}}cola, auditoria, geosafras project, remote sensing,
multitemporal analysis, crop area estimative, agriculture
statistics, audit.",
abstract = "Estat{\'{\i}}sticas agr{\'{\i}}colas s{\~a}o importantes em
um pa{\'{\i}}s como o Brasil, onde a agricultura desempenha um
papel fundamental na economia brasileira. As metodologias
atualmente utilizadas s{\~a}o baseadas em dados subjetivos, e
apresentam um car{\'a}ter n{\~a}o probabil{\'{\i}}stico. No
intuito de aprimorar os resultados obtidos, o projeto Geosafras
utiliza dados de sensoriamento remoto associado a dados de campo
na gera{\c{c}}{\~a}o de estimativa de {\'a}reas
agr{\'{\i}}colas para as principais culturas existentes no
pa{\'{\i}}s. Por{\'e}m, este m{\'e}todo apresenta algumas
limita{\c{c}}{\~o}es quanto {\`a} valida{\c{c}}{\~a}o dos
dados provenientes do campo. Neste sentido, este trabalho tem como
hip{\'o}tese central a exist{\^e}ncia de rela{\c{c}}{\~a}o
entre os dados coletados em campo no presente e
informa{\c{c}}{\~o}es de uso do solo em tempos passados. Assim,
o principal objetivo {\'e} avaliar como a din{\^a}mica de
{\'a}reas agr{\'{\i}}colas, pela abordagem multitemporal, pode
auxiliar o processo de previs{\~a}o, auditoria e
valida{\c{c}}{\~a}o de dados de campo. O estudo de imagens
multitemporais possibilitou a avalia{\c{c}}{\~a}o da
din{\^a}mica de {\'a}reas agr{\'{\i}}colas e do padr{\~a}o de
uso do solo na {\'a}rea de estudo. Foram utilizadas 24 imagens
dos sensores TM/Landsat-5 e ETM+/Landsat-7 no per{\'{\i}}odo de
2002 a 2005. As classes avaliadas neste trabalho - soja,
cana-de-a{\c{c}}{\'u}car, pasto e mata - foram bem discriminadas
visual e espectralmente. A an{\'a}lise da din{\^a}mica temporal
mostrou que cada classe possui padr{\~o}es distintos, que
est{\~a}o associados tamb{\'e}m ao calend{\'a}rio
agr{\'{\i}}cola da regi{\~a}o. A metodologia empregada neste
trabalho foi eficiente tanto na realiza{\c{c}}{\~a}o da
previs{\~a}o de uso do solo, como na indica{\c{c}}{\~a}o dos
pontos a serem auditados no painel amostral do projeto Geosafras,
apontando os erros que podem ser cometidos em campo e depurados
pela utiliza{\c{c}}{\~a}o das imagens de sat{\'e}lite.
ABSTRACT: Agricultural statistics are important in a country like
Brazil, where agriculture plays an important role over the
economy. The methodologies for crop area estimates are commonly
based on subjective data, and they present a non-probabilistic
profile. In order to increase the quality of the results, the
Geosafras Project uses remote sensing data associated to field
data for estimating the area of agricultural crops for the main
crop types existing in the country. However, this method presents
some limitations regarding the field data validation. This work
tackled this aspect and has as central hypothesis the existence of
relation between the field data colleted in the present and
information about the land-use in the past. Thus, the main
objective is to evaluate how the agricultural crop land use
dynamics, evaluated here by remote sensing multitemporal analysis,
can assist the early estimation process, auditing and field data
validation. The analysis of multitemporal images showed that it
was possible the validation of the agricultural land dynamics and
the land-use patterns in the study area. In order to accomplish
the study, 24 TM/Landsat-5 and ETM+/Landsat-7 images in the
time-frame from 2002 to 2005 were used. The crop land use classes
evaluated in this work soybean, sugarcane, grassland and forest
were well distinguished visually and spectrally. The analysis of
the temporal dynamics showed that each class has a distinct
pattern, which is also associated to the agricultural
schedule/calendar of the region. The methodology used in this work
was efficient for the land-use prediction, as well as for the
indication of the plotted points to be audited in the sample panel
of the Geosafras Project. Also, it was possible to identify the
errors that can be committed during field sampling and corrected
them by using multitemporal satellite images.",
committee = "Formaggio, Antonio Roberto (presidente) and Epiphanio, Jos{\'e}
Carlos Neves (orientador) and Kuplich, Tatiana Mora and Luiz,
Alfredo Jos{\'e} Barreto and Xavier, Alexandre C{\^a}ndido",
copyholder = "SID/SCD",
englishtitle = "Remote sensing for multitemporal analysis of crop land use
dynamics",
language = "pt",
pages = "107",
ibi = "6qtX3pFwXQZGivnJSY/Mg6TV",
url = "http://urlib.net/ibi/6qtX3pFwXQZGivnJSY/Mg6TV",
targetfile = "publicacao.pdf",
urlaccessdate = "09 maio 2024"
}