@MastersThesis{Coldibeli:2021:EsCaPa,
author = "Coldibeli, Matheus Costa",
title = "Estimativas geoestat{\'{\i}}sticas para a constru{\c{c}}{\~a}o
de modelos digitais do terreno de grandes {\'a}reas: estudo de
caso a partir do adensamento de amostragens provenientes de
mapeamentos com VANT de baixo custo",
school = "Instituto Nacional de Pesquisas Espaciais (INPE)",
year = "2021",
address = "S{\~a}o Jos{\'e} dos Campos",
month = "2020-12-15",
keywords = "MDT, VANT, krigagem ordin{\'a}ria, PEC-PCD, valida{\c{c}}{\~a}o
cruzada, DTM, UAV, ordinary kriging, cross validation.",
abstract = "A representa{\c{c}}{\~a}o altim{\'e}trica da
superf{\'{\i}}cie terrestre {\'e} essencial para a
compreens{\~a}o de diversos processos naturais que ocorrem sobre
a paisagem, al{\'e}m de fornecer informa{\c{c}}{\~o}es para
estudos relacionados a fatores ambientais, projetos de engenharia,
planejamento e gest{\~a}o urbana, entre outros. Uma das formas
tradicionais para o reconhecimento do terreno {\'e} baseada em
amostras pontuais obtidas por levantamentos topogr{\'a}ficos
convencionais para a elabora{\c{c}}{\~a}o de produtos
cartogr{\'a}ficos, tais como os Modelos Digitais do Terreno
(MDTs). Alguns estudos sobre a superf{\'{\i}}cie terrestre
exigem MDTs representativos em grandes escalas e, no caso de
regi{\~o}es de grandes dimens{\~o}es, {\'e} necess{\'a}ria uma
quantidade satisfat{\'o}ria de amostras sobre o terreno. No
entanto, dependendo das caracter{\'{\i}}sticas do terreno,
m{\'e}todos convencionais de medi{\c{c}}{\~a}o podem tornar o
processo custoso, demorado e perigoso. A partir de t{\'e}cnicas
recentes de fotogrametria e sensoriamento remoto, mapeamentos da
superf{\'{\i}}cie terrestre podem ser obtidos atrav{\'e}s de
Ve{\'{\i}}culos A{\'e}reos N{\~a}o Tripulados (VANTs) de baixo
custo e que apesar de uma baixa autonomia de voo, possibilitam a
gera{\c{c}}{\~a}o de modelos tridimensionais do terreno de forma
detalhada. Desta forma, esse tipo de mapeamento pode, a
princ{\'{\i}}pio, ser utilizado conjuntamente com amostras de
levantamentos topogr{\'a}ficos convencionais para a
elabora{\c{c}}{\~a}o de MDTs de grandes {\'a}reas, mantendo um
certo grau de representatividade do terreno. Este trabalho
prop{\~o}e uma metodologia para a constru{\c{c}}{\~a}o e
avalia{\c{c}}{\~a}o de MDTs de grandes {\'a}reas a partir de
amostras pontuais, derivadas de levantamentos topogr{\'a}ficos, e
de amostras extra{\'{\i}}das de representa{\c{c}}{\~o}es do
terreno oriundas de mapeamentos com VANT de baixo custo
(MDT-VANT). A constru{\c{c}}{\~a}o dos MDTs de grandes
{\'a}reas {\'e} realizada a partir de estimativas por krigagem
ordin{\'a}ria e a acur{\'a}cia posicional destes MDTs {\'e}
avaliada de acordo com o Padr{\~a}o de Exatid{\~a}o
Cartogr{\'a}fica para Produtos Cartogr{\'a}ficos Digitais
(PEC-PCD) na escala 1:10.000. A {\'a}rea de estudo localiza-se na
zona rural do munic{\'{\i}}pio de Jacare{\'{\i}}, estado de
S{\~a}o Paulo, Brasil, constituindo uma {\'a}rea de
aproximadamente 47,7km˛. O local de estudo foi dividido em quatro
regi{\~o}es, com base nas suas caracter{\'{\i}}sticas
topogr{\'a}ficas, nas quais foram realizados oito mapeamentos
aerofotogram{\'e}tricos com um VANT Phantom 3 Advanced para a
constru{\c{c}}{\~a}o dos MDTs-VANT. As acur{\'a}cias posicional
planim{\'e}trica e altim{\'e}trica dos MDTs-VANT foram avaliadas
atrav{\'e}s de valida{\c{c}}{\~o}es cruzadas e classificadas de
acordo com o PEC-PCD na escala 1:1.000. Os MDTs das quatro
regi{\~o}es foram constru{\'{\i}}dos com resolu{\c{c}}{\~a}o
espacial de 4m em dois cen{\'a}rios distintos, o primeiro
considerando somente o conjunto inicial de amostras
topogr{\'a}ficas de cada regi{\~a}o (MDTs Iniciais) e o segundo
utilizando, al{\'e}m do conjunto inicial, novas amostragens
extra{\'{\i}}das de cada MDT-VANT (MDTs Finais). Os modelos
finais apresentaram uma melhora significativa na qualidade
posicional altim{\'e}trica, enquadrando-os nas classes do PEC-PCD
na escala 1:10.000, al{\'e}m de uma representa{\c{c}}{\~a}o
mais detalhada das caracter{\'{\i}}sticas do terreno nos locais
mapeados com VANT. Um estudo hipot{\'e}tico de corte e aterro
para o planejamento de estradas foi realizado usando os modelos
finais e um modelo digital de superf{\'{\i}}cie (MDS) do
programa espacial SRTM (Shuttle Radar Topography Mission), com
resolu{\c{c}}{\~a}o espacial de 30m, com o intuito de avaliar o
impacto de MDTs com diferentes resolu{\c{c}}{\~o}es espaciais
nas estimativas realizadas. Os resultados indicaram
diferen{\c{c}}as nas estimativas de corte e aterro utilizando
modelos com diferentes configura{\c{c}}{\~o}es, nos quais os
MDTs resultantes tendem a indicar um melhor planejamento e
gest{\~a}o para o estudo. ABSTRACT: The altimetric representation
of the earth's surface is essential for understanding several
natural processes that occur on the landscape, besides providing
information for studies related to environmental factors,
engineering projects, planning and urban management, among others.
Digital Terrain Models (DTMs) are traditional cartographic
products used in earth's surface recognition. Generally, they are
derived from point samples obtained by conventional topographic
surveys. Some terrain studies require representative DTMs on large
scales and huge amounts of ground samples. However, depending on
the characteristics of the terrain, conventional measurement
methods can make the process expensive, time-consuming and
dangerous. On the other hand, based on photogrammetry and remote
sensing techniques detailed three-dimensional surface models can
be obtained using a low cost Unmanned Aerial Vehicle (UAVs).
Therefore, in this context, This work proposes a methodology for
the construction and evaluation of MDTs of large areas from point
samples, derived from topographic surveys and samples extracted
from representations of the land from mapping with low-cost UAV
(UAV-DTM). The construction of large area MDTs is performed based
on estimates by ordinary kriging and the positional accuracy of
these MDTs is evaluated according to the Cartographic Accuracy
Standard for Digital Cartographic Products (PEC-PCD) on the
1:10.000 scale. The study area has 47.7 km˛ and it is located at
Jacare{\'{\i}} City, S{\~a}o Paulo state, Brazil. The study
area was divided in four regions, based on their topographic
characteristics, in which eight aerophotogrammetric mappings were
performed with a UAV Phantom 3 Advanced for the construction of
the UAV-MDTs. In addition, the planimetric and altimetric
positional accuracies of each UAV-DTM were evaluated through a
cross validation process and according to the PEC-PCD on the
1:1000 scale. The study area was divided in four major regions and
for each region was built a DTM having 4m of spatial resolution.
It was considered two scenarios in the generation of the DTMs: (a)
using only the initial set of topographic ground samples (called
by Initial DTM) and (b) using the initial set of topographic
ground samples added to new samples extracted from DTMUAV (called
by Final DTM). The altimetric quality of final models showed a
significant improvement and according to PEC-PCD's classification
they could be placed in the scale 1:10000. It was also observed a
more detailed representation of the terrain characteristics in the
locations mapped with UAV. The generated DTM was compared with a
Digital Surface Model (DSM) of the SRTM (Shuttle Radar Topography
Mission), having 30m of spatial resolution in a hypothetical study
involving land movement and topographic profiles for road
planning. This comparison was carried out in order to assess the
impact in the road planning of height estimates provided by DTMs
having different spatial resolutions. The results indicated
significant differences in the cut and filled estimates, using
models presenting different configurations and besides that the
generated DTMs should allow better planning and management on the
land movement and topographic profiles study.",
committee = "Renn{\'o}, Camilo Daleles (presidente) and Sant’Anna, Sidnei
Jo{\~a}o Siqueira (orientador) and Felgueiras, Carlos Alberto
(orientador) and Ortiz, Jussara de Oliveira and Oliveira, Cleber
Gonzales de",
englishtitle = "Geostatistic estimates for the construction of digital terrain
models of large areas: case study from the densification of
samples by low-cost UAV mapping",
language = "pt",
pages = "189",
ibi = "8JMKD3MGP3W34R/43SB255",
url = "http://urlib.net/ibi/8JMKD3MGP3W34R/43SB255",
targetfile = "publicacao.pdf",
urlaccessdate = "30 abr. 2024"
}