对比航空高光谱VNIR-SWIR、LWIR与组合数据的地质测绘结果
需积分: 10 197 浏览量
更新于2024-09-07
收藏 11.19MB PDF 举报
"这篇研究文章比较了来自机载高光谱VNIR-SWIR、LWIR以及两者结合数据的地质映射结果。通过使用航空高光谱成像系统(AISA)的可见光和短波红外(SWIR)以及空间增强宽带(LWIR)数据进行遥感地质测绘。研究关注了不同红外波段在地质类型识别和制图中的表现，并采用了小波分析方法。文章详细讨论了数据采集、处理方法和结果的比较，强调了在地质调查中的应用价值。"
这篇文章是《国际地球观测与地理信息学》期刊2018年的一篇学术论文，探讨了利用遥感技术进行地质岩石分类地图的制作。主要研究了三种不同的数据来源：机载高光谱成像系统（AISA）的可见光和短波红外(VNIR-SWIR)数据、长波红外(LWIR)数据，以及这两者的组合数据在地质映射中的应用效果。
研究的核心在于比较这些不同红外波段的数据在识别和区分不同地质类型时的性能。短波红外和长波红外分别在地质特征的探测上有其独特优势，短波红外通常对矿物的吸收特性敏感，能提供丰富的矿物信息，而长波红外则可能更适用于地表温度和湿度的监测，从而反映地表条件。
文章中提到了小波分析这一方法，这可能是用来提取和解析不同波段数据中的有用信息，以提高地质特征的识别精度。小波分析能够对信号进行多尺度分析，有助于在复杂的数据中发现隐藏的模式和特征。
此外，研究还涉及了数据的获取和处理过程，包括遥感图像的校正、辐射校正、大气校正等预处理步骤，以及特征选择和分类算法的应用。这些步骤对于从原始遥感数据中提取准确的地质信息至关重要。
最后，作者比较了不同数据源和处理方法的映射结果，评估了它们在地质制图的准确性、分辨率和可操作性。这为未来的遥感地质研究提供了重要的参考，特别是在优化数据融合策略和提高地质映射效率方面。
关键词包括短波红外、长波红外、小波分析、机载高光谱成像、地质制图，表明了本文主要探讨的科学问题和技术手段。这项工作对地质学家、遥感专家以及地球科学领域的研究人员来说具有很高的实用价值和理论意义。
342
J.
Feng
et
al.
/
International
Journal
of
Applied
Earth
Observation
and
Geoinformation
64
(2018)
340–353
Fig.
1.
Regional
geology
of
Nunavik,
Quebec
and
study
area
location.
Source:
modified
from
Rogge
et
al.
(2014).
the
Archean
gneisses
of
the
Superior
Province
in
the
south
from
the
Churchill
Province
gneisses
to
the
north
(Narsajuaq)
(Modeland
et
al.,
2003).
Only
sparse
vegetation
comprising
grasses,
mosses
and
shrubs
occur
in
the
region,
with
rock
encrusting
lichens
predomi-
nant
on
bedrock
surfaces.
For
geological
mapping
this
is
particularly
challenging
for
packages
of
rocks
with
similar
mineralogy
(e.g.
mafic
to
ultramafic).
The
Cape
Smith
Belt
is
the
fourth
largest
magmatic
Ni-Cu-
platinum
group
element
(PGE)
resource
in
Canada
(Lydon,
2007)
where
deposits
are
associated
with
regionally
metamorphosed
lower
greenschist
ultramafic
sills
and
intrusions
(50–200
m)
that
outcrop
discontinuously
(dunite,
peridotite,
pyroxenite,
gabbro)
within
the
study
area
(Fig.
2a).
The
geology
of
the
Cape
Smith
belt
is
well
established
owing
to
the
long-standing
economic
interest,
which
started
with
the
first
showings
in
1898
located
in
the
west-
ern
part
of
the
belt
(Lesher,
2007)
and
continues
today.
However,
in
many
parts
exploration
and
development
is
at
an
early
stage
which
has
led
to
the
deployment
of
a
large
scale
SWIR
and
LWIR
hyperspectral
surveys
that
is
used
in
this
study.
3.
Data
acquisition
and
preprocessing
3.1.
Airborne
data
Two
airborne
surveys
were
conducted
from
late
July
to
early
September,
2008,
that
comprised
SWIR
data
collected
with
the
Specim
AISA
system
and
LWIR
data
acquired
by
the
SEBASS
sen-
sor.
For
the
AISA
data
20
flight-lines
were
used,
each
approximately
20
km
long
and
650
m
in
width
having
an
average
overlap
of
30%.
The
data
was
atmospherically
corrected
using
ATCOR4
(Richter
and
Schlapfer,
2002)
and
the
resulting
at
surface
reflectance
data
con-
tains
178
bands
(0.40–2.45
m)
at
a
spectral
resolution
of
∼10
nm
and
a
spatial
resolution
of
2
m.
A
total
of
128
bands
were
retained
for
further
analysis
with
bands
having
atmospheric
effects
and
poorer
signal
to
noise
(1.37–1.52
and
1.80–2.01
m)
or
anomalous
fea-
tures
(∼1.0–1.015
m)
being
removed.
Pre-processing
steps
that
included
dark
pixel
masking
and
steps
for
generation
of
the
AISA
mosaic
including
geocorrection,
cross-track
illumination
correc-
tion,
and
spectral
line
levelling
are
detailed
in
a
previous
study
(
Rogge
et
al.,
2014).
For
the
LWIR
analysis,
ten
flight
lines
of
SEBASS
data
(30%
overlap)
were
selected.
The
SEBASS
data
has
a
spatial
resolution
of
4
m,
comprises
128
bands
within
the
7.65–13.55
m
range.
The
in
scene
atmospheric
compensation
method
(ISAC,
Young
et
al.,
2002;
Johnson,
1998)
was
used
for
a
first
order
cor-
rection
of
atmospheric
effects
and
to
convert
the
data
to
at-surface
emissivity
before
mosaicking.
A
total
of
82
contiguous
bands
from
8.08
to
12.05
m
were
retained
for
further
analysis
owing
to
the
fact
that
this
spectral
region
contains
diagnostic
spectral
features
related
to
the
lithological
units
of
interest.
Bands
outside
this
region
had
poor
signal
to
noise.
A
spatial
region
approximately
5.5
km
by
3.5
km
where
the
SWIR
and
LWIR
image
mosaics
overlap
was
retained
as
the
study
area
and
includes
various
ultramafic,
mafic
and
sedimentary
lithological
units
(Fig.
2b).
3.2.
Field
samples
and
XRD
data
In
2009
and
2010
over
500
field
sites
were
characterized
with
respect
to
rock
type
and
lichen
cover,
which
included
the
collec-
tion
of
52
rock
samples
for
laboratory
spectral
measurements
and
mineralogical
determination
with
X-Ray
Diffraction
(XRD)
analysis.
Rogge
et
al.
(2014)
provides
a
description
of
the
sample
data
set,
XRD
results
and
collection
of
spectral
data
and
resulting
spectral
libraries.
In
brief,
the
sample
suite
is
dominated
by
four
rock
types
namely
basalt,
gabbro,
pyroxenite,
and
peridotite,
but
also
includes
a
number
of
representative
sedimentary
rocks.
Samples
comprised
natural
weathered
surfaces
(larger
than
20
×
20
cm)
to
allow
com-
parison
between
laboratory
spectra
and
spectra
extracted
from
the
airborne
imagery.
Previous
research
(Rogge
et
al.,
2014)
on
these
samples
indicates
that
the
weathered
surfaces
of
the
varied
ultramafic
and
mafic
rocks
are
comprised
of
similar
mineralogy,
but
the
rock
units
differ
in
the
relative
abundance
of
dominant
minerals,
which
complicates
their
discrimination
and
mapping.
For
example
the
peridotites
are
comprised
primarily
of
antigorite,
±diopside,
±clinochlore,
and
magnetite,
whereas
pyroxenites
are
comprised
of
antigorite,
±talc,
clinochlore,
Fe-amphibole,
and
mag-
netite.
The
gabbros
and
basalts
have
Ca-amphibole
(actinolite),
clinochlore,
±phlogopite,
±epidote,
and
albite/orthoclase.
Mapping
the
occurrence
and
abundance
of
specific
minerals
(e.g.
antig-
orite,
actinolite
and
clinochlore)
was
crucial
for
mapping
variability
within
mafic/ultramafic
units.
Sedimentary
rocks
included
a
variety
of
siliceous
and
calcareous
types.
XRD
analysis
was
not
conducted
on
these
samples
as
the
original
research
was
focused
on
the
dis-
crimination
of
mafic
and
ultramafic
rocks
of
relevance
to
mineral
exploration.
3.3.
Laboratory
spectral
collection
SWIR
spectral
measurements
of
the
sample
natural
weathered
surfaces
were
collected
on
multiple
random
spots
(4–11spots,
2
cm
diameter
per
spot)
using
a
ASD
FieldSpec
Pro
®
spectrometer,
which
has
2151
bands
ranging
from
0.35
to
2.5
m
with
a
2
nm
spectral
resolution.
Each
spectrum
was
the
results
25
co-adds.
The
spectra
of
the
measured
spots
of
each
sample
were
averaged
to
obtain
a
representative
spectrum
of
the
exposed
surface.
LWIR
reflectance
spectra
(bidirectional)
were
then
collected
over
the
same
spots
using
an
MB100
Fourier
transform
InfraRed
(FTIR)
spectrometer.
Each
spectrum
is
the
outcome
of
32
co-adds
and
has
2
cm
−1
reso-
lution
from
1.7
to
14.0
m.
The
laboratory
SWIR
and
LWIR
spectra
were
first
used
to
label
spectral
features
based
on
minerals
deter-
mined
from
XRD
analysis.
The
spectra
were
then
used
to
assess
if
these
spectral
features
could
be
detected
from
airborne
AISA
and
SEBASS
data
and
then
to
label
the
image
endmembers
used
for
regional
mapping.
4.
Spectral
analysis
methodology
In
order
to
test
the
usefulness
of
CWA
applied
to
SWIR,
LWIR
and
combined
spectral
imagery
for
geological
mapping
a
single
processing
chain
was
implemented
(Fig.
3).
This
also
allows
for
剩余13页未读，继续阅读
点击了解资源详情
点击了解资源详情
点击了解资源详情
2022-09-23 上传
2022-07-13 上传
2021-03-06 上传
2010-12-15 上传
2015-06-29 上传
woodfishye
- 粉丝: 0
- 资源: 4
我的内容管理
展开
最新资源
- Python中快速友好的MessagePack序列化库msgspec
- 大学生社团管理系统设计与实现
- 基于Netbeans和JavaFX的宿舍管理系统开发与实践
- NodeJS打造Discord机器人：kazzcord功能全解析
- 小学教学与管理一体化：校务管理系统v***
- AppDeploy neXtGen：无需代理的Windows AD集成软件自动分发
- 基于SSM和JSP技术的网上商城系统开发
- 探索ANOIRA16的GitHub托管测试网站之路
- 语音性别识别：机器学习模型的精确度提升策略
- 利用MATLAB代码让古董486电脑焕发新生
- Erlang VM上的分布式生命游戏实现与Elixir设计
- 一键下载管理 - Go to Downloads-crx插件
- Java SSM框架开发的客户关系管理系统
- 使用SQL数据库和Django开发应用程序指南
- Spring Security实战指南：详细示例与应用
- Quarkus项目测试展示柜：Cucumber与FitNesse实践
