Introduction to Remote Sensing for Geothermal Exploration

1
Introduction to Remote SensingforGeothermal
Exploration

• Gregory D. Nash, Ph.D.
• Research Associate Professor
• Energy Geoscience Institute
• University of Utah
• gnash_at_egi.utah.edu

2
1.0 Discussion Outline

• 2.0 The basics remote sensing concepts.
• 3.0 The data what kind do I need?
• 4.0 The software how can I process the data?
• 5.0 Geothermal exploration- the applications.

3
2.0 The Basics Electromagnetic Radiation
Passive Sensors
4
2.0 Electromagnetic Radiation Spectrum
5
2.0 Electromagnetic Radiation Spectrum- Some
Concepts

• Most common regions for remote sensing
• 400 nm 2500 nm (visible, near infrared, and
  short wave infrared -- VNirSWIR)
• 8 12 micrometers (thermal infrared)
• 75 107 cm (radar)

6
2.1 EMR Scattering Attenuation

• The atmosphere prevents transmission of some
  electromagnetic radiation
• CO2 strongly absorbs mid and far infrared -
  strongest from 13 µm to 17.5 µm
• Water vapor major absorptions in several bands
  between 1.0 µm and 7 µm and above 27 µm

7
2.1 EMR Scattering Attenuation

• Rayleigh scattering (clear atmosphere)
  atmospheric particle lt wavelength, i.e. N2 O2
• Affects shorter ? to a higher degree
• Affects sunsets
• Mie scattering larger atmospheric particles
  (dust, pollen, water droplets, smoke)
• Affects a broad part of EMR at and near visible ?
• Nonselective scattering large water droplets and
  dust particles

8
2.1 EMR Scattering Attenuation

• Idealized and averaged radiation (short wave)
• 3 absorbed by ozone
• 25 reflected by clouds
• 19 absorbed by dust and gasses
• 8 reflected by ground
• 45 absorbed by ground

9
2.1 EMR Scattering Attenuation
10
2.2 Atmospheric Correction

• Atmospheric correction conversion to apparent
  reflectance (critical for hyperspectral data)
• Software
• ACORN Analytical Imaging and Geophysics
• http//www.imspec.com/page4.html
• ENVI FLAASH
• http//www.ittvis.com/index.asp
• Both are Modtran-4 based
• Most digital image analysis software has basic
  atmospheric correction capability

11
2.2 Atmospheric Correction Comparisons
Atmospheric correction results on 10 nm
hyperspsectral data (AVIRIS) kaolinite spectrum.
Top internal average reflectance
(IAR) Middle ATREM Bottom ACORN Each
method, from top down, represents an improvement.
12
2.3 Why Does Remote Sensing Work?

• Different materials at the Earths surface have
  different reflectance or emitted spectra
• Major influences on reflectance spectra
• Vegetation
• Chlorophyll, spongy mesophyll, and water
• Minerals
• Fe, Fe, OH-, CO3 and water

13
2.3 Clay Mineral Spectra Examples
Ammonio-smectite
Montmorillonite
Alunite
Illite
Kaolinite
Dickite
Halloysite
14
2.4 Remote Sensing Data Formats

• How are remotely sensed data received?
• Digital Numbers
• Generally 8 to 16 bit values
• Can be converted to radiance
• Radiance
• Watts per meter squared per steradian (W m-2
  sr-1)
• Can be converted to reflectance
• Reflectance
• Used in absorption spectroscopy
• All of the above can be useful for multispectral
  or panchromatic imagery, however hyperspectral
  should be converted to reflectance.

15
3.0 Remote Sensing Data Types

• Multispectral (several relatively broad bands)
• Hyperspectral (many narrow bands)
• Thermal Infrared (TIR can be multispectral)
• Panchromatic (gray scale single very broad
  band)
• Radar (microwave)
• LIDAR (LIght Detection and Ranging - laser)

16
3.1 Multispectral Data Characteristics

• Low to high spatial resolution
• Low to medium spectral resolution
• Several bands covering parts of EMR spectrum from
  400 nm 2500 nm
• Sometimes includes TIR (thermal infrared) bands
• Low to medium cost

17
3.1 Multispectral Data Uses

• Regional to sub-regional
• Structure mapping
• Hydrothermal alteration mapping
• Lithology mapping
• General vegetation mapping
• Vegetation anomaly mapping
• Environmental baseline mapping/Change detection

18
3.1.1 Landsat Thematic Mapper (TM 4 5 and 7
ETM satellite data)

• Relatively low spectral resolution
• 6 broad visible, near, and short wave infrared
  bands at 28.5 m spatial resolution
• ETM has a 15 m spatial resolution panchromatic
  band
• One TIR band (90 m spatial resolution)
• ETM (Landsat 7) cost 600.00 250.00 US for
  additional scenes
• Landsat 4 5 TM cost 425.00 200.00 US for
  additional scenes
• 185 x 185 km coverage
• Very low cost per unit area
• Excellent coverage and availability
• Good for regional geologic characterization
• http//edc.usgs.gov/products/satellite.html

19
3.1.1 TM Full Scene Zoom to Full Resolution
Full image 185 km x 185 km
Zoom 15.5 km x 15.5 km
20
3.1.2 ASTER multispectral

• 3 - 15 m spatial resolution VNIR bands
• 6 - 30 m spatial resolution SWIR bands
• 5 - 90 m spatial resolution TIR bands
• SWIR bands have higher spectral resolution than
  TM/ETM data
• Can be used to identify and map some specific
  hydrothermal alteration minerals
• TIR can be used to map thermal anomalies and
  silica rich rocks/quartz
• Extremely low cost per unit area (60 km x 60 km
  scene _at_ 85.00 US)
• Fair coverage but images for some areas may not
  available (pretty good U.S. coverage)
• Good for regional and sub-regional geologic
  characterization
• http//edc.usgs.gov/products/satellite.html

21
3.1.2 ASTER Full Scene Zoom to Full Resolution
Full 60 km x 60 km image (SWIR - 30 m resolution)
Zoom 7.5 km x 7.5 km (VNIR 15 m resolution)
22
3.2 Quick Bird Pan-Sharpened
This image has four bands (VNIR). Useful
primarily for structure and general vegetation
mapping. Quick bird is available in resolutions
to 0.6 m. This is commercial imagery available
from DigitalGlobe - http//www.digitalglobe.com/
23
3.3 Hyperspectral Data

• Medium to high spatial resolution
• High spectral resolution by definition
• Many bands covering the EMR spectrum from 400 nm
  2500 nm
• Sometimes includes TIR (thermal infrared) bands
• Low to high cost (generally high)
• Poor spatial coverage with low cost data

24
3.3 Hyperspectral Data Uses

• Subregional to local applications
• Mineralogic mapping
• Lithologic mapping
• Structure mapping
• Soil-mineral anomaly mapping
• Geobotany/vegetation anomaly mapping
• Geologic mapping
• Environmental baseline mapping and change
  detection

25
3.3.1 AVIRIS (Airborne Visible/Infrared Imaging
Spectrometer)

• 224 channels, 10 nm band-widths across 400 nm
  2450 nm
• 20 m or better spatial resolution
• Archive data inexpensive (500.00/scene)
• Poor coverage - relatively few areas have been
  flown
• New data acquisition can be expensive
• Very useful
• Somewhat difficult to process
• US Government (NASA-JPL)
• http//aviris.jpl.nasa.gov/

26
3.3.1 AVIRIS ImageAVIRIS Full Image (airborne -
20 m Resolution 12.3 km x 10.2 km image 10
km wide)
VNirSWIR image. Good lithologic discrimination.
Vegetation highlighted as green.
SWIR image with kaolinite spectrum inset.
27
3.3.2 HyVista Hymap (Airborne)

• 126 bands covering 450 nm 2500 nm, 15-20 nm
  band-widths
• Spatial resolution variable to fit needs
• Relatively expensive
• Flights can be contracted
• Atmospherically corrected
• Geocorrected
• Very useful
• Commercial - HyVista Corporation
• http//www.hyvista.com/

28
3.3.2 HyMap Image
SWIR image with kaolinite spectrum inset. Image
is 1.3 km wide.
29
3.3.3 HyVista ARGUS

• 300 channels from 370 nm 2500 nm
• 100 TIR channels 7.8 µm - 12 µm
• Relative expensive
• Atmospherically corrected
• Geocorrected
• Likely to be extremely useful
• Commercial - HyVista Corporation

30
3.4 Radar

• Uses
• Geologic structure
• Interferometry (InSAR)
• Subsidence
• Crustal dynamics/tectonics
• High resolution topography (DEM)
• Vegetation studies (soil moisture content)
• Can be acquired through cloud cover
• L-band penetrates the surface a few meters
• Generally moderately priced

31
3.4 Radar Data Availability

• Radarsat 1 2
• 8 to 100 m spatial resolution
• C-band
• Commercial MDA
• http//gs.mdacorporation.com/

32
4.0 Software Examples

• RSI ENVI (particularly suited to hyperspectral
  data analysis)
• http//www.rsinc.com/envi/
• ERDAS Imagine (excellent vector GIS integration
  capabilities)
• http//gis.leica-geosystems.com/products/
• ER Mapper (superior geophysical data integration
  capabilities)
• http//www.ermapper.com
• PCI Geomatics
• http//www.pci.on.ca
• ArcGIS (primarily vector GIS, but handles raster
  images and raster-vector integration)
• www.esri.com
• Some free or low-cost software (search web)

33
5.0 Applications Geothermal Exploration

• Idealized remote sensing application
• Regional targeting
• Use inexpensive multispectral data for
• General hydrothermal alteration mapping
• Geologic structure mapping support
• Lithologic mapping support
• Use TIR data to identify thermal anomalies
  (ASTER)
• Select local targets
• Local assessment
• Hyperspectral data analysis (preferably high
  spatial resolution)
• Mineralogic mapping
• Soil-mineral anomaly mapping
• Identify pathfinder minerals
• Potentially identify blind geothermal systems
• Vegetation health anomaly mapping
• Identify hidden structures
• Potentially identify blind geothermal systems
• Lithologic mapping support

34
5.1 Multispectral Data Applications

• Hydrothermal alteration mapping
• Band ratios
• Spectral unmixing
• Structure mapping
• Use conventional photo-geologic methods
• Lineaments
• Lithology
• Tonal differences
• Band ratio enhancements
• Statistical analysis may help in some areas

35
5.1.1 Hydrothermal Alteration Mapping Example
(ASTER)

• Step 1 - Atmospheric correction
• Step 2 Image georegestration (may be done after
  ratios)
• Step 3 - Vegetation reduction
• generate vegetation index NDVI (TM4
  TM3)/(TM4 TM3)
• Set minimum vegetation threshold on NDVI
• Set all values gt threshold to 0
• Set all values lt threshold to 1
• Multiply through the original image
• All pixels above the threshold are be set to 0
• Step 4 - Cut out water bodies and clouds if
  present

36
5.1.1 Vegetation, Water and Cloud Masked
ImageExample (right)
37
5.1.1 Step 5 preprocessing is complete - time to
calculate band ratios

• Problem
• Need Kaolinite and Gypsum map
• Solution
• Divide high reflectance band(s) in masked imagery
  with absorption bands to create a 3 band false
  color composite ratio image

38
5.1.1 Know Your Data what bands should I use?

• ASTER VnirSWIR band intervals
• 1 0.520 µm 0.600 µm
• 2 0.630 µm 0.690 µm
• 3 0.760 µm 0.860 µm
• 4 1.600 µm 1.700 µm
• 5 2.145 µm 2.185 µm
• 6 2.185 µm 2.225 µm
• 7 2.235 µm 2.285 µm
• 8 2.295 µm 2.363 µm
• 9 2.360 µm - 2.430 µm

39
5.1.1 Find appropriate Absorption Features within
ASTER bands
B4
B5
B6
B7
B8
B9
Gypsum
Gypsum (yellow) and kaolinite (red) spectra with
ASTER SWIR bands approximately located. Band 5
and 6 are absorptive for both. Band 7 shows
absorption for gypsum, but not kaolinite.
Kaolinite absorptions2165 nm (Band 5)2205 nm
(Band 6)2315 nm (Band 8)2386 nm (Band 9)
Gypsum absorptions 2175 nm (Band 5)2215 nm
(Band 6)2265 nm (Band 7)
40
5.1.1 Do the math

• Rational
• divide a large number by a small number and
  derive a large number. Conversely divide a large
  number by a large number and derive a small
  number (enhances absorption features with small
  brightness values and subdues non-absorptive
  materials)
• Possible ratios for solution
• 4/5 (kaolinite and gypsum)
• 4/6 (kaolinite and gypsum)
• 4/7 (gypsum)
• Another possibility
• 7/5
• 7/6
• 4/7

41
5.1.1 Results
Raw ASTER image on the left and band ratio image
on the right with 4/6 as red 4/7 as green and
4/5 as blue. White areas are gypsiferous, magenta
areas are kaolinitic, red areas are
undifferentiated clays.
42
5.2 Hyperspectral Data

• Hydrothermal alteration mineral mapping
• Soil mineral anomalies
• Geothermal path finder mineral mapping
• Geobotanical anomaly mapping
• Vegetation stress detection as related to
  hydrothermal convection systems
• Structure mapping
• Use conventional photo-geologic methods
• Minimum noise fraction transform isolates texture
• Lithology
• Tonal differences
• Spectral differences

43
5.2.1 Hyperspectral Mineralogy Mapping

• Step 1 - High level atmospheric correction and
  conversion to reflectance required
• Step 2 - Linear spectral unmixing example (ENVI
  software)
• Data reduction
• Use only necessary bands
• Minimum noise fraction (MMF)
• Pixel purity index (PPI) generation
• Spectral end-member extraction
• Mixture-tuned matched filtering (MTMF)
• Field validation
• Nash, G. D. and G. W. Johnson, 2004,
  Hyperspectral detection of geothermal system
  related soil mineralogy anomalies in Dixie
  Valley, Nevada A tool for exploration.
  Geothermics, v. 33, issue 6, 695-711.

44
5.2.1 Hyperspectral Mineralogy Mapping MNF noise
free (left) and noise (right) images noise
bands can be removed prior to further processing
45
5.2.1 Hyperspectral Mineralogy Mapping Pixel
Purity Index Image
High values in the PPI image (bottom) show
pixels representing unique spectra (not
necessarily pure). The analyst can link the PPI
and reflectance images to check each spectrum.
Those considered to be relatively pure
end-members are used as a guide to create a
spectral library from MNF spectra.
46
5.2.1 Hyperspectral Mineralogy Mapping Spectral
Analyst (spectra correlation analysis)
The Spectral Analyst can be used to match an
image spectrum with those similar in a spectral
library. This can aid the analyst in determining
spectral end-members, However, the spectra with
the highest correlation values are not always
correct. The analyst must carefully interpret the
results to assign the end-member to the correct
mineral.
A pure kaolinite library spectrum (dashed) shown
with an image spectrum (solid).
47
5.2.1 Hyperspectral Mineralogy Mapping
Mixture-tuned matched filtering (Relative
Abundance Mapping)
Spectral end-members are used as input for MTMF
processing. The MTMF processing results in an MF
score image (top) and an infeasibility image
(bottom). Where MF scores are high and
infeasibility scores are low there is a good
probability that the spectral end-member has been
mapped correctly. This image represents kaolinite
from AVIRIS data
48
5.2.1 Geothermal Path Finder Minerals

• A few examples of what you can map
• Ammonium minerals
• Buddingtonite
• Ammonioalunite
• Others
• High temperature alteration minerals
• Alunite
• Pyrophyllite
• Others
• Ferrihydrite (ferric gels)
• basin/playa margins
• Sulfate mineral examples
• Thenardite
• Mirabilite
• Others

49
5.2.2 Soil-Mineral Anomaly Mapping

• Anomaly examples
• Calcium carbonate, kaolinite, other alteration
  minerals
• Buried hot spring deposits
• Buried fumarole deposits
• Buried hydrothermally altered rock associated
  with faults
• Sulfate minerals
• abundant in Emigrant prospect, Nevada
  associated with H2S fumarole
• Borate minerals
• May indicate buried structures
• May indicate past surficial geothermal activity
• Will be useful in detecting blind systems
• Nash, G. D. and G. W. Johnson, 2004,
  Hyperspectral detection of geothermal system
  related soil mineralogy anomalies in Dixie
  Valley, Nevada A tool for exploration.
  Geothermics, v. 33, issue 6, 695-711.

50
5.2.2 Soil-mineral anomaly mappingExample
C calcium carbonate. K kaolinite. This
soil-mineral anomaly is spatially correlated with
a covered piedmont faults. Background is a
calcite relative abundance image.
51
5.2.3 Vegetation Health Anomaly Mapping

• Vegetation stress/health anomalies can be related
  to
• Permeable faults
• Geothermal reservoir degassing
• Gas toxicity
• Soil acidification
• Hydrothermal mineralization
• Toxic minerals
• Thermal anomalies
• Non-geothermal phenomena

52
5.2.3 Vegetation Health Anomaly Mapping

• Field spectroscopy
• Use in areas with low to high vegetation density
• Airborne hyperspectral data
• Use in areas with medium high to high vegetation
  density

53
5.2.3 Vegetation Health Anomaly Mapping
54
5.2.3 Vegetation Health Anomaly Mapping

• Spectral parameters indicating stress/damage
• Spectral position of visible green maximum
  reflectance (Red-shift)
• Point of inflection of the red-edge (blue-shift)
• 699nm/765nm ratio (relatively high values)

55
5.2.3 Vegetation Health Anomaly Mapping
Vegetation health anomaly map for the Cove
Fort-Sulphurdale, Utah thermal anomaly. Note the
spatial relationship of the anomalies to faults.
56
5.2.3 Vegetation Health Anomaly Mapping
Cover Fort-Sulphurdale vegetation health
anomalies in relation to faults mapped from
geophysical data. Nash, G. D., J. N. Moore, and
T. Sperry, 2003. Vegetal-spectral anomaly
detection at the Cove Fort-Sulphurdale thermal
anomaly, Utah, USA implications for use in
geothermal exploration. Geothermics, v. 32, p.
109-130.
57
5.3 Thermal Anomaly Mapping

• ASTER TIR data can be used
• What is needed
• Daytime kinetic temperature data
• Nighttime kinetic temperature data
• Albedo image
• can be created from ASTER multispectral data
• Digital elevation model (DEM)

58
5.3 Thermal Anomaly Mapping

• Apparent Thermal inertia is calculated
• ATI (1 Albedo) / ?T
• Where albedo reflectance in daytime visible
  data and ?T change in temperature between the
  day and night thermal imagery
• ATI and nighttime kinetic temperature used in
  combination
• Highest ATI values generally indicate areas that
  are not anomalous
• High nighttime kinetic temperature values that
  correlate spatially with medium to low ATI values
  may indicate thermal anomalies (areas to field
  check)
• Dudley, E.A. and G.D. Nash, 2003. Using Thermal
  Infrared (TIR) Data to Identify Geothermal
  Anomalies. Geothermal Resources Council
  Transactions vol. 27, 645-647.

59
ATI (left) and Nighttime Kinetic Temperature
(right) Images
These images can be geographically linked in most
image processing software packages so that pixel
values can be precisely compared. Image
normalization and differencing can be used to map
thermal anomalies.
60
5.4 Geologic Mapping

• Data fusion for superior base mapping imagery
• Increase spatial resolution of spectral data
• Input data
• ASTER multispectral (15 m 30 m)
• all input data resampled to 15 m
• Digital Orthophotos (DOQs 1 m)
• Several fusion methods available
• Brovey method used in this example
• (A1/(A1A2A3)) x DOQ New band 1
• (A2/(A1A2A3)) x DOQ New band 2
• (A3/(A1A2A3)) x DOQ New band 3
• Where A the ASTER multispectral input band
  number

61
5.4.1 Brovey Fused Imagery
The top image is a 15 m spatial resolution ASTER
image. The bottom image shows an inset of 1 m
spatial resolution fused data lying on the
original ASTER image. The white box in the top
image shows the area of the bottom image.
62
5.4.2 Geologic Mapping Base 120,000
Brovey fused mapping base. The UTM grid is used
in the field with a GPS to improve map spatial
accuracy. Note the excellent discrimination of
rock types. This area has light vegetation cover.

63
5.4.3 Geologic Mapping Base 16,000
Lithologic contacts and faults can be added as
polygons and lines in a GIS system prior to field
work. Large scale map base images are printed
with an inkjet using glossy photo quality paper
for field use (excellent). Editing can easily be
done after field validation and mapping using GIS.
64
5.4.4 Result
65
Remote Sensing Data Sources

• Multiple Platform
• EROS data center (USGS) http//edcwww.cr.usgs.gov
  /
• Airborne
• HyVista http//www.hyvista.com/
• HyMap hyperspectral
• ARGUS hyperspectral including TIR
• Hyperspectral Data International
  http//www.hdi.ns.ca/
• CASI hyper- and multispectral
• SpecTIR http//www.spectir.com/
• HyperSpecTIR (HST) hyperspectral
• Satellite
• Space Imaging http//www.spaceimaging.com/
• IKONOS 1m panchromatic
• IKONOS 4m multispectral
• IRS (Indian Remote Sensing) 5m panchromatic
• IRS Multispectral 23.5m
• Others