Geothermal exploration using remote sensing techniques

In The Name Of God
Remote Sensing Applications in
Geothermal Studies
Supervisor : Dr Akhoondzadeh
Presented by : Sepideh Abadpour
University Of Tehran
Faculty Of Engineering
Department Of Surveying And Geomatics Engineering

Goal:
A brief study on how to use satellite images in order to :
Discover capable geothermal regions for generating
power plants
Discover different characteristics of geothermal
regions with the help of remote sensing techniques
Also we can have a brief discussion on various applications
of geothermal resources , please notice background
images and ask your questions.

A brief literature review:
Early studies:
Hodder in 1970 used visible and IR photogrammetric
images
Lee in 1978 used thermal band images of LANDSAT
Recent studies:
Coulbaugh in 2003 proposed to use TIMS and AVIRIS
images to identify geothermal zones then he developed
his ideas in his PHD thesis in 2007 using ASTER images.

A brief literature review:
Recent studies:
Maccune & Koen proposed a new way to omit
topographic affect from thermal images
Savage in 2009 used LANDSAT images
Iranian Studies:
There has not been a major effort
Tavakkoli in 1380 tried to use statistical analysis on
AVHRR images and concluded that remote sensing
techniques are not sufficient and we do need field data.

Common ways:
Magnettotellurics ( MT)
Time Electromagnetics ( TEM )
DC Resistivity ( Direct Current Resistivity )
Seismic Tomigrapghy
Disadvantages:
Time consuming and expensive
Troubles to access all regions and drill borehole well
(because of hard topograpy)

Using multispectral
images to identify
regions which have
thermal anomalies
Geothermal minerals
exploration using
multispectral data
Geothermal exploration
with hyperspectral data
Monitoring and mapping
land deformation in
geothermal areas using
SAR interferometry
My studies
concentration

Studying thermal anomalies caused by
geothermal activities using multispectral
images

Thermal anomalies :
1 – Topography
2 – Albedo
3 – Thermal Inertia
4 – Internal heat resources
( radioactive isotopes and
geothermal energy)

Anomaly = Modeled Temperature – Observed Temperature
Ways to calculate LST:
1 – TES ( ASTER )
2 – NEM
3- Alpha Residual
4 – Thermal Infrared Single Band
5 – Split Window
6 – Day & Night ( MODIS)

Problems LST
algorithms deal with:
Inaccurate atmospheric corrections
Accuracy depends on:
1. Accuracy of radiation transfer
models
2. Atmospheric parameters such
as exact amount of water vapor
in the atmosphere
Emissivity diversity in different spectral
bands and different materials of the
earth surface

Ways to reduce
atmospheric effects:
1 – Simply neglect it (classification)
2 – calibrate the data with the help of field
measurements taken at the time of satellite’s
transit (phenomenon with weak reflectance )
3 – Model a hypothetical atmosphere based
on parameters like latitude and longitude,
flight height and etc
4 – Model a hypothetical atmosphere based
on satellite's measurements taken at data
acquisition time
5 – Create several different views of a single
object or measure a phenomena in different
spectral bands

Split – window:
Surface emissivity is variable for different types of land cover

Split – window:
But in 10.5 – 12.5 micrometer range of wavelength, the
characteristics of emissivity is relatively constant
Bands 31 and 32 of MODIS are in this range

Conclusions:
1 – All of materials have an emissivity
more than 0.825 in bands 31 and 32
2- We have a general equation -0.012 ≤ ɛ32
- ɛ31 ≤ 0.023 for all of the cases except
fresh rocks, melting ice and deciduous
trees
3- Surfaces of vertical leaves have more
emissivity than surfaces of horizontal
leaves
4 – Emissivity is in the range 0.97 – 0.99 for
most types of land covers except rocks and
sands

Split – window:
So, emissivity is relatively constant in the bands 31 and 32
of MODIS
General form of the algorithm:
Ts = Land surface temperature
T4 & T5 = Brightness temperature of the bands 4 & 5 of the AVHRR
sensor
A & B = Coefficients related to atmospheric effects, incidence angle and
surface emissivity which are calculated empirically

Ways to
calculate
A & B
Use atmospheric
profiles simulator
software like
LOWTRAN 7
Using field
measurements

So , we can always enhance the split window algorithm
Price ( 1984 )
Becker and Li ( 1990 )
Sobrino and Caselles ( 1991 )
Coll et al ( 1994 )
Becker and Li ( 1995 )
Wan and Dozier ( 1996)

Split – window:
The equation above is specialized for the MODIS satellite
The coefficient A1 is not constant and is given as a table in different
atmospheric conditions and incident angles
We can always improve split-window algorithm by improving our
databases about the coefficients in different water vapor atmospheric
conditions and different incidence angles
In general developing split-window algorithms depends on exact
information about emissivity for surfaces in bands 31 and 32 of MODIS
The more accurate emissivities will result more accurate temperatures
We will use emissivities that are calculated within NDVI-based methods

Ways to estimate
emissivity:
1 – Temperature Independent Spectral Indices
2 – Band Radiance Estimation
3 – Reference
4 – Normalization
5 – Renormalization
6 – Spectral Ratio
7 – Alpha Emissivity
8 – Normalization Difference Vegetation Index
9 – Classification – based Emissivity

Normalization Difference Vegetation Index ( NDVI )
NDVI is calculated based on radiance in visible and near infrared
red wavelengths:
ƥ1 and ƥ2 are the reflectance in bands 1 and 2 of the AVHRR sensor
NDVIC is corrected from the effects induced by solar zenith angle and
atmosphere
ɛ4 and ɛ5 are the emissivities in bands 4 and 5 of AVHRR

Normalization Difference Vegetation Index ( NDVI )
The manner supposes that NDVI and emissivity is constant for
each pixel over the day
But notice that moisture and rain can change emissivity
So NDVIC is the maximum emissivity over the entire day
The equation may not be useful for bare soils
But it is useful for vegetation covered regions
You can replace bands 31 and 32 of MODIS by bands 4 and 5 in
the above equations

LST modeling
Weather
temperature-based
Daolan Zheng
Kang
Energy
balance -based
Coolbaugh
Stephan
Savage

Modeling Land Surface Temperature ( LST )
Weather temperature – based : They can not be used
in Iran because of lack of meteorological information
Energy balance – based : Mr.Zohary has to the point
that Stephan is the best for modeling LST, so we will just
cover that

Stephan Model:
The energy which remains in the body and is spent to increase
its temperature is calculated as:
Where I is the solar insolation
i : The angle between the incident ray and the normal to the
surface

Stephan Model:
A, B and C are coefficients calculated for each region
considering the solar energy reached to the earth
T : Surface Temperature
Ɛ : Emissivity
δ: Stephan – boltzman constant
Parameter X determines the energy skipped from the surfaces
due to processes like conducting with neighbor pixels, weather
around and the below surface

Stephan Model:
The body’s internal energy changes causes decreasing or
increasing of its temperature over the time which is presented
by :
m : Mass of the body
A : Area of the body
C : Heat capacity of the body

Stephan Model:
Substituting the above in the first equation, yields :
Due to the fact that the parameter observed in remote sensing
is the brightness temperature:

Stephan Model:
We’ll need to solve a first – order differential equation in order
to calculate the modeled temperature from the equation above

Geothermal Exploration Using Remote
Sensing in the South of Baja California
Sur, Mexico

Objective : Define favorable area for utilization an
exploration of geothermal energy
Studied Region : In Baja California Sur, the area from
Ciudad Constitucion to Los Cabos was studied using a mosaic
of four Landsat ETM+ images.
Atmospheric corrections has been done to each single image
before pasting it to the mosaic

Problem :
Landsat ETM+ images is useful for :
Identification of hydrothermal alteration mineral like
oxides and hydroxyls
Identification of main geological structures like faults
and fractures
But spectral response of vegetation masks the response of
minerals
Vegetation has high reflectance in the middle and near
infrared just like hydroxyls.

Solution:
Use image processing to :
Enhance oxides and hydroxyls spectral features
Subdue vegetation spectral characteristics
Field Verification would be necessary
Some hydrothermal minerals are also observed in weathered
rocks, so the results are just a guideline for field verification

Methodology:
Six of the seven bands ( 1, 2, 3, 4, 5 and 7 ) of the Landsat
ETM+ images were used
Enhance oxides and hydroxyls spectral features
Subdue vegetation spectral characteristics
Field Verification would be necessary
Some hydrothermal minerals are also observed in weathered
rocks, so the results are just a guideline for field verification

Spectral Enhancement:
The table shows :
Oxides have high reflectance in the red band ( ETM3 ) and high
absorption in the blue band ( ETM1 )
Hydroxyls have high reflectance in the middle infrared ( ETM5 )
and high absorption in far infrared ( ETM7 ) hence we have not used
the thermal infrared band ( ETM6 )
Vegetation have high reflectance in the near infrared ( ETM4 ) and
high absorption in the red band ( ETM3 )

Spectral Enhancement:
Done by band ratios and band subtractions
The image in the background has RGB – 432 color
composition to enhance and visualize vegetation in red

Hydroxyle
band ratios present some areas in the image with very bright
pixels, they give a better contrast, but band subtraction
achieves better results for the whole image.

With the three band ratios and three band subtractions, two
color composites were done. The best results were obtained by
the color composite made with the band subtraction

Spatial Enhancement:
Done to heighten geological structures like faults and
fractures by photointerpretation
Five filters were applied:
Laplacian
Sobel Edge Enhancement
Three directional filters
oNW
oNE
oE

The results for the Laplacian and Sobel filters

Several images result after the application of the filters. These images
were analyzed and structural features were recognized and draw by
photointerpretation

Conclusions:
Vegetation is more abundant than expected in the southern
tip
band ratio is a non-linear operation and band subtraction is
a linear operation. This is important because a contrast
stretch will be required for the images resulting from the
band ratios
band difference extracts spectral contrast between two
bands, but it does so in a linear way. The resulting image can
then be linearly stretched, with no loss of information

The color composite above is for band subtractions

Conclusions:
Band 4 minus 3 made for enhancement of vegetation is red
Band 3 minus 1 made for enhancement of oxides is green
Band 5 minus 7 made for enhancement of hydroxyls is blue
Vegetation is well identified with the color composite RGB-
432 shown in the background
Band ratios did not work for the objectives of this research
probably because of nonlinear math and subsequent loss of
information
The band subtraction (RGB-(4-3) (3-1) (5-7)) gave the finest
results

Conclusions:
As it has been observed that some of the hydrothermal
minerals are also related with weathering, the areas marked
as hydrothermally altered must be taken as guidelines for
future field work.

Geothermal exploration using remote sensing techniques

More Related Content

What's hot (20)

Similar to Geothermal exploration using remote sensing techniques (20)

Recently uploaded (20)

Geothermal exploration using remote sensing techniques