Content uploaded by Najam Ul Hassan Syed
Author content
All content in this area was uploaded by Najam Ul Hassan Syed on Nov 18, 2018
Content may be subject to copyright.
Spectral quality assessment of Landsat 8 and Sentinel 2 bands for glacier identification in
Upper Indus Basin
Syed Najam ul Hassan (1,2,3), Mohd Nadzri Md. Reba (1,2), Dostdar Hussain (3), Aftab Ahmed (3)
1Geoscience & Digital Earth Centre (INSTeG), Research Institute for Sustainability & Environment
(RISE), Universiti Teknologi Malaysia, 81310 Johor Bahru, Johor, Malaysia.
2Faculty of Geoinformation and Real Estate, Universiti Teknologi Malaysia, 81310 Johor Bahru, Johor,
Malaysia.
3Department of Computer Science, Karakoram International University, Pakistan.
Email: syed.najam@kiu.edu.pk; nadzri@utm.my; dostdar.hussan@kiu.edu.pk; aftab.ahmed@kiu.edu.pk
*corresponding; syed.najam@kiu.edu.pk
KEY WORDS: Glaciers, Upper Indus Basin, Sen2Core, Operational Land Imager, Multi Spectral Instrument.
ABSTRACT: Glacier studies of Hindu Kush Karakoram Himalaya (HKKH) are inadequate where, the stability of
glaciers in the Upper Indus Basin (UIB) of HKKH is known for anomaly studies. Despite of satellite based synoptic
measuring schema, the quality of glacier anomaly estimate is always on debate. The advancement in Operational
Land Imager (OLI) and Multi Spectral Instrument (MSI) offers the potential future of glacier measurement in UIB.
Therefore, this study assesses the quality of OLI and MSI in mapping the glacier anomaly for glaciers of Hunzza in
UIB. The methodology is based on acquisition of OLI Level 2 data, while for Sentinel MSI Level 2A data was derived
using Level 1C, however, considering Landsat Enhanced Thematic Mapper Plus (ETM+) is motivation to support
the sensors for their calibrated data with uncertainty of 3% as compare to 5% of the raw ETM+. Glacier outlines
extracted from the Randolph Glacier Inventory and the snow line altitude (SLA) demarcated through contour
generation from Global Digital Elevation Model (GDEM) to differentiate permanent snow and clear ice in the overall
glacier polygon. Reflectance of each band was derived and Normalized Snow Differential Index (NDSI) calculated.
Statistics applied in spectral quality assessment for glacier parameters. Overall glacier surface exhibited range of
reflectance about 0.08 to 1.39, 0.07 to 1.39 and 0.04 to 1.44 at visible bands of OLI that was differed about 27%,
29% and 25% than that of MSI. Where, for SWIR band both sensors agreed by the mean reflectance of 0.10.
Reflectance correlation between both sensors derived as 0.73 to 0.80 at visible band and 0.31 to 0.38 at SWIR which,
allows clear discrimination between the clear ice and snow. But the overlap of reflectance within 0.20 to 0.35 in
infrared bands of MSI may led to erroneous identification. To complement the results, NDSI of OLI with -0.01 to
0.95 becomes good indicator to distinguish different glacier features with disadvantage of inconsistent in MSI. These
results clearly show that OLI and MSI have promising capability to map glacier anomaly and both variants can be
synergized for better interpretation in climacterically intrinsic high-altitude zone of UIB.
1. INTRODUCTION
Glacier is the best climate indicator that is responsive to even very small climate variation. Recent climate change
issue is completely related to the ice melting on glaciers and as a result, the glacier extends and their morphology
changes were interconnected to climate conditions (Paul et al., 2015). The Hindu Kush Karakoram Himalayas
(HKKH) is a home to wide range of highest mountain glaciers and vulnerable to unprecedented climatic change
(Najam ul Hassan, Md Reba, Hussain, & Ali, 2018). International and regional initiatives have put full attention on
HKKH and glacier inventory is a major concern. Remote sensing plays important role in glacier studies in last two
decades yet the inconsistent findings about glacier behaviors were evident in different ranges of this region. For the
HKKH, glaciers in the Upper Indus Basin (UIB) known as Karakoram Pamir Anomaly is the most interesting site
due to its stability behavior (A. Racoviteanu, 2011; Gardelle, Berthier, Arnaud, & Kääb, 2013). Despite of single
mission optical remote sensing by Landsat has been exploited for continuous glacier monitoring, there are some
technical limitation in spectral, spatial, temporal and radiometric resolution were reported in last few decades (Paul,
Winsvold, Kääb, Nagler, & Schwaizer, 2016); (Zhu, Woodcock, Holden, & Yang, 2015).
Recently, new generation optical sensors of Operational Land Imager (OLI) and Multispectral Instrument (MSI) with
12-bit quantization provide higher quality land surface monitoring application (Mandanici & Bitelli, 2016) and are
recommended for glacier monitoring mission by exploiting the reflectance properties of these multispectral sensors
for glacier surface extraction (Paul et al., 2016). The improved spectral and spatial resolution of OLI and MSI which
are far better than the previous Landsat variant offer an opportunity to apply for UIB glaciers monitoring (Minora et
al., 2013). The improved optical properties of OLI and MSI enhance the glacier extraction method such as Normalised
Differential Index (NDSI) by offering multiple reflectance at different optical bands so that different glacier faces
can be mapped at higher accuracy (A. E. Racoviteanu, Paul, Raup, Khalsa, & Armstrong, 2009; Minora et al., 2016).
Currently, the spectral reflectance properties assessment of both sensors and potential bands combination for NDSI
were understudied in line with the UIB climatic dependent mountainous glaciers research. Therefore, the study
focuses to determine the significant spectral band of OLI and MSI that can be used for snow and ice indexing. The
objectives of this study are (1) to assess the spectral quality of both sensors by comparing the reflectance estimates
at different band and (2) to develop the NDSI variant of which the best visible and short wave infrared (SWIR) bands
used to distinguish different mountainous glacier features in UIB were defined. The relationship of selected bands is
the major highlight towards the future research for data fusion in glacier mapping.
2. STUDY AREA
For this study, one of the river basins in the Upper Indus Basin (UIB) named Hunzza sub basin which located in north
east of HKKH is selected. The Hunzza sub-basin covers an area of 13715 km2 and about 25% of this area is the
glacier of UIB that contributes water discharge to the Indus river. Almost 15% of the glaciated area is mainly
distributed in the entire mountainous range of Karakoram and enclosed with heterogeneous glacier behavior (T.
Bolch et al., 2012; Quincey & Luckman, 2014; Tobias Bolch, Pieczonka, Mukherjee, & Shea, 2017). The study
focuses on mountainous glaciers layout shown in Figure 1.
Figure 1: Image of Sentinel-2 of MSI (purple outline) and Landsat 8 of OLI (green outline) show the glacier and snow
cover with the corresponding GDEM presenting the various glacier altitudes of HKKH. Glacier outline in red line taken
from RGI gives visual polygon of mountainous glacier used as the main reference in this study.
3. MATERIAL AND METHODOLOGY
The methodology in this study is designed to assess OLI and MSI bands for their spectral quality to monitor the
glacier surfaces in mountainous area of UIB. Selected glacier (depicted in Figure 2) of Hunzza sub basin with
diversified elevation profile was used for intrinsic discussion for this study.
3.1 Data Acquisition
Glacier outlines showing the latest extend in year 2005±3 (Bajracharya et al., 2015) were obtained from glacier
inventory of Randolph Glacier Inventory (RGI) and Global Land Ice Measurements from Space (GLIMS) (RGI
Consortium, 2015). While selected sample glaciers from the Hunzza Sub basin is debris cover glaciers (Khan, Naz,
& Bowling, 2015) covering an area of 34.30 km2 however, the elevation range of overall glacier is 2700 ~ 7700
meters (Figure 1) with a heterogenous nature of terrain which consists debris cover, clear ice and snow and proved
to be best sample to test and compare both sensors for mountainous glacier studies. The permanent snow and clear
ice of overall glacier were differentiated by referring Snow Line Altitude (SLA) the delineated with the Global
Digital Elevation Model (GDEM) at the accuracy of 17 meters ((A. Racoviteanu, 2011) GDEM Version 2, 2011).
Level 2 scientific data of OLI and Level 1C data of MSI acquired during the end of ablation period of 2017 are the
primary data in this study of which radiometrically and geometrically were corrected and formed in Top of
Atmosphere (TOA) reflectance product (USGS, 2018; ESA, 2015) and the snow cover was at minimum during the
period of acquisition (Khan et al., 2015). To determine the spectral matching between MSI and OLI bands, the
Relative Spectral Response Function (RSRF) was downloaded from European Space Agency (ESA) and the National
Aeronautics and Space Administration (NASA) web portals respectively. Table 1 provides the comprehensive
description of the data acquired for this study.
Table 1: Detail description of data for the study.
Sensor/Sou
rce
Type of
data
Bands
Resolution
RSRF*
Acquisition
Time
Band
Band No
OLI
Optical
imaging
Blue
B2
30 m
0.60
18
September
2017
Green
B3
30 m
0.56
Red
B4
30 m
0.54
NIR
B5
30 m
0.38
SWIR
B6
30 m
0.45
SWIR
B7
30 m
0.56
MSI
Optical
imaging
Blue
B2
10 m
0.62
20
September
2017
Green
B3
10 m
0.68
Red
B4
10 m
0.76
NIR
B8A
10 m
0.64
SWIR
B11
10 m
0.75
SWIR
B12
10 m
0.67
GDEM
Microwa
ve
imaging
-
-
30 m
-
11 February
2000
Glacier
outlines
Vector
database
-
-
-
-
2005±3
*RSRF as function taken as radiance response taken at pre-calibration stage.
3.2 Data Preparation
All satellite images were transformed into new geographical projection of Universal Transverse Mercator (UTM) to
minimize the geometric distortion when comparing to the local DEM map (Paul et al., 2015). Comparison between
MSI and OLI requires a resampling of MSI imageries by aggregating the pixels at raw spatial resolution of 10 m to
a single 30 m pixel. The raw imageries of MSI and OLI are presented in the digital count and must be converted into
the truth Top of Atmospheric (ToA) reflectance within the range of 1 to 10 by using the quantification number
expressed in Eq.(1). The quantification number was used to reduce the processing capacity during the data retrieval
via online and it is provided in the metadata of each image (European Space Agency, 2015); (USGS, 2018a).
(1)
where,
is the TOA reflectance of the respective wavelength, , the respective spectral band with the pixel
location of (i,j) and the correction of solar angle for each sensor (k), and is the digital count provided in each
sensor data where ToA is computed and valid between 0 to 104.
The RSRF was used to redefine the quality of band similarity between both sensors and it is representing the “true”
spectral response of each band estimated during the pre-calibration process (Mandanici & Bitelli, 2016). The
similarity bands between both sensors was assessed by using histogram and the best band matching is later applied
in ice and snow indexing.
3.3 The Normalized Differential Snow Index (NDSI)
The optimum reflectance (peak) is determined and visualized completely from RSRF and based on this result, the
regression between band of OLI and MSI at the visible and the short-wave infrared (SWIR) wavelength is carried
out. To examine more on the range of reflectance, the minimum and the maximum of the spectral range is determined
so that the exact spectral overlapped is completed distinguished. From this band selection, list of similar spectral
response of wavelength at visible to SWIR is extracted.
Band ratio is commonly used in satellite data processing for glacier mapping and in this study, the Normalised
Differential Snow Index (NDSI) was applied. The NDSI is based on the ratio of band combination between visible
and SWIR of which the visible bands give higher spectral response on snow and the SWIR reflectance remains higher
even in infrared region (A. E. Racoviteanu, Williams, & Barry, 2008). It also has been proved as the promising tool
in delineating glaciated regions (Khan et al., 2015). The NDSI was extracted within the glacier polygon based on the
following expression.
(2)
where
is the Normalized Differential Snow Index (NDSI) of each pixel location, (i,j), for sensor, k (OLI or
MSI), is the reflectance at visible band and is the reflectance at the short wave infrared band. All
reflectance is presenting the total reflectance at ToA. From both sensors, there are several numbers of NDSI formation
and this study examines the best VIS-SWIR combination using statistical evaluation.
4. RESULTS AND DISCUSSION
The RSRF of OLI and MSI dataset suggests that most of the bands in both sensors is not identical with mere
radiometric difference (Mandanici & Bitelli, 2016). Though there are significant spectral match between
corresponding bands at visible, near infrared and SWIR wavelength (Zhang et al., 2018) and these bands have
potential to exploit further in glacier extraction through NDSI. Therefore, the histogram of all possible combination
of visible (blue, green and red) and SWIR bands was designed and shown in Figure 4. By using the glacier outline of
GLI, the reflectance of OLI in visible and SWIR spectra varies between 0.07 to 1.44 and 0.0 to 1.37 respectively
while MSI is from 0.07 to 1.41 and 0.02 to 1.37 respectively.
(a)
(b)
(c)
(d)
Figure 2: Histogram of pixel reflectance at visible bands in (a) Blue, (b) Green, (c) Red wavelengths and at (d)
SWIR band for OLI (in dashed red line) and MSI (in solid black line).
0
200
400
600
800
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40
Pixel Freq
Reflectance
MSI (B2)
OLI (B2)
0
200
400
600
800
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40
Pixel Freq
Reflectance
MIS (B3)
OLI (B3)
0
200
400
600
800
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40
Pixel Freq
Reflectance
MSI (B4)
OLI (B4)
0
200
400
600
800
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40
Pixel Freq
Reflectance
MSI (B11)
OLI (B6)
Later, the NDSI was computed for all combination of visible and SWIR bands with variation range of -0.01 to 0.95
and -0.05 to 0.97 for OLI and MSI sensor respectively. This result is clearly illustrated in Figure 5 for visible and
first SWIR combination and Figure 6 is for visible and second SWIR combination.
(a)
(b)
(c)
Figure 3: NDSI Trend for possible combinations of VIS and SWIR-1 bands for OLI and MSI (a) Blue & SWIR-1 (b)
Green & SWIR -1 (c) Red & SWIR-1.
(a)
(b)
(c)
Figure 4: NDSI Trend for possible combinations of VIS and SWIR-2 bands for OLI and MSI (a) Blue & SWIR-2 (b)
Green & SWIR -2 (c) Red & SWIR-2.
A comprehensive summary of reflectance range for individual band of OLI and MSI is tabulated in Table 2. The
result suggests that OLI and MSI reflectance have difference of 27%, 29% and 25% in their respective visible bands
of blue, green and red, however, the difference in infrared spectrum is considerably high with mean reflectance of
0.10 in SWIR. It also noted that MSI has explicit reflectance range of 0.02 to 1.35 and 0.00 to 0.35 in NIR and SWIR
respectively which is an advantage to discriminate glacier extends while overlapping reflectance values in said range
(0.02 to 0.35) may lead to an erroneous identification.
Table 2: Reflectance range of OLI and MSI over the glaciated surface of Hunzza sub basin.
Landsat 8 (OLI)
Sentinel-2 (MSI)
Bands (ID)
Reflectance
Bands (ID)
Reflectance
Blue (B2)
0.08 to 1.39
Blue (B2)
0.07 to 1.38
Green (B3)
0.07 to 1.39
Green (B3)
0.05 to 1.36
Red (B4)
0.04 to 1.44
Red (B4)
0.05 to 1.41
NIR (B5)
0.02 to 1.37
NIR (B8A)
0.02 to 1.35
SWIR-1 (B6)
0.00 to 0.24
SWIR-1 (B11)
0.00 to 0.25
SWIR-2 (B7)
0.00 to 1.20
SWIR-2 (B12)
0.00 to 0.19
Table 3 summarises the NDSI range for all possible combinations in which the maximum pixel frequency in OLI and
MSI image was determined (Dorothy K Hall & George A Riggs, 2011). Those NDSI ranges are explicit and thus
they are promising advantage to discriminate the glacier extents particularly for glacier surface observed with pixel
shift between both sensors.
0
200
400
600
800
0.00 0.20 0.40 0.60 0.80 1.00
Pixel Freq
Reflectance
MSI (B3 & 11)
OLI (B3 & 6)
0
200
400
600
800
0.00 0.20 0.40 0.60 0.80 1.00
Pixel Freq
Reflectance
MSI (B2 & 11)
OLI (B2 & 6)
0
200
400
600
800
0.00 0.20 0.40 0.60 0.80 1.00
Pixel Freq
Reflectance
MSI (B4 & 11)
OLI (B4 & 6)
0
200
400
600
800
0.00 0.20 0.40 0.60 0.80 1.00
Pixel Freq
Reflectance
MSI (B4 & 12)
OLI (B4 & 7)
0
200
400
600
800
0.00 0.20 0.40 0.60 0.80 1.00
Pixel Freq
Reflectance
MSI (B3 & 12)
OLI (B3 & 7)
0
200
400
600
800
0.00 0.20 0.40 0.60 0.80 1.00
Pixel Freq
Reflectance
MSI (B2 & 12)
OLI (B2 & 7)
Table 3: Normalized Differential Snow Index (NDSI) for the selected glacier in Hunzza
Sub Basin.
Landsat 8 (OLI)
Sentinel-2 (MSI)
Band Combination
NDSI
Band Combination
NDSI
B2 & B6
-0.10 to 0.95
B2 & B11
-0.13 to 0.96
B2 & B7
-0.02 to 0.95
B2 & B12
-0.05 to 0.97
B3 & B6
-0.09 to 0.94
B3 & B11
-0.13 to 0.96
B3 & B7
-0.01 to 0.95
B3 & B12
-0.06 to 0.97
B4 & B6
-0.08 to 0.94
B4 & B11
-0.15 to 0.96
B4 & B7
-0.01 to 0.95
B4 & B12
-0.07 to 0.97
To assess the relation between matching band of OLI and MSI, correlation of pixels representing the glacier was
generated and shown in Figure 7(a). Here, the correlation between 0.72 to 0.80 is evident for visible band and this
not a case for SWIR band with the lower correlation between 0.31 to 0.38. However, the correlation in NIR band is
higher than SWIR and such correlation (0.860) is consistently existed with visible band and this suggests that the
NIR and visible bands are responsive to the same snow reflectance.
(a)
(b)
Figure 5: Plot of (a) degree of correlation for each OLI-MSI band combination from visible to SWIR band; and histogram
of pixels representing the selected glacier which later being used for correlation estimation in (a).
5. Conclusion
This study was carried out to assess the spectral quality of OLI and MSI bands for mountain glacier extend mapping.
Pixels of each band of both MSI and OLI located within the selected glacier of RGI and specific GDEM were taken
as the primary sample. Prior to data processing, the pre-calibration data of reflectance namely RSRF dataset for both
sensors were assessed to determine the degree of similarity between bands. Highly attention is put to visible and
SWIR bands as they are going to be used in the snow indexing of NDSI. The ToA reflectance in visible bands has
slight difference in spectral estimate but it is much higher in SWIR band. The correlation at visible band is higher
than the SWIR and this gives prospective for accurate clear ice and snow discrimination. The results have shown both
sensors provide promising spectral response for monitoring the mountainous glaciers.
6. References
Bajracharya, S. R., Maharjan, S. B., Shrestha, F., Guo, W., Liu, S., Immerzeel, W., & Shrestha, B. 2015. The
glaciers of the Hindu Kush Himalayas: current status and observed changes from the 1980s to 2010.
International Journal of Water Resources Development, 31(2), 161–173.
https://doi.org/10.1080/07900627.2015.1005731
Bolch, T., Kulkarni, A., Kaab, A., Huggel, C., Paul, F., Cogley, J. G., … Stoffel, M. 2012. The State and Fate of
Himalayan Glaciers. Science. https://doi.org/10.1126/science.1215828
Bolch, T., Pieczonka, T., Mukherjee, K., & Shea, J. 2017. Brief communication: Glaciers in the Hunza catchment
0.73 0.72 0.80 0.86
0.38 0.31
0.00
0.20
0.40
0.60
0.80
1.00
Blue Green Red NIR SWIR-1 SWIR-2
Correlation
0
200
400
600
800
Blue Green Red NIR SWIR-1 SWIR-2
Pexil Freq
OLI
MSI
(Karakoram) have been nearly in balance since the 1970s. Cryosphere, 11(1). https://doi.org/10.5194/tc-11-
531-2017
Dorothy K Hall, & George A Riggs. 2011. NORMALIZED-DIFFERENCE SNOW INDEX (NDSI). In
Encyclopedia of Snow, Ice and Glaciers. https://doi.org/10.1007/978-90-481-2642-2
European Space Agency. (2015). SENTINEL-2 User Handbook. European Space Agency.
Gardelle, J., Berthier, E., Arnaud, Y., & Kääb, A. (2013). Region-wide glacier mass balances over the Pamir-
Karakoram-Himalaya during 1999–2011. The Cryosphere, 7(4), 1263–1286.
https://doi.org/10.5194/tc-7-1263-2013
Khan, A., Naz, B. S., & Bowling, L. C. 2015. Separating snow, clean and debris covered ice in the Upper Indus
Basin, Hindukush-Karakoram-Himalayas, using Landsat images between 1998 and 2002. Journal of
Hydrology. https://doi.org/10.1016/j.jhydrol.2014.11.048
Mandanici, E., & Bitelli, G. 2016. Preliminary Comparison of Sentinel-2 and Landsat 8 Imagery for a Combined
Use. Remote Sensing, 8(12), 1014. https://doi.org/10.3390/rs8121014
Minora, U., Bocchiola, D., D’Agata, C., Maragno, D., Mayer, C., Lambrecht, A., … Diolaiuti, G. 2013. 2001 -;2010
glacier changes in the Central Karakoram National Park: a contribution to evaluate the magnitude and rate of
the "Karakoram anomaly" The Cryosphere Discussions, 7(3), 2891–2941.
https://doi.org/10.5194/tcd-7-2891-2013
Najam ul Hassan, S., Md Reba, M. N., Hussain, D., & Ali, A. 2018. Elevation dependent thickness and ice-volume
estimation using satellite derived DEM for mountainous glaciers of Karakorum range. IOP Conference
Series: Earth and Environmental Science, 169(1), 012115. https://doi.org/10.1088/1755-1315/169/1/012115
Paul, F., Bolch, T., Kääb, A., Nagler, T., Nuth, C., Scharrer, K., … Van Niel, T. 2015. The glaciers climate change
initiative: Methods for creating glacier area, elevation change and velocity products. Remote Sensing of
Environment. https://doi.org/10.1016/j.rse.2013.07.043
Paul, F., Winsvold, S. H., Kääb, A., Nagler, T., & Schwaizer, G. 2016. Glacier remote sensing using Sentinel-2.
part II: Mapping glacier extents and surface facies, and comparison to Landsat 8. Remote Sensing, 8(7).
https://doi.org/10.3390/rs8070575
Quincey, D. J., & Luckman, A. 2014. Brief communication: On the magnitude and frequency of Khurdopin glacier
surge events. Cryosphere. https://doi.org/10.5194/tc-8-571-2014
Racoviteanu, A. 2011. Himalayan glaciers: combining remote sensing, field techniques and indigenous knowledge
to understand spatio-temporal patterns of glacier changes and their impact on water resources. University of
Colorado. Retrieved from http://scholar.colorado.edu/geog_gradetds
Racoviteanu, A. E., Williams, M. W., & Barry, R. G. 2008. Optical remote sensing of glacier characteristics: A
review with focus on the Himalaya. Sensors. https://doi.org/10.3390/s8053355
RGI Consortium. (2015). Randolph Glacier Inventory – A Dataset of Global Glacier Outlines: Version 5.0:
Technical Report, Global Land Ice Measurements from Space. Colorado, USA. Digital Media.
https://doi.org/https://doi.org/10.7265/N5-RGI-50
USGS. 2018a. PRODUCT GUIDE: LANDSAT 8 SURFACE REFLECTANCE CODE (LASRC) PRODUCT.
USGS. 2018b. USER GUIDE EARTH RESOURCES OBSERVATION AND SCIENCE (EROS) CENTER
SCIENCE PROCESSING ARCHITECTURE (ESPA) ON DEMAND INTERFACE. Department of the
Interior U.S. Geological Survey.
Zhang, H. K., Roy, D. P., Yan, L., Li, Z., Huang, H., Vermote, E., … Roger, J.-C. 2018. Characterization of
Sentinel-2A and Landsat-8 top of atmosphere, surface, and nadir BRDF adjusted reflectance and NDVI
differences. Remote Sensing of Environment, 215, 482–494. https://doi.org/10.1016/J.RSE.2018.04.031
Zhu, Z., Woodcock, C. E., Holden, C., & Yang, Z. 2015. Generating synthetic Landsat images based on all
available Landsat data: Predicting Landsat surface reflectance at any given time. Remote Sensing of
Environment. https://doi.org/10.1016/j.rse.2015.02.009
USING DTMs TO DELINEATE ACTIVE FAULTS OF THE PROXIMAL PART OF THE GANGA PLAIN,
UTTARAKHAND, INDIA
Pradeep K Goswami
Centre of Advanced Study, Department of Geology,
Kumaun University, Nainital – 263002, India
*E-mail : drpgoswami@yahoo.com,
KEY WORDS: Geomorphology, Tectonics, DTM, IRS imagery
ABSTRACT: The present study pertains to effective use of Digital Terrain Models (DTMs) in identifying and
mapping active faults in a large, proximal part of the Ganga (also called Gangetic) plain, where field geological
investigations are mostly refrained due to inaccessibility owing to dense, multistoried forest cover.
The SRTM (Shuttle Radar Topography Mission) 90 m DEMs were used to make a reconnaissance of the
study area, but for detailed investigations the DTMs were prepared from relief information given in toposheets.
Several hydrologically correct, grid-based DEMs were prepared in a Geographic Information System (GIS) for
different resolutions. 2-D profiles along a number of longitudinal and transverse sections were drawn. Several 3-D
perspective views were generated by draping the enhanced IRS imagery over the DEMs, for different exaggeration
factors of the z-value, Sun azimuth and Sun angles to emphasize subtle topographic variations. These DTMs were
then visually analyzed in conjunction with the satellite imagery to delineate the morpho-tectonic features.
Throughout the analysis of DTMs, special emphasis was placed on drainage characteristics. Subsequently, the maps
were verified during extensive fieldwork, and required corrections were made by incorporating the field data.
The investigations reveal that the area is traversed by a number of criss-crossing lineaments. Some of these
lineaments are active faults. Most of the active faults are generally concealed below the alluvium, but they could be
identified and demarcated on the basis of their geomorphic characteristics as discernible on the satellite image or
various DTMs. In the north is the active Himalayan Frontal Thrust (HFT), which defines the northern structural
limit of the Ganga basin against the Himalayan Mountains. Parallel to HFT in the south is the blind Najibabad Fault
(NF); however, it is identifiable only in the western part of the area. The HFT is offset by a number of dip-slip,
oblique-slip and strike-slip faults, some of which extend northward into Himalaya and are related to the basement
structures of the basin. Ongoing activities along these faults/thrusts have pronounced control on the river dynamics
and landscape of the Ganga foreland basin and adjoining Himalayan mountain-front.
1. INTRODUCTION
The Ganga Plain represents the central part of the alluvium filled Indo-Gangetic foreland basin system (Fig.1a). Its
western and eastern limits are defined by the NE-SW trending Delhi-Hardwar and Monghyr-Saharsa basement
ridges respectively (Sastri et al. 1971; Karunakaran and Ranga Rao, 1979). There are a number of ridges, spurs,
depressions and faults in the basement of the Ganga Basin (Sastri et al., 1971), some of which are extensions of the
faults of the Peninsular India and extend further northwards into the Himalaya (Sastri et al., 1971; Raiverman et al.,
1983). The northern limit of the Ganga Basin is sharply defined by the Himalayan Frontal thrust (HFT), whereas
the southern limit is diffuse and marked by the highlands of the Peninsular Craton.
Geomorphically, the Ganga Plain is subdivided into three broad zones extending apparently parallel to
the axis of the basin. These are, the Piedmont Zone (PZ), Central Alluvial Plain (CAP) and Marginal Alluvial Plain
(MAP) (Fig.1b) (Singh, 1996). The PZ is formed to the south of the Himalayan foothills as a result of the
coalescence of several alluvial fans and talus deposits. This is generally a 10-50 km wide, southwards sloping zone,
gravelly in the proximal part and silty-fine sandy in the distal part. The CAP is a gently south-eastwards sloping,
generally sandy zone extending between PZ and axial river of the basin (Yamuna River in the western part and
Ganga River in the eastern part). The sediments of the CAP are derived from the Himalaya. The MAP is a north to
north-eastwards sloping zone extending between the Peninsular Craton and axial river of the basin. It is composed
of sediments derived from the Peninsular Craton.
Despite that a substantial amount of information has been generated over the past few years on various
geological aspects of different parts of the Ganga Plain, the information on the active faults of the PZ has not yet
been compiled. The field identification and demarcation of morpho-tectonic features in the PZ is impeded because
of the thick, multi-storeyed forest cover, extensive anthropogenic modifications in the landscape and easily
reworkable nature of the gravels. The dense forest cover also makes a large part of the area inaccessible. Therefore,
this study is based on analysis and interpretation of satellite imagery and Digital Terrain Models (DTMs) together
with detailed fieldwork.
Following our previous studies (i.e., Goswami et al., 2009; Goswami and Yhokha, 2010; Goswami,
2012, 2017; Goswami and Mishra, 2014a,b), the main purpose of this paper is to demonstrate the successful
utilization of RS and GIS technologies in mapping of active faults in densely forested plain areas. For this,
compiled here is the data of a part of the densely forested PZ of the western Ganga Plain. We hope that the present
study will provide a model for carrying out similar investigations in densely forested plain areas of other parts of
the world.
2. MATERIALS AND METHODS
The active faults in the present study area have been demarcated on the basis of their geomorphic expressions, as
enunciated by many workers (e.g. Keller and Pinter, 1996; Burbank and Anderson, 2001; Arrowsmith and Zielke,
2009). The flatness of the surface, thick forest cover and anthropogenic modifications of the landscape and the
friable and easily reworkable nature of the unconsolidated sediment fill make difficult the delineation of subtle
surface deformations through traditional field methods. Therefore, the investigations are based on analysis and
interpretation of satellite imagery, stereopairs of black & white aerial photographs and Digital Terrain Models
(DTMs) together with detailed fieldwork.
The active faults of the study area have been delineated through visual interpretation of the enhanced
images and DTMs on the basis of geomorphic features and terrain characteristics, such as linear stream courses,
linear valleys, linear pattern of deflected stream courses, linear train of active landslides, linear zones of slope
break, linear pattern of tonal variations in satellite images. Escarpments, terraces and topographic undulations are
easily identifiable on aerial photo stereograms under a zoom stereoscope. A number of lineaments become clear in
shaded relief maps at different sun illumination, whereas in slope and slope aspect maps the faults can be identified
with fair confidence because in many sectors they are represented as linear zones of sharp break of slope of
comparatively uniform slope aspect. In 2-D cross-sections the faults exhibit abrupt increase in the gradient. The
interpreted maps were subsequently verified in the field, and required corrections were made by incorporating the
field data.
DTMs, like shaded relief, slope, slope aspect, two-dimensional (2-D) cross-sections and Digital Elevation
Models (DEMs) formed the primary data for the present study. The SRTM (Shuttle Radar Topography Mission) 90
EMs (available from the United States Geological Survey (USGS) server at http://dds.cr.usgs.gov/srtm/ and
http://seamless.usgs.gov/) have been used to make a reconnaissance of the study area, but for detailed investigations
Figure 1 (a) Extent of the Indo-Gangetic Plain (IGP) between the Himalaya and the Indian craton (after Singh
1996). The extent of Ganga Plain is marked by two thick lines on the figure. (b) Simplified map of the Ganga Plain
showing Piedmont Zone (PZ), Central Alluvial Plain (CAP) and Marginal Alluvial Plain (MAP) (simplified after Singh
1996). Rectangle marks the location of the study area. HFT: Himalayan Frontal Thrust; R: River.
the DTMs were prepared in ArcGIS Geographic Information System (GIS) from relief information provided in the
Survey of India toposheets. Several hydrologically correct, grid-based DEMs were prepared for different
resolutions (25 m, 50 m 100 m and 150 m grid sizes) using the TOPOGRID command in ArcGIS software. Several
3-D perspective views were generated by draping the enhanced satellite imagery over the DEMs, for different
exaggeration factors (ranging from 1 to 25) of the z-value (elevation), Sun azimuth and Sun angles to identify subtle
topographic variations.
The DTMs were analysed in conjunction with LISS III digital imagery, having a spatial resolution of
23.5m, from the Indian Remote Sensing Satellites (IRS) using the ArcGIS software. The imagery were
georeferenced and enhanced using ERDAS IMAGINE software. The linear contrast stretched NIR (Near Infrared)
band grey-scale image, linear contrast stretched NIR/Red band-ratioed grey-scale image and a False Colour
Composite (FCC) image generated by coding linear contrast stretched SWIR (Shortwave Infrared), NIR and red
bands in red, green and blue colour planes respectively are found most suitable for the present investigations.
Subsequently, the maps were verified by fieldwork, and required corrections were made by incorporating the field
data.
3. RESULTS AND DISCUSSION
3.1 Active faults of the area
The active faults of the study area are shown in figure 2, and a brief description of the geomorphic features
expressing them, as discernible in sundry DTMs and verified in field, is given below:
3.1.1 Longitudinal faults
Trending parallel to the axis of the basin are the HFT in the north and Najibabad Fault in the south. The
HFT defines the northern structural limit of the PZ, but larger segments of it are concealed below alluvia. In the
DTMs it is expressed by relief anomalies, upwarps and drainage deflections and anomalies, and terraces (Fig. 3–5).
At places the HFT is expressed by scarps, which are 7–70 m high in different stretches; for example, a 7–18 m high
scarp between Malin River and Kotwali Rao stream, 50–70 m high scarp to the west of Ramganga River, and 18 m
Figure 2. Map showing active faults of the Piedmonte Zone of the study area. HFT, Himalayan Frontal Thrust; MBT, Main
Boundary Thrust; NF, Najibabad Fault; 1–12 faults transverse to Himalayan strike; a–p, rivers/streams: 1, Sukh Rao
Fault; 2, Khoh River Fault; 3, Kalagarh Fault; 4, Jhirna Fault; 5, Ramnagar Fault; 6, Dabka Fault; 7, Nihal Fault; 8,
Unchapul Fault; 9, Haldwani Fault; 10, Kathgodam Fault; 11, Chorgallia Fault; 12, Kalaunia Fault; 13, Tanakpur Fault;
a, Kotwali Sot stream; b, Malin River; c, Sukh Rao stream; d, Khoh River; e, Ramganga River; f, Dhara River; g, Banaili
River; h, Phika River; i, Dhela River; j, Kosi River, k, Nihal River; l, Bhakhra River; m, Gola River; n, Sukhi River; o,
Nandhaur River; p, Kalaunia River. xx’, yy’, zz’ and x’’y’’ are lines across which 2-D profiles shown in figure 3 are drawn
Figure 2. Map showing active faults of the Piedmonte Zone of the study area. HFT, Himalayan Frontal Thrust; MBT, Main
Boundary Thrust; NF, Najibabad Fault; 1–12 faults transverse to Himalayan strike; a–p, rivers/streams: 1, Sukh Rao
Fault; 2, Khoh River Fault; 3, Kalagarh Fault; 4, Jhirna Fault; 5, Ramnagar Fault; 6, Dabka Fault; 7, Nihal Fault; 8,
Unchapul Fault; 9, Haldwani Fault; 10, Kathgodam Fault; 11, Chorgallia Fault; 12, Kalaunia Fault; 13, Tanakpur Fault;
a, Kotwali Sot stream; b, Malin River; c, Sukh Rao stream; d, Khoh River; e, Ramganga River; f, Dhara River; g, Banaili
River; h, Phika River; i, Dhela River; j, Kosi River, k, Nihal River; l, Bhakhra River; m, Gola River; n, Sukhi River; o,
Nandhaur River; p, Kalaunia River. xx’, yy’, zz’ and x’’y’’ are lines across which 2-D profiles shown in figure 3 are drawn
high scarp to the west of the Sarda River. At many other places gravel ridges have developed along the HFT zone;
for example a 70–80 m high ridge to the northwest of Dhela River, 5 m high ridge to the west of Nandhaur River,
and 50 m high ridge to the west of Sarda River (Fig 4a). These ridges may be fault-bend/fault-propagation folds
related to the blind segments of the HFT. In the western part of the study area, the drainages show marked
deflections along the trace of the HFT; the Kotwali Sot flows towards SSW in the Siwalik terrain but on reaching
the PZ it takes a sharp, knee-bend turn to flow towards W along the HFT and again turns sharply towards SW along
a lineament (Fig 5b). Ground tilting associated with HFT has produced paired or unpaired terraces along the
proximal parts of most of the rivers, and palaeochannels along the right bank of the Ramganga River. Most of the
mountain-fed streams enter into the PZ through deeply cut V-shaped valleys, suggesting accelerated undercutting in
response to faster ground uplift along the HFT.
The Najibabad Fault traverses through the south of the western part of study area. It’s a blind fault with
no surface exposure, but in DTMs it’s delineated on the basis of drainage, surface gradient and relief anomalies. In
2D profiles it is expressed by surface upwarping and sharp break in surface gradient, suggesting subsidence of the
southern fault block (Fig. 3a,b). Terrain upwarping along the NF has also caused accelerated erosion and thus
development of badland zone in the west of Ramganga River. The Ramganga River’s consistently rightward and
leftward migratory trends on the northern and southern fault blocks respectively indicate left-lateral strike-slip
component on the NF. The similar migratory trends of Kosi River further corroborate to a left-lateral strike-slip
component on the NF. The NF as such is an oblique-slip fault, and seems to be what Yeats and Thakur (2008) call
as the Piedmont Fault.
3.1.2 Transverse faults
There are many faults that trend transverse to the axis of the basin; most of these faults offset the HFT
(Fig. 2). These faults have been identified and the sense of movement on them ascertained on the basis of drainage
and relief anomalies discernible in the DTMs. In the western part of the study area, the N-S trending, branching
Sukh Rao Fault and Koh River Fault offset the HFT left-laterally and right laterally, respectively. Admittedly, the
area between these two faults produces a marked indent into the adjoining mountain-front. Eastward, the NE-SW
trending Kalagarh Fault and Jhirna Fault left-laterally offset the HFT. Analyses of DTMs reveal that the Jhirna
Fault has oblique-slip movement with ~20 m up-throw of the left-laterally displaced western fault block.
In the central part of the study area, the NNE–SSW trending Ramnagar Fault and Dabka Fault offset the
HFT right-laterally and left-laterally, respectively. While the former of these controls the anomalously straight
course of the Kosi River, the latter controls the course of the Dabka River, compelling its flow to take a knee-bend
turn from E–W to N–S direction, and thus preventing it from joining the Kosi River that flows just 500 m ahead.
Figure 3. 2-D profiles drawn across the lines shown in figure 2. The faults/thrusts are drawn
irrespective of dips, just to highlight their surface expressions.
Further eastward, NNE-SSW trending Nihal Fault controls the trend of Nihal River in the Siwalik as well as the
PZ. In DEMs, the NF is seen to right-laterally offset a linear ridge associated with HFT by ~800 m, indicating
oblique slip along this fault with western block upthrown to north. East of Nihal Fault the NE-SW trending
Unchapul Fault traverses the proximal PZ, which is expressed in DEM by a 6 km long and up to 15 high scarp
with western fault block uplifted up to 15 m. Parallel to the Unchapul Fault is the blind Haldwani Fault. Trending
NNE-SSW, it is delineated mainly on the basis of drainage deflections in PZ and adjoining Siwalik Hills. The
presence of a NNE-SSW trending palaeochannel downstream of Haldwani, in line with the active channel
upstream, suggests that movements along this fault have caused uplift of the western fault block so that the Gola
River has shifted eastwards following an avulsion just downstream of Haldwani. A comparison of the toposheet
(surveyed in 1964-65) with the satellite data of 1997 and 2004 reveals that, over the period of forty years, the width
of the river channel in the northeast of Haldwani has increased from ~522 m to ~632 m as a result of lateral-cutting,
and has shifted up to 360 m eastwards. This may also be due to uplift of the right bank of the river along the
Haldwani Fault. The deflection in the course of Gola River near the Haldwani indicates that this fault also has a
right-lateral strike-slip component of movement. As such, it is an oblique-slip fault with western block upthrown to
northeast (Fig. 3c). East of the Haldwani Fault, the NW–SE trending Kathgodam Fault controls the straight
course of the Sukhi River. In its upper reaches the Sukhi River descends southwestwardly from the adjoining
Siwalik, but in the PZ it sharply turns towards SSE along the Kathgodam Fault, rather than joining the Gola River
flowing just ~1.3 km away (Fig.2). All streams descending from the adjoining mountain between the northeast of
Haldwani and west of Chorgallia join this fault controlled Sukhi River almost perpendicularly, giving rise to a
gridiron drainage pattern (Fig. 4b).
In the eastern part of the study area, the NNE-SSW trending Chorgallia Fault dextrally offsets the HFT
as borne out in a DEM by the across-trend displacement of a gravel ridge related to a blind segment of the HFT.
These geomorphic features indicate that the Chorgallia Fault is an oblique-slip fault with western block upthrown
to the north. Eastward, the NNE–SSW trending Kalaunia Fault controls the course of Kalaunia River and its
tributaires, givin rise to gridiron-type drainage pattern. The southeastwardly flowing Kalaunia River takes a sharp
turn towards SSW along this fault, rather than joining the Sarda River flowing just ~2.5 km away. Along the eastern
extremity of the study area, the NE–SW trending Tanakpur Fault is expressed in a DTM by steeper south-easterly
gradient of the surface, indicating that the eastern fault block is downthrown (Fig. 3d, 5a). The displacement of
gravel ridges in the PZ indicates that the transverse Tanakpur and Kalaunia faults have dextrally offset the
longitudinal HFT.
Figure 4. (a) SRTM DEM showing gravel ridges marking surface expression of the HFT between Kalaunia and Sarda
rivers (highlighted by red ellipse. (b) IRS LISS III FCC (bands 3,4,5 in BGR colour planes, respectively), showing
gridiron drainage pattern marking the surface expression of the Kathgodam Fault (KF)
All the faults and thrusts identified on the DTMs have also been validated by fieldwork at a number of
places.
4. CONCLUSION
The present study demonstrates successful use of DTMs in mapping active tectonic features in flat,
densely forested and cultivated terrains. The best method to delineate active tectonic features in such terrains is to
first identify the lineaments on imagery and DTMs, following the geomorphic criterion, and then verify them in the
field. The 2-D sections and perspective views at grater vertical exaggerations are in particular very useful in
delineating subtle ground undulations associated with the active faults. Many lineaments become clear in shaded
relief maps at different sun illumination. Similarly, many faults are identifiable in slope maps (angle and aspect) due
to their characteristic linear zones of steeper surface gradient with a comparatively uniform aspect, or in across-
strike profiles due to abrupt increase in the gradient.
REFERENCES
Arrowsmith, J.R. and Zielke, O., 2009. Tectonic geomorphology of the San Andreas fault zone from high
resolution topography: An example from the Cholame segment. Geomorphology, 113, pp. 70–81.
Burbank, D. W. and Anderson, R. S., 2001. Tectonic Geomorphology. Blackwell Science, Massachusetts.
Goswami P.K., 2012. Geomorphic evidences of active faulting in the northwestern Ganga Plain, India: Implications
for the impact of basement structures. Geosciences Journal, 16, pp. 289-299.
Goswami, P.K., 2017. Controls of basin margin tectonics on the morphology of alluvial fans in the western Ganga
foreland basin’s piedmont zone, India. Geological Journal, DOI: 10.1002/gj.3010.
Goswami, P.K. and Yokha, A., 2010. Geomorphic evolution of the Piedmont Zone of the Ganga Plain, India: a
study based on remote sensing, GIS and field investigation. International Journal of Remote Sensing, 31, pp.
5349-5364
Goswami, P.K. and Mishra, J.K., 2014a. Tectonic and climatic controls on the Quaternary landscape evolution of
the Piedmont Zone of the Ganga Plain, India. Zeitschrift für Geomorphologie 58, pp. 367-384.
Goswami, P.K. and Mishra, J.K., 2014b. Morphotectonic evolution of the Piedmont Zone of the west Ganga Plain,
India. Zeitschrift für Geomorphologie 58, pp. 117-131.
Goswami, P.K., Pant, C.C. and Pandey, S., 2009. Tectonic controls on the geomorphic evolution of alluvial fans in
the Piedmont Zone of the Ganga Plain, Uttarakhand, India. Journal Earth System Science, 118, pp. 245-259.
Karunakaran, C. and Ranga Rao, A., 1979. Status of exploration for hydrocarbons in the Himalayan region –
contribution to stratigraphy and structure. Geological Survey of India Publication, 41, pp. 1-66.
Keller, E.A. and Pinter, N., 1996. Active tectonics: Earthquakes, uplift and landscape. Prentice Hall, New Jersey.
Nakata, T., 1972. Geomorphic History and Crustal Movements of the Foothills of the Himalaya. Tohoku University
Science Reports, 7th Series, Japan 22, pp. 39-177.
Figure 5. (a) 3-D perspective view obtained by draping IRS LIII FCC (band 2,3,4 in BGR colour planes, respectively) depicts
grounding tilting caused by the Tanakpur Fault (TF). (b) Drainage deflections marking surface trace of the Himalayan
Frontal Thrust (HFT) in the western part of the study area.
Raiverman, V, Kunte, S.V. and Mukherjea, A., 1983. Basin geometry, Cenozoic sedimentation and hydrocarbon in
north western Himalaya and Indo-Gangetic plains. Petroleum Asia Journal, 6, pp. 67-92.
Sastri, V.V., Bhandari, L.L., Raju, A.T.R. and Dutta, A.K., 1971. Tectonic framework and subsurface stratigraphy of
the Ganga Basin. Journal of the Geological Society of India, 12,pp. 222-233.
Singh, I.B., 1996. Geological evolution of Ganga Plain- an overview. Journal of the Palaeontological Society of
India 41, pp. 99-137.
Yeats, R.S., and Thakur, V.C., 2008. Active faulting south of the Himalayan Front: establishing a new plate
boundary. Tectonophysics, 453, pp. 63–73.
1
DETECTION OF HYDROTHERMAL ALTERATION ZONES AND LINEAMENTS
ASSOCITED WITH OROGENIC GOLD MINERALIZATION USING ASTER REMOTE
SENSING DATA IN SANANDAJ-SIRJAN ZONE, EAST IRAN
Abdollah Sheikhrahimi1, Amin Beiranvand Pour2, Mazlan Hashim*3 and Danboyi Joseph Amusuk3
1 Department of Geography and Urban Planning, Tabriz University, Tabriz, Iran
2 Korea Polar Research Institute (KOPRI) Songdomirae-ro,Yeonsu-gu, Incheon 21990,
Republic of Korea
3Geoscience and Digital Earth Centre (INSTeG), Universiti Teknologi Malaysia (UTM), 81310 UTM
Skudai, Johor Bahru, Malaysia
Email: abdola.sheikhrahimi@gmail.com; beiranvand.amin80@gmail.com; mazlanhashim@utm.my
KEY WORDS: Saqqez; ASTER; PCA; SAM; SID; Alteration zones; Lineaments; Gold exploration
ABSTRACT: The Sanandaj-Sirjan Zone (SSZ) is considered as an important region for exploration of orogenic gold
mineralization in the eastern sector of Iran. Mountainous topography and relatively lack of accessible route are
challenging for researchers and costly for mining companies for gold exploration in the SSZ. Gold mineralization
mainly occurs as irregular to lenticular sulfide veins along shear zones in extremely altered and deformed mafic to
intermediate metavolcanic and metasedimentary rocks. In this investigation, the Advanced Spaceborne Thermal
Emission and Reflection Radiometer (ASTER) satellite data were used for mapping indicator hydrothermal alteration
minerals and geological structural features associated with orogenic gold mineralization in the Saqqez plot of the SSZ.
Image transformation techniques such as specialized band ratioing and Principal Component Analysis were used to
delineate lithological units and alteration minerals. Supervised classification, namely Spectral Angle Mapper (SAM)
and Spectral Information Divergence (SID) supervised classification methods were used to detect subtle differences
between indicator alteration minerals associated with gold mineralization in the study area. Directional filtering was
implemented to trace structural features. Results demonstrate that the integration of image transformation techniques
and supervised classification derived from ASTER remote sensing analysis with fieldwork and previous stream
geochemical study has a great ability for targeting new prospects of gold mineralization in the Saqqez plot of the SSZ.
1. INTRODUCTION
Hydrothermal alteration mineral detection and lithological and structural geology mapping is one of the most
prominent applications of remote sensing satellite data for regional ore exploration programs during last decade (Pour
et al.,2017a,b). The Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) is a high spatial,
spectral and radiometric resolution multispectral remote sensing sensor. The ASTER data consist of three separate
subsystems with a total of 14 spectral bands: (a) the visible–near infrared (VNIR) subsystem contains three bands
(0.52–0.86 μm) with 15 m spatial resolution; (b) the shortwave infrared (SWIR) subsystem has six bands (1.60–2.43
μm) with 30 m spatial resolution; and (c) the thermal infrared (TIR) obtains five bands (8.12– 11.65 μm) with 90 m
spatial resolution. ASTER swath width is 60 km that each individual scene is cut to a 60 × 60 km2 area (Abrams,
2000). Iron oxide/hydroxide minerals such as limonite, goethite, jarosite and hematite tend to have diagnostic
absorption features due to charge transfer and crystal-field processes in the VNIR region (0.4 to 1.1 μm) of the
electromagnetic spectrum (Hunt and Ashley 1979). Thus, these spectral characteristics can be used to map iron oxide/
hydroxide minerals at the Earth’s surface using the VNIR bands of ASTER remote sensing data (Noda and
Yamaguchi, 2017). Hydroxyl-bearing minerals including clay and sulfate groups as well as carbonate minerals
present diagnostic spectral absorption features due to vibrational processes of fundamental absorptions of Al–O–H,
Mg–O–H, Si–O–H, and CO3 groups in the SWIR region (Hunt and Ashley 1979). Therefore, spectral bands of
ASTER in the SWIR region (1.60 μm to 2.5 μm) have great ability to map hydrothermal alteration mineral zones
associated with ore mineralization and alteration of the rocks surface (Pour et al., 2013; Safari et al., 2017). The
ASTER TIR bands (10–14) are useful for detecting silicate and carbonate rocks (Ninomiya et al., 2005).
The area of Saqqez plot (46˚ to 46˚.30' east longitude and 36˚ to 36˚.30' north latitude), North West of Kurdistan is
2
located in the SSZ, east Iran (Fig. 1). It is a high potential zone for orogenic gold mineralization due to infiltration of
granitoid masses related to Precambrian and Mesozoic into Archaic metamorphosed stones and Paleozoic
carbonate–debris coverage (Fig. 2). A comprehensive remote sensing investigation has not been reported for gold
mineral exploration and prospecting in the SSZ especially for Saqqez plot, yet. Accordingly, the objectives of this
investigation are: (1) to map indicator hydrothermal alteration minerals and geological structural features such faults
and fracture (lineaments) associated with orogenic gold mineralization in the Saqqez plot of the SSZ using the
Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) satellite data.
Figure 1. The location of the Sanandaj-Sirjan Zone (SSZ) and the Saqqez plot in the Kurdistan province, east Iran.
2. MATERIALS AND METHODS
2.1 Remote sensing data
In this research, a cloud-free ASTER level 1T (AST_L1T_00309062002075134_20150424203448_10828) covering
Saqqez plot was obtained from the USGS EROS (https://earthexplorer.usgs.gov) in 2002/09/06 (Path-168 and
ROW-35 and. The ASTER Level 1 Precision Terrain Corrected Registered At-Sensor Radiance (AST_L1T) data
contains calibrated at sensor radiance, which corresponds with the ASTER Level 1B (AST_L1B), that has been
geometrically corrected, and rotated to a north up UTM projection
(https://lpdaac.usgs.gov/dataset_discovery/aster/aster_products_table/ast_l1t). The image has been
pre-georeferenced to UTM zone 38 North projection with the WGS-84 datum.
2.2 Data analysis
The main purpose of the methodology is to apply image processing techniques that are capable detecting subtle
hydrothermal alteration minerals and mapping structure elements associated with orogenic gold mineralization in the
study area using VNIR+SWIR spectral bands of ASTER. Image transforms are commonly used to reduce the
dimensionality of the input dataset and processing time, focus processing on the information of interest within the
input file and removing noise. Each output band of a transformed image is a linear combination of every input image,
thus helping to identify those spectral bands that are most important for finding targets of interest, or which bands
contribute the most noise (Research Systems, Inc. 2008). To reduce the effects of topography and enhancing the
spectral differences between bands, band ratioing technique was selected in this analysis. It is a technique where the
digital number value of one band is divided by the digital number value of another band. Band ratios are very useful
3
for highlighting hydrothermal alteration minerals (Abrams et al., 1983). Dividing one spectral band by another
produces an image that provides relative band intensities, this is able to minimize the illumination differences due to
topography. Therefore, this technique is particularly applicable for highly exposed areas and rugged terrains in arid
and semi-arid environments. Moreover, Red-Green-Blue (RGB) color composites technique could be easily applied
on band ratios to produce image map of lithological units of the study area.
Multispectral data bands are often highly correlated; the principal components (PC) transformation could be used to
produce uncorrelated output bands, segregate noise components and reduce the dimensionality of data sets. This is
done by finding a new set of orthogonal axes that have their origin at the data mean and that is rotated so the data
variance is maximized. PC bands are linear combinations of the original spectral bands and are uncorrelated. The
principal component analysis is a well-known method for alteration mapping in metallogenic provinces for mineral
exploration objectives (Safari et al., 2017). In this study, the forward PC rotation was applied to VNIR+SWIR and
TIR bands of ASTER covering the study area. It uses a linear transform to maximize the variance of the data. Table 1
show the eigenvector values for VNIR+SWIR bands, respectively, which obtained using a covariance matrix.
Convolution filters produce output images in which the brightness value at a given pixel is a function of some
weighted average of the brightness of the surrounding pixels (Research Systems, Inc. 2008). The directional filter is a
first derivative edge enhancement filter (convolution filter) that selectively enhances image features having specific
direction components (gradients). The sum of the directional filter kernel elements is zero. The result is that areas with
uniform pixel values are zeroed in the output image, while those that are variable are presented as bright edges. 5*5
kernel matrix was selected in this study to enhance semi-smooth and smooth/rough features. Four principal
Directional filters N–S, E–W, NE–SW, and NW–SE with a 5 × 5 kernel size were applied to band 6 of ASTER.
Directional filter angles were adjusted as N–S: 0°, E–W: 90°, NE–SW: 45° and NW–SE: 135°. North (up) is 0° and
the other angles are measured in the counterclockwise direction. Image Add Back value was entered 70%. The Image
Add Back value is the percentage of the original image that is included in the final output image. Adding back part of
the original image to the convolution filter results helps preserve the spatial context and is typically done to sharpen
an image.
3. RESULTS AND DISCUSSION
The combination of band ratios is a robust method for information extraction of specific hydrothermal alteration
zones and reducing the effects of topography. Therefore, a specialized band ratio image map derived from image
spectra were developed by assigning 6/8, 4/6 and 4/5 in RGB for mapping and discriminating the argillic, phyllic and
propylitic zones in the study area (Fig. 2). Band ratio of 6/8 was employed for identifying Fe, Mg-OH rich area
(propylitic zone). Band ratio of 4/6 was used for identifying muscovite/illite (phyllic zone) by virtue of their
reflectance in band 4 and absorption features in band 6 of ASTER data. Band ratio of 4/5 was added to the RGB
composites for identifying kaolinite and alunite (argillic zone) and reducing the effect of unaltered/silicate rocks of
the background. Figure 7 shows the resultant image map. Propylitic alteration zone appears in magenta to pink color
due to strong absorption of Mg-OH minerals in band 8 of ASTER and argillic and phyllic alteration zones appear in
green and whitish yellow because of strong absorption of Al-OH minerals in bands 5 and 6 of ASTER. Blue tone is
attributed to unaltered/silicate lithological units. Argillic and phyllic alteration are mostly recognizable in the
southwestern and northeastern part of the study area, while propylitic alteration distributed in the background of the
scene (Fig. 2).
For detail mapping of alteration minerals and lithological units, PCA transformation was applied to VNIR+SWIR and
TIR bands of ASTER. The statistic results of PCA transformation for VNIR+SWIR bands of ASTER (Table 1) show
that the PC1 is composed of a negative weighting of all total bands, which corresponds with overall scene brightness
(albedo) and the strong correlation between image bands (Loughlin, 1991). According to Loughlin, 1991, the PC that
contains loadings of similar signs on both input band, explains the variance due to similarities in the spectral
responses of the interfering component and the component of interest. The other PC, whose loadings are of different
signs on either of the input band images, highlights the contributions unique to each of the components. The sign of
the loadings dictates whether the component of interest is represented as bright or dark pixels in the PC image.
Considering the eigenvector loadings in PC2 for enhancing alteration minerals in VNIR+SWIR bands of ASTER in
the scene, this PC only shows the difference between the visible and infrared bands (bands 1, 2, and 3 as negative
eigenvector loadings) and shortwave infrared bands (band 4, 5, 6, 7, 8 and 9 as positive eigenvector loadings) (Table
1). Therefore, the remaining seven PCs could be considered to have information related to the spectral response of
iron oxides and hydroxyl-bearing minerals components.
4
Figure 2. RGB color composite image map of the Saqqez plot derived from band ratios 6/8, 4/6 and 4/5.
Table 1. Eigenvector matrix of VNIR+SWIR bands provides to the principal component analysis for the study area.
Eigenvector Band 1 Band 2 Band 3 Band 4 Band 5 Band 6 Band 7 Band 8 Band 9
PCA1 -0.320855 -0.387854 -0.301641 -0.339137 -0.319744 -0.338363 -0.326673 -0.346945 -0.311140
PCA2 -0.412968 -0.438544 -0.526861 0.076469 0.265347 0.226342 0.274314 0.293055 0.266339
PCA3 -0.412042 -0.451051 0.710046 0.327660 0.036797 0.052432 0.001750 -0.063192 -0.084551
PCA4 -0.172872 0.286868 -0.326148 0.683599 0.130490 0.135934 -0.116134 -0.351494 -0.376300
PCA5 -0.612885 0.506871 0.052764 0.001191 -0.269497 -0.123264 -0.186849 0.059803 0.488218
PCA6 0.344701 -0.307718 -0.076463 0.258405 -0.112259 0.149391 -0.435074 -0.341308 0.610877
PCA7 -0.175322 0.131155 0.100178 -0.479944 0.297927 0.682738 -0.150000 -0.364049 -0.041894
PCA8 0.040450 -0.049919 -0.047171 0.082836 -0.760040 0.553595 0.050323 0.257065 -0.184501
PCA9 -0.005055 0.014379 0.001155 -0.022089 -0.239416 -0.054230 0.745193 -0.589512 0.190213
A PC image contains strong eigenvector loading for diagnostic reflectance and absorption bands of mineral with
opposite signs enhance that mineral. If the loading is positive in reflectance band of a mineral the image tone will be
bright, while if negative the image tone will be dark for target mineral (Loughlin, 1991). Considering the eigenvector
loadings for bands 1, 2 and 3 in PC3 where these loadings are also in opposite sign, iron oxides could be mapped due
to strong positive contribution of band 3 (0.710046) and strong negative loadings of bands 1 (-0.412042) and 2
(-0.451051) as bright pixels in PC3 (Table 1). Iron oxide minerals have low reflectance in visible and higher
reflectance in near-infrared corresponding bands 1, 2 and band 3 of ASTER data. Eigenvector loadings for PC4
indicate that PC4 has high potential to enhance Fe-Mg-OH minerals (propylitic alteration zone: chlorite, biotite and
epidote). The strong positive contribution of band 4 (0.683599) and negative strong loadings of band 8 (-0.351494)
and band 9 (-0.376300) in the PC4 emphasize for enhancing the Fe-Mg-OH minerals as bright pixels in the PC4
image (Table 1). Fe, Mg(OH)-bearing minerals such as chlorite, epidote and biotite contain high reflectance in band 4
(1.60-1.70 μm) and distinctive absorption in bands 8 and 9 (2.29–2.43 μm) of ASTER data.
After analyzing the eigenvector loadings for PC5 and PC6, it seems that they do not contain desired information
5
related to Al-OH alteration minerals. Therefore, these PCs are uninformative for alteration mineral mapping.
Eigenvector loadings for PC7 indicate high potential of this PC for mapping Al-OH alteration minerals (argilic and
phylic alteration zones: kaolinite, alunite, muscovite and illite) due to the strong negative loading of band 4
(-0.479944) and strong positive weighting of band 6 (0.682738) (Table 1). Hence, they appear as dark pixels in the
PC7. Al(OH)-bearing minerals such as kaolinite, alunite, muscovite and illite show major absorption in band 6
(2.185-2.228 μm) of ASTER. Eigenvector loadings of PC8 and PC9 do not show appropriate contributions of
reflectance and absorption bands for enhancing alteration minerals.
Accordingly, RGB color composite was assigned to PC3, PC4 and PC7 image map to represent the surface
distribution of iron oxide, Fe-Mg-OH and Al-OH mineral groups in the study area. It must be noted that before
applying the RGB color composite, dark pixels in PC7 were inverted to bright pixels by multiplication to -1. Figure 9
shows the resultant image map. Surface distribution of iron oxide minerals appears in magenta color, which mostly
observable in the northeastern, eastern and southwestern parts of the study area associated with carbonate rocks.
Fe-Mg-OH minerals depict in green color in the northwestern, northeastern and central southern parts of the study
area, where the outcrops of different sedimentary rocks, metamorphic rocks and andesitic volcanic rocks are
observable. Surface distribution of blue pixels (Al-OH minerals) is less in abundance. Yellow pixels might show the
surface distribution of Al-OH mineral groups that they mixed with other mineral groups in the northeastern and
southern part of the study area (Fig. 3).
Figure 3. RGB color composite image map of the Saqqez plot derived from PC3, PC4 and PC7 of VNIR+SWIR
bands.
After analyzing four principal directional filters to band 6 of ASTER, the most pronounced trends and lineaments
were mapped in the study area. Figure 15 shows the resultant lineament map. Two dominant trends, including
NE–SW and W–E sets of lineaments are identified in the study area. Several curvilinear structures indicate
open-upright fold systems with N–S axial plane in central and southern parts of the image map (Fig. 4). Intersections
of lineament and curvilinear elements can be seen in central, southern and southwestern parts of the scene. The main
trend of lineaments in the anomaly zones is NE–SW. However, intersections of structural elements are favorable sites
for intrusions and mineralization in the study area, which are also mapped in the vicinity of the anomaly zones. Shear
zone, mylonite, cataclasite and igneous intrusive coincident with hydrothermal alteration zones are especially
important for gold exploration. Therefore, in terms of structural analysis, the southern and northeastern parts of the
study are more suitable for potential gold mineralization.
6
Figure 4. Lineament map of the Saqqez region.
4. CONCLUSIONS
The new information extracted from specialized band ratioing, PCA, SAM, SID and directional filtering (DF) defines
several potential zones for gold exploration in the Saqqez region. The results overlap with published stream
geochemical surveys and show good coincidence. Statistical assessment and fieldwork data also verified the
consistency of the results. Accordingly, it is concluded that the structural traps (valley intersections) south of the
Saqqez town is a highly prospective area for future gold exploration program. This investigation emphasizes that a
remote sensing approach using ASTER data should be considered as a strong reconnaissance tool for targeting high
potential gold mineralization zones before costly field-data-required techniques in the SSZ.
Acknowledgements
This study was conducted as a part of KOPRI research grant PE17160. KOPRI grants PE17050 was also
acknowledged for supporting the research. We are thankful to Korea Polar Research Institute (KOPRI) for providing
all the facilities for this investigation. University Technology Malaysia (UTM) also appriciated.
REFERENCES
Abrams, M.J., Brown, D., Lepley, L. and Sadowski, R., 1983. Remote sensing of porphyry copper deposits in
Southern Arizona. Economic Geology, 78, 591-604.
Abrams, M., 2000. The Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER): data products
for the high spatial resolution imager on NASA‘s Terra platform. International Journal of Remote Sensing, 21,
847-859.
Hunt, G.R., and Ashley, P., 1979. Spectra of altered rocks in the visible and near infrared. Economic Geology, 74,
1613-1629.
Loughlin, W.P., 1991. Principal components analysis for alteration mapping. Photogrammetric Engineering and
Remote Sensing 57, 1163–1169.
Ninomiya, Y., Fu, B. and Cudahy, T.J. (2005). Detecting lithology with Advanced Spaceborne Thermal Emission and
Reflection Radiometer (ASTER) multispectral thermal infrared radiance-at-sensor‖data. Remote Sensing of
7
Environment, 99 (1-2), 127-139.
Noda, S., Yamaguchi, Y., 2017. Estimation of surface iron oxide abundance with suppression of grain size and
topography effects. Ore Geology Reviews 83, 312–320.
Pour, A.B., Park, Y., Park, T.S., Hong, J.K. Hashim, M., Woo, J., Ayoobi, I. 2018a. Evaluation of ICA and CEM
algorithms with Landsat-8/ASTER data for geological mapping in inaccessible regions. Geocarto International,
doi.org/10.1080/10106049.2018.1434684
Pour, A.B., Park, Y., Park, T.S., Hong, J.K. Hashim, M., Woo, J., Ayoobi, I. 2018b. Regional geology mapping using
satellite-based remote sensing approach in Northern Victoria Land, Antarctica. Polar Science, 16, 23-46.
Research Systems, Inc., 2008. ENVI Tutorials. Research Systems, Inc., Boulder, CO.
Safari, M., Maghsodi, A., Pour, A.B., 2017. Application of Landsat-8 and ASTER satellite remote sensing data for
porphyry copper exploration: a case study from Shahr-e-Babak, Kerman, south of Iran. Geocarto International
http://dx.doi.org/10.1080/10106049.2017.1334834.
1
SPACE-BORNE SATELLITE SENSORS FOR MINERAL EXPLORATION IN HIGH
ARCTIC REGIONS
Amin Beiranvand Pour1, Tae-Yoon S. Park1, Mazlan Hashim*2, Yongcheol Park1, Jong Kuk Hong1
1 Korea Polar Research Institute (KOPRI) Songdomirae-ro,Yeonsu-gu, Incheon 21990,
Republic of Korea
2Geoscience and Digital Earth Centre (INSTeG), Universiti Teknologi Malaysia (UTM), 81310 UTM
Skudai, Johor Bahru, Malaysia
Email: beiranvand.amin80@gmail.com; typark@kopri.re.kr; mazlanhashim@utm.my; ypark@kopri.re.kr;
jkhong@kopri.re.kr
KEY WORDS: Arctic regions; Landsat-8; ASTER; PALSAR; The Franklinian Basin; Zinc exploration; North
Greenland
ABSTRACT: The Franklinian Basin in North Greenland has a distinctive potential for exploration of world-class
zinc deposits. In this study, image processing algorithms are implemented on satellite remote sensing datasets define
hydrothermal alteration halos associated with Zn-Pb±Ag sulfide mineralization in the trough sequences and
shelf-platform carbonate of the Franklinian Basin. Directed Principal Component Analysis (DPCA) is applied to
selected Landsat-8 mineral indices to map carbonate and clay alteration. Major lineaments, intersections, curvilinear
structures and sedimentary formations are traced by the application of Feature-oriented Principal Components
Selection (FPCS) to cross-polarized backscatter PALSAR ratio images. PC image with strong textural variations was
selected as input band for directional filtering and consequently mapping geological structures. Mixture Tuned
Matched Filtering (MTMF) algorithm is applied to ASTER VNIR/SWIR bands for subpixel detection and
classification of hematite, goethite, jarosite, alunite, gypsum, chalcedony, kaolinite, muscovite, chlorite, epidote, and
carbonate. The resultant MF score images were subsequently used for virtual verification. We identified several high
potential zones with distinct alteration mineral assemblages and structural fabrics that could represent undiscovered
Zn-Pb sulfide deposits in the region.
1. INTRODUCTION
Geological investigations and mineral exploration in the Arctic have been naturally hampered by its remoteness and
climatic conditions. Greenland has a variety of mineral resources (Fig. 1) and mineral exploration tradition since 18th
century, but mineral exploration activities have been only focused on certain regions with variable intensity and
density of data collection, leaving most of the parts of Greenland largely underexplored compared to other areas with
similar geology elsewhere in the Arctic (Kolb et al., 2016). The Franklinian Basin as a highly prospective part of
Greenland for zinc exploration extends for more than 2,500 km E-W through the Canadian Arctic Islands and
northern Greenland (Fig. 1). Greenland's coastline is cut by numerous deep fjords; the topography and lack of roads
require helicopter support for accessibility of inland areas and plateaus for mineral exploration. Moreover, the time
frame for fieldwork varies largely with the geography, which is July to August in the northern parts (High Arctic
environment) of Greenland. Sea ice in the north breaks up during May to June and results in a wide pack ice girdle
along the east coast, which may hinder access to land (Kolb et al., 2016). Subsequently, the remote nature and
environmental challenges posed by the Arctic environment reduces the capacity to economically explore and locate
mineral resources by using traditional techniques. Satellite remote sensing data are capable of providing key
information for mineral exploration community to explore larger areas, reduce exploration costs and focus on key
hydrothermal alteration mineral assemblages, lithological units and structural features that are associated with
different types of ore mineralization (Safari et al., 2017; Pour et al., 2018 a,b).
Since no report on comprehensive remote sensing investigation is available for Zn-Pb exploration purposes in the
Franklinian Basin. However, key geological criteria for zinc exploration using satellite remote sensing data could be
considered as specific alteration mineral zones and lineament trends. Consequently, the main objectives of a
satellite-based remote sensing investigation for zinc exploration in the Franklinian Basin are then set: (i) to apply
robust image processing algorithms for detecting pixels/sub-pixels contain spectral features related to key alteration
minerals and assemblages (gossan, hydrated sulfate, clay and carbonates) that may represent potential undiscovered
2
Zn-Pb mineralization zones in the Franklinian Basin using spectral bands of Landsat-8 and ASTER remote sensing
satellite data; and (ii) to map and highlight the major lineaments (faults and fractures), intersections, curvilinear
structures and sedimentary formations in the prospective target regions using PALSAR remote sensing satellite data.
Figure 2. Simplified geological map of the Greenlandic part of the Franklinian Basin showing the distribution of
known Zn-Pb occurrences. Inset: location of the Franklinian Basin within the wider context of Greenland and Arctic
Canada. Abbreviations to zinc occurrences: BE-Børglum Elv; CF-Citronen Fj; C-Cass Fj; HB-Hand Bugt;
KB-Kayser Bjerg; KL-Kronprins Christian Land; KS-Kap Schuchert; KW-Kap Wohlgemuth; LE-Løgum Elv;
NF-Navarana Fj; P-Petermann; RH-Repulse Havn; TE-Tvillum Elv (modified from Kolb et al.,2016).
2. MATERIALS AND METHODS
2.1 Remote sensing data
In this investigation, Landsat-8, ASTER and PALSAR datasets were used to map
lithological-mineralogical-structural features hosting CD and MVT Zn-Pb mineralization in the trough sequences and
carbonate shelf-platform of the Franklinian Basin at both regional and district scales. Two Landsat-8 level 1T (terrain
corrected) images covering the trough sequences and carbonate shelf-platform of the eastern part of the Franklinian
Basin were obtained through the U.S. Geological Survey Earth Resources Observation and Science Center (EROS)
(http://earthexplorer.usgs.gov). The images (LC80402442016199LGN00; Path/Row 040/244) and
(LC80402452016199LGN00; Path/Row 040/245) covering Peary Land, Amundsen Land, Johannes V. Jensen land,
Nansen land and Freuchen Land were acquired on July 17, 2016. Scene cloud cover was 17.55 % and 9.15 % for the
Landsat-8 images, respectively. During acquisition time (19:44:49 to 19:45:44) of the images, sun elevation recorded
as 21.380 and 22.680 and sun azimuth were -91.670 and -103.265, respectively. For district scale
lithological-mineralogical-structural mapping in the eastern part of the Franklinian Basin, several available scenes of
ASTER surface reflectance VNIR-SWIR data (Level-2B07 or AST-07) and PALSAR Fine mode Level 1.5 dual
polarization (HH + HV polarization) were used in this investigation. ASTER (AST-07) scenes contain
atmospherically corrected data were obtained on-demand from USGS Earth Resources Observation and Science
(EROS) center (https://earthdata.nasa.gov/). They acquired under favourable conditions of minimal could- and
snow-cover. PALSAR Fine mode Level 1.5 scenes were obtained from the Earth and Remote Sensing Data Analysis
Center (ERSDAC) Japan (http://gds.palsar.ersdac.jspacesystems.or.jp/e/). The Level 1.5 product used in this study
has a high-resolution mode with 6.25 m pixel spacing and dual polarization (HH + HV), which is geo-reference and
geo-coded. Nominal incident angle is 7.9-60.0.
3
2.2 Data analysis
To accomplish the objectives, directed principal components analysis (DPCA) and directed independent component
analysis (DICA) (Hyvarinen, 2013) were selected and implemented for mapping the target minerals at the pixel level.
Correspondingly in this analysis, Mixture Tuned Matched Filtering (MTMF) algorithm (Boardman, 1998) was
selected for applying to VNIR/SWIR spectral bands of ASTER surface reflectance data for subpixel detection and
classification of the alteration minerals using the reference spectra of selected end-member minerals extracted from
the USGS spectral library version 7.0 (Kokaly et al., 2017). Therefore, several mineralogic band ratio indices derived
from spectral bands of Landsat-8 and ASTER were selected on the basis that one band ratio index contains
information related to the component of interest (e.g. target alteration minerals). The indices were considered as input
image datasets for implementing DPCA and DICA in this analysis. Normalised Difference Snow Index (NDSI) (band
ratio of 3-6/3+6), ferric iron oxide index (band ratio of 4/2), ferrous iron oxide index (band ratio of 6/5) clay minerals
index (band ratio of 6/7) and thermal radiance lithology index (TRLI) (band ratio of 10+11×11) were used for
Landsat-8. NDSI attends as a significant snow/ice indicator whereas band ratios of 4/2, 6/5, 6/7 and 10+11×11 are
utilized to enhance ferric and ferrous iron oxide/hydroxide, clay/carbonate minerals and lithological units using
Landsat-8 spectral bands, respectively.
For implementing MTMF to VNIR+SWIR ASTER data, Minimum Noise Fraction (MNF) statistics is required.
Forward MNF to spectra was applied to transform endmember spectra into MNF space for use in MTMF. New
covariance statistics were computed. Subspace background was enabled for removing anomalous pixels before
calculating background statistics. Background threshold was adjusted 0.800 in this analysis for the fraction of the
background in the anomalous image. The MTMF output is a set of rule images equivalent to both the MF score and
the infeasibility score for each pixel matched to each endmember spectrum. For extracting radar information to map
the major lineaments (faults and fractures), intersections, curvilinear structures and sedimentary formations in the
Franklinian Basin using PALSAR data a developed image processing technique using the combination of HH
(co-polarized) and HV (cross-polarized) polarizations is required. As different polarizations are sensitive to ground
surface features of different dimensions, they collectively bring out greater geological-geomorphological-structural
detail.
3. RESULTS AND DISCUSSION
A regional view of the eastern part of the Franklinian Basin was constructed using mosaic of the Landsat-8 ratio
images assigned to NDSI (b3-b6/b3+b6), ferric iron oxide index (b4/b2) and clay minerals index (b6/b7) as
Red-Green-Blue (RGB) color composites, respectively (Fig. 2). The image map provides a color-based classification
of pixels with intense H2O, Fe3+, Al-OH and CO3 absorption features. It highlights snow/ice in dark orange to gold
colors, surface distribution of ferric iron oxide/hydroxide minerals in light green to green tones, clay and carbonate
minerals in blue color. The areas with admixture of iron oxide/hydroxide, clay and carbonate minerals represent in
cyan color (Fig. 5). NDSI allows discriminating snow/ice from exposed lithologies due to the fact that snow reflects
visible radiation (in 0.544 -0.565 μm) more strongly than it reflects radiation in the middle-infrared region (in
1.628-1.652 μm). Rock exposure produces very low or negative NDSI values as rocks are generally less reflective in
the visible and near-infrared portion. For mineral constituents, combinations and overtones of H2O or OH
fundamentals and CO3 can produce absorption features in the 2.1 μm to 2.5 μm (e.g., Hunt 1977; Hunt and Ashley
1979), which coincide with band 7 (2.11-2.29 µm) of Landsat-8. Hydroxo-bridged Fe3+ results in absorption bands in
the 0.43 to 0.9 μm regions coinciding with bands 2 and 4 of Landsat-8. Therefore, the combination of NDSI, ferric
iron oxide and clay minerals indices as RGB color composites enhances the target geologic materials contain
distinctive spectral characteristics in the study area.
The trough clastic sediments of Amundsen Land group (Lower Ordovician to Lower Silurian) host CD Zn-Pb
mineralization of Citronen Fjord deposit. The CD deposits tend to occur in clusters within their host stratigraphy and
second order basins focusing hydrothermal fluids. Thus, the Amundsen Land group and time-equivalent horizon
associated with synsedimentary faults could be considered as highly prospective strata in the trough sequences.
Moreover, the CD mineralization is pyrite rich and yields iron oxide/hydroxide and clay alteration minerals due to the
oxidation and acid weathering. Tureso Formation (Upper Ordovician to Lower Silurian – Morris Bugt Group) of the
carbonate platform was documented as a favorable stratigraphic horizon for MVT Zn-Pb mineralization (Rosa et al.,
2014). It is characterized by pale and dark-weathering dolostones with 150-180 m thick that often distinctly
burrow-mottled. Accordingly, some spatial extents of the Landsat-8 images covering potential tracts for CD and
MVT Zn-Pb deposits in the Peary Land, Amundsen Land and Nansen Land were selected for detail mapping of target
alteration minerals and related lithologies. Several significant stream sediment anomalies have been reported in the
selected regions by Rosa et al. (2014).
In this analysis, spatial subsets covering the Citronen Fjord Zn-Pb deposit (northeastern part of the Peary Land and
southeastern part of Johannes V. Jensen Land), the southern part of the Amundsen Land and the central part of the
4
Nansen Land were considered as potential tracts for CD mineralization. Three spatial subsets of the southern part of
Peary Land, namely the Erlandsen Land (SW Peary Land), the Melville Land (SE Peary Land) and the Melville Land
(2) (SE Peary Land) contain MVT Zn-Pb mineralization (Rosa et al., 2014) were also selected. DPCA/ICA analysis
was implemented to the NDSI, ferric iron oxide, ferrous iron oxide, clay minerals and TRLI indices of the selected
spatial subset scenes of Landsat-8. Statistical analysis has been limited to the selected regions to evaluate the resultant
images (Fig. 3).
Figure 2. Mosaic of the Landsat-8 ratio images as RGB color composites assigned to NDSI, ferric iron oxide and clay
minerals indices, showing a regional view of the eastern part of the Franklinian Basin.
Structurally, CD Zn-Pb mineralization is the product of local development in a sub-basin controlled by syn-genetic
faults and metal-bearing fluids derived from underlying fractures in the sea floor (Leach et al., 2010). In the
Franklinian basin, sulfide mineralization in the trough is associated with faults and fault splays with specific trends for
instance NW-SE TLFZ mineralization trend in Citronen Fjord deposit and E-W fractures in the Navarana Fjord Zn
occurrence. MVT Zn-Pb mineralization produce from basinal brine (mineralizing fluids) could have ascended and
precipitated metals in extensional fault systems within the carbonate-platform (Leach et al., 2010). Important
lineaments for MVT Zn-Pb mineralization in the Franklinian basin are long lasting structures such as Central Peary
Land Fault Zone (CPLFZ) striking N70 and other parallel structures. Moreover, structures related to strike slip
movements and extensional domains such as synclines with negative flower structure and lineaments strike N110 can
be considered amongst important structural groups (Rosa et al., 2014). For mapping the significant structural trends
associated with CD and MVT Zn-Pb mineralization, FPCS technique was applied to available scenes of PALSAR
data for the study area. The Amundsen Land and the Nansen Land of the trough sequence and Erlandsen Land and the
Melville Land (2) of the carbonate shelf succession were selected as spatial subsets (almost similar size with the
Landsat-8 subsets) for FPCS image analysis. Statistical factors were calculated for the cross-polarized backscatter
ratio images, namely HH/HV, HH+HV, HH-HV, HH+(HH+HV)/HV and HH+(HH-HV)/HV (Fig. 4).
5
Figure 3. ICA image maps of Landsat-8 spatial subset scenes considered for CD Zn-Pb mineralization. (A) ICA2 image map
showing ferric iron oxide/hydroxide minerals (gossan) for the subset of Citronen Fjord Zn-Pb deposit; (B) ICA5 image map
showing clay minerals for the subset of Citronen Fjord Zn-Pb deposit; (C) ICA2 image map showing ferric iron oxide/hydroxide
minerals (gossan) for the subset of the Amundsen Land; (D) ICA5 image map showing clay minerals for the subset of the
Amundsen Land; (E) ICA5 image map showing ferric iron oxide/hydroxide minerals (gossan) for the subset of the Nansen Land;
(F) ICA3 image map showing clay minerals for the subset of the Nansen Land.
Figure 5 (A-C) shows resultant alteration mineral maps derived from the ASTER MF score rule images for the SE
Citronen Fjord deposit, SW Amundsen Land and SW Nansen Land zones that contain a high potential for CD Zn-Pb
mineralization. Large areas of the alteration zones in the selected regions exhibit a high sub-pixel abundance of
chlorite and/or epidote and hematite and/or goethite, while muscovite and alunite and/or kaolinite have low surface
distribution. Gypsum shows very low surface abundance among the detected minerals in the selected subsets, which
only detected in some small zones in Citronen Fjord deposit subset (Fig 5 (A-C)). It is discernable that the association
of iron oxide minerals (hematite/goethite) and clay mineral assemblages (chlorite/epidote, muscovite and
alunite/kaolinite) possibly point to numerous target zones for CD Zn-Pb mineralization in the selected regions. For
example, several alteration zones in the vicinity of Trolle Land Fault Zone (TLFZ) in the Citronen Fjord deposit
subset (Fig 5 (A)), a strip between the Odin Fjord and Heimedal Icecap, some alteration zones in the northwestern and
central parts of the Amundsen Land subset (Fig. 5 (B)) and the northeastern corner and southern part (adjacent to the
J.P.Koch Fjord) of the Nansen Land subset (Fig. 5 (C)) could be considered as target zones for CD Zn-Pb
mineralization.
6
Figure 4. (A) FCC image map of PCA2, PCA3 and PCA5 for the PALSAR subset of the Amundsen Land; (B) NW–SE (135°)
directional filter image map derived from PCA2 for the subset of the Amundsen Land; (C) FCC image map of PCA2, PCA3 and
PCA4 for the subset of the Nansen Land; (D) NW–SE (135°) directional filter image map derived from PCA2 for the subset of the
Amundsen Land. Lithostratigraphic units are abbreviated as Pm= Polkorridoren Group, VG=Volvedal Group, ME=Merqujoq
Formation; AG=Amundsen Land Group, Qa= Quaternary alluvium, and PS= Paradisfjeld Group.
4. CONCLUSIONS
The results of this investigation demonstrate that analysis of combined remote sensing mapping techniques has great
capability as an exploration tool for mapping potential occurrences of Zn-Pb deposits in the Franklinian Basin, North
Greenland. Numerous potential zones for Zn-Pb deposits were mapped using the combined remote sensing
techniques. Their identification was based on small-scale (~100 - 200 m) mineral zoning patterns between hydrous
silica, jarosite, kaolinite, and smectite, which graded into background rocks dominated by ferric-iron, chlorite and
sericite. This investigation indicate that combined remote sensing mapping techniques can aid in identifying unknown
Zn-Pb sulfide deposits in the High Arctic Franklinian Basin by improving how and where permissive and/or favorable
mineral occurrence zones are located.
Acknowledgements
This study was conducted as a part of KOPRI research grant PE17160. KOPRI grants PE17050 was also
acknowledged for supporting the research. We are thankful to Korea Polar Research Institute (KOPRI) for providing
all the facilities for this investigation. University Technology Malaysia (UTM) also appriciated.
7
Figure 5. Alteration mineral maps derived from the ASTER MF score rule images for selected ASTER spatial subsets
in the trough sequence. (A) SE Citronen Fjord deposit; (B) SW Amundsen Land; and (C) SW Nansen Land.
REFERENCES
Boardman, J. W. 1998. Leveraging the high dimensionality of AVIRIS data for improved sub-pixel target unmixing
and rejection of false positives: mixture tuned matched filtering, in: Summaries of the Seventh Annual JPL Airborne
Geoscience Workshop, Pasadena, CA, p. 55.
Hyvärinen, A. 2013. Independent component analysis: recent advances. Phil Trans R Soc A 371: 20110534.
http://dx.doi.org/10.1098/rsta.2011.0534.
Kokaly, R.F., Clark, R.N., Swayze, G.A., Livo, K.E., Hoefen, T.M., Pearson, N.C., Wise, R.A., Benzel, W.M.,
Lowers, H.A., Driscoll, R.L., and Klein, A.J. 2017. USGS Spectral Library Version 7: U.S. Geological Survey Data
Series 1035, 61 p., https://doi.org/10.3133/ds1035.
Kolb, J., Keiding, J.K., Steenfelt, A., Secher, K., Keulen, N., Rosa, D., Stengaard, B.M. 2016. Metallogeny of
Greenland. Ore Geology Reviews 78, 493-555.
Leach, D. L., Bradley, D. C., Houston, D., Pisarevsky, S. A., Taylor, R. D., Gardoll, S. J. 2010. Sediment-hosted
lead-zinc deposits in earth history. Economic Geology, 105, 593–625.
Pour, A.B., Park, Y., Park, T.S., Hong, J.K. Hashim, M., Woo, J., Ayoobi, I. 2018a. Evaluation of ICA and CEM
8
algorithms with Landsat-8/ASTER data for geological mapping in inaccessible regions. Geocarto International,
doi.org/10.1080/10106049.2018.1434684
Pour, A.B., Park, Y., Park, T.S., Hong, J.K. Hashim, M., Woo, J., Ayoobi, I. 2018b. Regional geology mapping using
satellite-based remote sensing approach in Northern Victoria Land, Antarctica. Polar Science, 16, 23-46.
Rosa, D., Rasmussen, J.A., Sørensen, E.V. and Kalvig, P., 2014. Reconnaissance for Mississippi Valley-type and
SEDEX Zn-Pb deposits in the Franklinian Basin, Eastern North Greenland – Results of the 2013 Season, 2014/6.
Geological Survey of Denmark and Greenland Report, Copenhagen, 41 pp.
Safari, M., Maghsodi, A., Pour, A.B., 2017. Application of Landsat-8 and ASTER satellite remote sensing data for
porphyry copper exploration: a case study from Shahr-e-Babak, Kerman, south of Iran. Geocarto International
http://dx.doi.org/10.1080/10106049.2017.1334834.
CHANGE MONITORING OF BHAGIRATHI & ALAKHNANDA BASIN
GLACIER USING SATELLITE IMAGE
DarshitSavani*(1), R. D. Shah(1), I. Bahuguna (2), B.P. Rathor(2)
1Department of Geology, M. G. Science Institute, Gujarat University, Ahmedabad, India.
2Space Applications Centre, ISRO, Ahmedabad, India
E-mail: darshit.savani1994@gmail.com
Abstract: The history of glacier length fluctuations serves as a reliable indicator of the past
climate. In this paper, a numerical flow line model has been used to study the relationship
between length variations of the Himalayan glacier and local climate since 1876. The front
positions of Alakhnanda & Bhagirathi area drained by a river huge mass of ice are in
agreement with those followed. After a successful test run that appears to the real thing of the
past retreat, the model was also used to describe a possible future event over time of the huge
mass of ice for the next 100 years under different climatic situations. This work puts into
numbers huge mass of ice different versions in the Alakhnanda & Bhagirathi basin area of
the northern Himalaya by integrating huge mass of ice existing in satellite data from IRS
LISS and Landsat series of different years, which are 2001 and 2016. Glacier variations were
mapped and analysed; discrepancies between images could be detected and removed from
the integrated data using remap tables in Arc/Info grid both graphically and numerically. Our
results show that glaciers in the region both retreated and advanced during the last 15 years;
difference between the year 2001 and 2016, average Alakhnanda & Bhagirathi basin glacier
area decreased from 1.71m2& 1.11m2. On average, during the period 2001 and 2016
respectively, suggesting that glacier retreat has an expedition.
Keywords: Himalayan glacier, Alakhnanda & Bhagirathi basin area, retreated and advanced,
Introduction
The Himalaya comprise one of the largest collections of glaciers outside the polar regions,
with a total glacier cover of 33 000km2 [1] and around 9600 glaciers exists in the Indian
Himalaya [2]. Himalayan glaciers are the important source of fresh water for the innumerable
rivers that flow across the Indo-Gangetic plains. The rivers flow trans-boundary and meet the
potable water, irrigation, hydropower, fishery, inland navigation and other needs of more
than 1.3 billion people living downstream. With about 9,575 small and large glaciers in the
Himalayas [3], they hold the largest reserves of water in the form of ice and snow outside the
Polar Regions [4]. The Himalayas are thus also referred to as the ‘water towers’ of Asia and a
‘third pole’ of the earth.
Monitoring of glaciers actuates scientific interest for two main reasons. First, Glaciers
change monitoring has been used for climatic change investigation. The surface area and
volume of individual glaciers are monitored to estimate future water availability.
Second, glaciers in Indian Himalayas, have been recognized as important water storage
systems for municipal, industrial and hydroelectric power generation purposes.
GLACIERS occur in the high-altitude regions of the moun-tains and in the polar regions of
the earth. They are vitalto mankind as they control the global hydrological cycle, maintain
the global sea levels and perennially supply freshwater to the rivers. In the wake of
climatic variations arising due to increasing concentration of greenhouse gases in the
atmosphere resulting in global warming and its implications on various resources, glaciers
are increas-ingly being monitored worldwide. The Himalayan moun-tain system to the
north of the Indian land mass with arcuate strike of NW–SE for about 2400 km holds one
of the largest concentration of glaciers outside the polar regions in its high-altitude
regions. Perennial snow and ice-melt from these frozen reservoirs is used in catch-
ments and alluvial plains of the three major Himalayan river systems, i.e. Indus, Ganga
and Brahmaputra for irri-gation, hydropower generation, production of bio-resources and
fulfilling the domestic water demand. Also, variations in the extent of these glaciers are
understood to be a sensitive indicator of climatic variations of the earth system and might
have implications on the availability of water resources in the river systems. Therefore,
mapping and monitoring of these natural, frozen freshwater re-sources is required for
the planning of water resources and understanding the impact of climatic variations.
However, ground-based studies on monitoring of the Himalayan glaciers require
enormous effort in terms of time and logistics due to lack of atmospheric oxygen in
high altitudes, trekking in rough terrain and cold climatic regimes. Despite these
difficulties, the efforts made by many expedition teams have led to the generation of vital
information on the fluctuations of Himalayan glaciers in terms of mass balance or
simply snout monitoring 1–9 . Remote sensing having the capability of providing synop-
tic view, multi-temporal coverage and multispectral char-acterization of earth surface
features has demonstrated its utility for glacier monitoring in different mountain regions
of the world, including the Himalaya 10–20 . The satellite data available in the public
domain such as Landsat TM 21 , topographic maps prepared in the past, ae-rial
photographs and recently released CORONA photo-graphs along with data from other
earth observation satellites such as IRS series, ASTER, etc. have been the main sources
for generating this information. However, it is seen that very few studies compare the
changes in gla-ciers from data of similar sources. The present study uses mainly data from
LISS III sensor of IRS satellites for an interval of about one decade between 2001 and
2016 for monitoring of 513 glaciers taken from dif-ferent parts of the Alakhnanda and
Bhagirathi basin area.
Study Area
Gangotri Glacier originates in the Chaukhamba massif (6853–7138 m a.s.l.) and flows
northwest towards Gaumukh. The equilibrium-line altitude (ELA) of Gangotri Glacier is
4875 m a.s.l. [11].The Gangotri glacier, one of the largest ice bodies in the Garhwal
Himalayas, is located in the Uttarkashi district of the state of Uttarakhand in India (See Fig
1). It is one of the most sacred shrines in India, with immense religious significance. Being
the main source of the river Ganga, it attracts thousands of pilgrims every year. The Gangotri
glacier is a vital source of freshwater storage and water supply, especially during the summer
season for a large human population living downstream. The discharge from the glacier
flows as the river Bhagirathi initially before meeting the Alaknanda River at Devprayag
to form the river Ganga. Snow and glaciers contribute about 29% to the annual flows
of the Ganga (up to Devprayag) and hence any impacts on these glaciers are likely to affect
this large river system [3].Numerous smaller glaciers join the main stream of the main glacier
to form the Gangotri group of glaciers. The study area is Gangotri glacier including it
tributary glaciers such as Maidani Glacier Swachand glacier, Sumeru Glacier, Ghanohim
Glacier and Kriti Glacier system. For ease in writing we abbreviated these all tributary
glacier as Gangotri Glacier System in this investigation.
This system covers an area of 156.587 sq km (ETM+2010).The area of the main trunk of the
glacier 62.412 sq km [12], Average width of the glacier is 1.847 km and glacier, lies
between 79o4’ 46.13” E-79o16’ 9.45” E and 30o43’ 47.00” N-30o55’ 51.05” N
(ETM+2000). It has varying elevation of 4082–6351 meters above sea level (SRTM Data
Analysis)
Fig. 1. False Color Composite (FCC) of RED (4) Green (3) Blue (2) in the
Gangotri Glacier, Subset of Landsat-7 ETM+ Image(G=Glacier, AB= Ablation zone and
AC= Accumulation Zone).
Data Sources
The multi-spectral satellite data of Landsat 8 for the year 2001 and 2016, Landsat TM5 data
for 2001 and 2016, Landsat ETM+ data for the years 2001 and 2016 have been procured in
the present study (see table 1). The Landsat data used in current investigation system was
downloaded for free from the USGS Global Visualization Viewer (GLOVIS). Also used the
advanced wide field sensor (AWIFS) these data used from the sac
Satellite Data
Date of acquisition
Spatial resolution (m)
Landsat MSS
26/10/2001
79
Landsat MSS
19/11/2016
79
Landsat TM 5
21/10/2001
30
Landsat TM 5
13/11/2016
30
Landsat ETM+
25/05/2001
30
Landsat ETM+
13/11/2016
30
AWIFS
25/05/2001
56
AWIFS
13/11/2016
56
Glacier mapping was undertaken employing digital elevation models (DEMs) i.e. ASTER
and SRTM DEM freely downloaded from ASTER GDEM and US Earth Explorer. In this we
investigation we find out that elevation values from ASTER DEM are higher than SRTM
DEM. However SRTM DEM looking smoother but problematic at moraines.
Methodology
The spatial-based detection of variations in glacial extent requires co-registration of
multi-temporal images with one another, a task easily achievable in GIS. GIS is an
efficient tool for analyzing current state and changes in glaciers (Li et al., 1998).
Other analyses such as classification and detection can also be carried out in GIS, as
can measure the glacier area and change in glacier termini. A database within a GIS
may be manipulated to yield information on changes in glacier size. Glaciers can be
mapped by supervised and unsuperevised classification method tested. Pre-processing
data / post processing data such as georeferencing and Orthorectification.
For Landsat MSS data
In this study, a Landsat MSS imagery of Oct 1972 and Nov 1976 covering the
Gangotri glacier system was downloaded for free from the USGS Global
Visualization Viewer (GLOVIS). The main glacier body is almost cloud free.
The terrain of Himalayan glaciers has undulating surface and steep slopes, so the
radiance reaching the sensor greatly depends on the orientation (slope and aspect) of
the target. The incoming radiance is highly depend on the orientation of the
object Therefore, for better recognition of the classes for effective mapping, the
DN numbers have to be converted into topographically correctedreflectance
images. AAR Estimation requires mapping of glacier extent and accumulation area,
Therefore to get accumulation area classification was required so in a second step,
both supervised and unsupervised classification are performed to extract four classes
rock, snow, ice and debris ice. In third step accumulation area and total glacier area
has been measured by visual demarcation and GIS techniques using ArcGIS 9.3.
For Landsat TM and ETM+ data
Earlier studies have shown that normalized-difference snow index (NDSI) and band
ratio methods could not differentiate debris covered glacier ice from surrounding rock
surface due to similar spectral signatures[14], However, when compared with manual
delineation, thresholding of NDSI and band ratio methods are better approaches for
mapping clean glacier ice [14]. It has been reported that for shaded areas with thin
debris cover the band ratio near-infrared/ shortwave infrared (NIR/SWIR) performs
better than red/ SWIR and NDSI [16]. In our study, several image band ratios (1/3; 3/
4), NDSI (1–4/1+4) and classifications were tested using Landsat imagery. Band ratio
NIR/SWIR was most suitable for mapping clean glacier ice. Ratio images have been
successfully used for the delineation of glaciers for the Swiss Glacier Inventory (SGI)
and a study in the Inner Tien Shan [16]. In a first step, TM4/TM5 ratio images were
calculated and segmented using a threshold value of 1 using raster math tool in
ArcGIS 9.3. Using Image enhancement techniques snow and snow free area classified
easily. Ratio Images was associated with ASTER Digital Elevation Model for
mapping glacial extent and accumulation area using visual demarcation. One major
problem in mapping glacier is related to the exact definition of glacier, whether
‘inactive’ bodies of ice above a bergschrund connected to a glacier should be
considered as part of the glacier. Currently, there is no consensus within the
glaciological community on these
issues. For example, some previous studies (e.g. 15) excluded the inactive
parts at the heads of glaciers, so concerning that problem we generated surface
slope image from ASTER DEM using Terrain and TIN surface tool. Now we overlay
delineated area of glacier extent derived from ratio image and surface slope image.
Accumulation area, ablation area and total area of glacier was measured much
accurately and misclassified pixels of peaks and rocky surface are eliminated. ESRI
ArcScene was used for visualization of glacial extent image with digital elevation
model.
Fig. 2.all are images are the change detection inventory on the image of 2001 and
2016
Fig. 3.all are images are the change ditection inventory of the 2001 and 2016 changes
Results and Discussions
The Gangotri glacier is one of the largest glaciers in the Himalayas. Numerous small
sized glaciers also join the main Gangotri glacier from all sides and form the Gangotri
group of glaciers. The main glaciers as well as its tributaries are valley glaciers .The
total ice cover is approximately 156.587 km 2 and estimated volume of ice is 36.17
km 3. The area and length of the main trunk of the glacier is 62.412 sq km and 29.38
km respectively [12]. The average width of the glacier is 1.85 km. The glacier, lies
between 79o4’ 46.17” E-79o16’ 10” E and 30o43’ 46.98” N-30o55’ 50.96” N
(ETM+2000). It has elevation range from 4,017–6,146 meters above sea level (SRTM
data analysis).
Table 2-Measured and estimated characteristics of Gangotri Glacier
Year
Total Area(Sq. km)
2001
1766.744
2016
1763.929
Table 3-Measured and estimated characteristics of Gangotri Glacier
Year
Basin
Total Area(Sq. km)
2001
alakhnanda
867.92
2016
alakhnanda
866.21
2001
Bhagirathi
898.92
2016
bhagirathi
897.71
Table 4- Elevation and topographic analysis of Gangotri Glacier
Characteristics of
Gangotri Glacier
ASTER DEM
Analysis
SRTM DEM analysis
Min altitude
4101
4082
Max altitude
6389
6351
ELA
4684
4649
Relief
2288
2269
Aspect
NW
NW
Slope
5.15 0
5.61 0
In this we investigation we find out that elevation values from ASTER DEM are higher than
SRTM DEM, However SRTM DEM looking smoother but problematic at moraines. Glacier
mapping using a TM4/TM5-ratio image in combination with a DEM was successfully
performed and the accumulation area, ablation area and total area of Gangotri along with
TG’s (Maidani Glacier, Swachand glacier, Sumeru Glacier, Ghanohim Glacier and Kriti
Glacier). Also Our results show that glaciers in the region both retreated and advanced during
the last 15 years; difference between the year 2001 and 2016, average Alakhnanda &
Bhagirathi basin glacier area decreased from 1.71m2& 1.11m2. On average, during the period
2001 and 2016 respectively, suggesting that glacier retreat has an expedition.
Conclusion
The recent advent of Geographic Information Systems (GIS) and Remote sensing
techniques have created an effective means by which the acquired data are analyzed
for the effective monitoring and mapping of temporal dynamics of glaciers.
Longitudinal variations in glacial extent have been detected from multi-temporal
images in GIS. A large number of researchers have taken advantage of remote
sensing, GIS and GPS in their studies of glaciers. In this study, accumulation area,
ablation area and total glacial extent of Gangotri Glacier with TG’s (Maidani
Glacier, Swachand glacier, Sumeru Glacier, Ghanohim Glacier and Kriti
Glacier) mapped using ratio images with the association of surface slope. This study
shows that the glaciers in the region both retreated and advanced during the last 15
years; The retreated and advanced study using remote sensing is good for
monitoring a larger area and the glacier those are difficult to access.
References
[1] Dyurgerov, M.B. and Meier, M.F. 2000.Twentieth century climate change:
Evidence from small glaciers. Proceedings of the National Academy of Sciences
97(4):1406-1411.
[2] Raina, V.K. and D. Srivastava. 2008. Glacier atlas of India. Bangalore, Geological
Society of India.
[3] Singh, P., Polglase, L. and Wilson, D., 2009. Role of Snow and Glacier
melt runoff modeling in Hydropower projects in the Himalayan Region. In (WEES
-2009).
[4] IPCC WGII Fourth Assessment Report 2004.
[5] Chaohai, L. and Sharma, C. K., Report on first expedition to glaciers in the
Pumqu (Arun) and Poiqu (Bhote-Sun Kosi) river basins, Xizang (Tibet), China.
Science Press, Beijing, 1988, p. 192.
[6] Kulkarni, A. V., Rathore, B. P. and Alex, S., Monitoring of glacial mass balance
in the Baspa basin using Accumulation Area Ratio method. Curr. Sci., 2004, 86, 101–
106.
[7] Kulkarni, A. V., Mass balance of Himalayan glaciers using AAR and ELA
methods. J. Glaciol., 1992, 38, 101–104.
[8] Bahuguna, I.M., A.V. Kulkarni, S. Nayak, B.P. Rathore, H.S. Negi and P.
Mather. 2007. Himalayan glacier retreat using IRS 1C PAN stereo data. Int. J.
Remote Sens., 28(2), 437–442.
[9] Kumar, R., G. Areendran and P. Rao. 2009. Witnessing change: glaciers in the
Indian Himalayas. Pilani, WWF-India and Birla Institute of Technology.
[10] Nainwal, H.C., K.S. Sajwan, I.M. Bahuguna and A.V. Kulkarni. 2008.
Monitoring of recession of the glaciers using satellite data: a case study from
Saraswati (Alaknanda) Basin, Garhwal Himalaya. In Proceedings of National
Seminar on Glacial Geomorphology and Palaeoglaciation in Himalaya, 13–14
March 2008.
[11] Ahmad, S. and S.I. Hasnain. 2004. Analysis of satellite imageries for
characterization of glaciomorphological features of the Gangotri Glacier, Ganga
headwater, Garhwal Himalayas. Geol. Surv. India Spec. Publ. 80, 60–67.
[12] Anul Haq and Kamal Jain.2011. Change Detection of Himalayan Glacier
Surface Using Satellite Imagery. In Regional Conference on Geomatics for G-
governance from 13 – 14 September, 2011.
[13] Li, Z., Sun, W . and Zeng, Q. 1998: Measurements of glacier variation in the
Tibetan Plateau using Landsat data. Remote Sensing of Environment 63, 258–64.
[14] Bolch, T., M.F. Buchroithner, A. Kunert and U. Kamp. 2007. Automated
delineation of debris-covered glaciers based on ASTER data. In Gomarasca, M.A.,
ed. GeoInformation in Europe. Proceedings of the 27th EARSeL Symposium, 4–6
June 2007, Bolzano, Italy. Rotterdam, Millpress, 403–410.
[15] Paul, F., A. Ka¨ a¨ b, M. Maisch, T. Kellenberger and W. Haeberli. 2002. The
new remote-sensing-derived Swiss glacier inventory: I. Methods. Ann. Glaciol., 34,
355–361.
[16] Georges, C. 2004. 20th century glacier fluctuations in the tropical
Cordillera Blanca, Peru. Arct. Antarct. Alp. Res., 36 (1), 100–107
[17] Meier, M. F. and Post, A., Recent variations in mass net budgets of glaciers in
western North America, IASH, 1962, 58, 63–77.
[18] Paterson, W. S. B., The Physics of Glaciers, Butterworth-Heinemann, 1998, pp.
26–53.
[19] Vohra C P (1981). Himalayan Glaciers. In: The Himalaya: Aspects of
Change. (Eds. J.S. Lall and Moddie), Oxford University Press, New Delhi, 138-
151.
