8/17/2019 Tf April 2012 Web
1/20
A monthly house bulletin of Defence Research & Development Organisation ■ Vol. 31 No. 7 ■ July 2011ISSN : 0971-4413BULLETIN OF DEFENCE RESEARCH AND
DEVELOPMENT ORGANISATIONVol. 20 No. 2 April 2012
Troops guarding the high altitude snowbound
border areas are always in perpetual danger
of losing life due to avalanche. Forecasting
of avalanche can help in minimising such incident.
One of the mandate of DRDO is to facilitate high
operational mobility of troops in snowbound high
altitude avalanche prone areas. Snow and Avalanche
Study Establishment (SASE), a constituent laboratory
of DRDO, provides precision avalanche forecastingsupport to the Services including advice on
avalanche control measures and enhance avalanche
forecasting through systematic data collection in
snowbound areas. The task involves collection of
snow and meteorological data at every hour and at
a closely spaced grid of less than a kilometer from
dierent altitude ranges of the Himalaya.
Given the ruggedness of the terrain and
inhospitable conditions, it is not possible to collect
data at high temporal and spatial resolution. Even
automatic weather stations (AWS) do not suce
since it is not possible to install and maintain
these at some places. In such a scenario, remote
sensing is the only alternative to collect data of a
larger area. SASE has made use of remote sensing
technology since 1998 for better understanding and
monitoring of cryospheric regions of the Himalayas.
Remote sensing has augmented the conventional
measurements, which SASE has been taking at
select points since 1969. SASE has also started
collaborative research with Space Application
Centre, Ahemdabad; Centre of Studies in Resources
Engineering, Indian institute of Technology, Bombay
(IIT-B), Mumbai; IIT-K, Kanpur; and Department of
Geomatics Engineering, IIT, Roorkee.
The Establishment has a strong foundation and
extensive understanding of the remote sensing
technology and its application to snow and
avalanche-related studies. It is fully equipped with
the state-of-the-art software and hardware facilities
to make use of optical and microwave satellite data.
The Establishment has developed methodologies/
algorithms for snow and avalanche-related
applications and has taken up research in Polarregions also.
Some of the new applications and technologies
pertaining to avalanche hazard assessment, snow
cover monitoring, topographic parameters using
multispectral, hyperspectral, microwave and other
remote sensing data are being highlighted in this
special issue of Technology Focus .
8/17/2019 Tf April 2012 Web
2/20
2
From the Special Editor
The mountain ranges of Himalayas form a distinct geographical divide that separates the Indian
subcontinent from Central Asia. The area is strategically very important and affects the socio-economic
development. The rugged foothills of Himalayas are carved into deep gorges and ravines by innumerable
streams. The road connectivity to these areas is rather short due to climatic conditions. The passes are
closed as they remain covered with snow during most of the year. The roads are plagued by frequent snow
avalanches, etc.
Himalaya experience severe snowfall of varying magnitudes during the winter period. The snow
accumulation depends on altitude and other geographical parameters of the region. The unstable snow
pack results in avalanches that cause danger and loss to human lives and property, and also causes a lot of
hindrance in transportation, communication and deployment of army personnel and inconvenience to civilian
population in these areas. The hazard potential of the region is a cause of concern for us at SASE especiallythe ones directly emanating from the interaction of snow, ice, terrain and weather. SASE has been providing
yeoman’s service for more than four decades in combating the hazards due to avalanches and the foremost
menace of avalanches have been controlled to a large extent due to the sustained efforts of its scientists and
the constant help rendered by the Indian Army.
Operational avalanche forecast primarily depends on the snow-met parameters. SASE’s network of
existing ground observatories, automatic weather stations and radio-based remote telemetry systems for
collection of snow and meteorological parameters provide only point information and are often very sparse
and unevenly spatially distributed, which needs to be extrapolated to cover a bigger region of interest. The
manual collection of data is extremely difcult in rugged nearly inaccessible terrains. SASE made a humble
beginning in establishing the remote sensing group in early nineties for better understanding and handling
of the avalanche problems and monitoring of snow-covered terrain of Indian Himalaya. Remote sensing
augments the conventional measurements as its repeativity is high and the area coverage is quite large.
Presently, SASE is fully geared with the latest state-of-the-art software and hardware facilities to deal with
optical and microwave satellite data, GIS-based applications and terrain visualisation. SASE also undertakes
the reception of the optical and microwave satellite imageries, analysis of received data, development of
methodologies and algorithms for snow and avalanche-related applications for the Indian Himalaya and
generation of nal products for real-time monitoring of the snow-cover and hazard analysis.
This special issue of Technology Focus brings out the various remote sensing technologies developed
by SASE in respect of high altitude and cold region engineering, avalanche forecasting, avalanche control
measures, virtual y through models, controlled release of avalanches, hazard zonation, etc.
(Ashwagosha Ganju)
Director
Snow & Avalanche Study Establishment, Chandigarh
8/17/2019 Tf April 2012 Web
3/20
3
Snow-met Parameters and Radiation Fluxes
Extraction of Atmospheric Profiles of
Temperature and Moisture
Moderate resolution imaging spectroradiometer
(MODIS) is a key instrument onboard the Earth
observing Terra and Aqua satellites. It provides high
density proles of temperature and humidity at a
resolution of about 0.1o in latitude, 0.25o in longitude,
and 20 pressure levels in the vertical. These proles
are useful for capturing the meso to micro scalesthermodynamic elds, which help in better simulation
of localised terrain-disrupted airow (related to the
temperature prole in boundary layer) and relative
humidity distribution.
SASE has developed an application for the extraction
and analysis of the atmospheric proles of temperature
and moisture from MODIS data. These proles were
compared with the radiosonde (upper air prole) data
for accuracy and are now being used in numericalweather prediction models.
TECHNOLOGY DEVELOPMENT
Application for MODIS profile extraction.
Soft Computing-based Satellite Data Segmenta-
tion for Snow and Land Features Extraction
Classication of a multispectral satellite image is a
challenging task and has a number of applications such
as feature identication, change detection, etc. Various
soft computing methods and articial intelligence
techniques like neural network, fuzzy logic and support
vector machine are being used for classication of
the optical satellite―MODIS and advanced wide eld
sensor (AWiFS)―
for snow-covered areas, its types, andother land features.
Schematic of the work flow.
GUI of the fuzzy classification module.
8/17/2019 Tf April 2012 Web
4/20
4
Classified map of an area.
Another application to implement these techniques
for classication of multispectral images has been
developed. This application comprises dierentmodules for dierent techniques like GUI of fuzzy
classication module and classied map using MODIS
data.
Snow depth map.
Snow water equivalent map.
Modelling for Snow Depth/Snow WaterEquivalent Estimation
Operational Algorithm for the Estimation of
Snow Depth and Snow Water Equivalent inWestern Himalayas
Varied inputs in the form of snow parameters (e.g.
snow water equivalent) are required in scientic
models pertaining to avalanche and weather
forecasting. These parameters are usually collected
from eld observatories at dierent locations in the
snowbound regions. However, the vast snow-covered
areas in high altitudes are remote and inaccessible.
The integration of remote sensing data supplemented
with eld snow-depth measurements is an eective
way for the accurate determination of snow depth and
snow water equivalent at spatial level.
SASE and IIT Roorkee have jointly developed a spatial
interpolation model for estimation of snow depth and
snow water equivalent at each 500 m MODIS pixel
resolution. The model uses discrete point data to
create a model of the snow depth from which a value
for any location can be estimated. The proposed spatial
interpolation method is based on snow depth and
altitude above mean sea level. The dependency is later
adjusted through in situ snow depth observations to
represent the local and regional characteristics of the
snow distribution. The model has been further rened
by snowline variations in specic areas retrieved from
MODIS sensor data. It estimates snow water equivalent
from snow depth maps generated using snow-covered
area maps retrieved at sub-pixel accuracy and snow-
covered density observations.
Snow depth estimation has also been attempted in
parts of north-west Himalayas in the GIS-framework.
A quasi physical method has been developed for
the estimation of snow depth. To generate snow
depth map from observation point, snow depth is
8/17/2019 Tf April 2012 Web
5/20
5
calculated at each pixel of the study area with respect
to observation point using the elevation information
from digital elevation model (DEM). A weighted
inverse distance technique has been used to calculate
the depth at each pixel from the observation point. The
nal snow depth map has been generated by taking
weighted sum of the snow depths of all observation
points. The calculated snow depth values have been
validated using data from AWS.Hyperspectral Remote Sensing for SnowCharacterisation
Hyperspectral Spectroscopy
Hyperspectral sensors capture data in contiguous
narrow bands (~10 nm spectral resolution) of the
electromagnetic spectrum and allow whole spectral
curves to be recorded with individual absorption
features. Hyperspectral data is, therefore, used for
characterisation, quantication, identication anddetection of subtle changes in snow in the snow-
covered areas. SASE has generated a library of spectral
signatures of dierent type of snow/ice and other
ambient objects by conducting eld investigations
using spectroradiometer (350-2500 nm).
Snow depth map from a hybrid model.
SASE has also carried out experiments to understand
the inuence of size of snow grain, contamination,
moisture, snow depth, slope aspect, and snow mixedobjects on spectral reectance and to determine the
sensitive/suitable wavelengths for mapping of snow
and estimation of snow characteristics using satellite
data.
Collection of spectral signature of snow using spectroradiometerin eld.
Hyperspectral Imaging
The Establishment has also explored space-borne
Hyperion sensor data for the estimation of snow-cover
characteristics in the Himalayan region. Snow grain
size was estimated using spectral angle mapper (SAM)
method. The retrieved grain sizes were compared/
validated with grain sizes retrieved from grain index
and asymptotic radiative transfer (ART) theory-based
methods. A very good overall accuracy of matching
snow-grain size classes was observed among dierent
methods.
The fractional snow-cover maps were also generated
using Hyperion data and image spectra were validated
using mix-object snow spectra collected from eld
experiments. Further, dierent level of vegetation/
8/17/2019 Tf April 2012 Web
6/20
6
soil mixed snow-cover areas were delineated using
Hyperion data and validated using high resolution
satellite data.
Modelling for Snow Albedo Estimation
Snow Albedo Estimation using Reflectance
Measurements
The Establishment has developed a modelto retrieve snow albedo from spectral reectance
measurements. The ART theory has been applied
to retrieve the plane and spherical albedo from the
reectance observations. The retrieved plane albedo
was compared with the measured spectral albedo and
a good agreement, with only ±10 per cent dierence,
was found between the two. Retrieved integrated
albedo was also found in good agreement, within ±6
per cent dierence, with ground-observed broadband
albedo.
This methodology was also implemented for the
retrieval of spectral albedo using single observation
from the satellite data, and found very useful for
operational snow-cover and glacier monitoring of the
Himalayan region using space- and air-borne sensors.
Algorithm to Estimate Broadband Albedo of
Snow
Algorithms to estimate narrowband to broadband
albedo (NBBA) of snow using AWiFS and MODIS
sensor images have been developed for the westernHimalayan region of India. The in situ measurements
of spectral reectance and transmitted spectral solar
irradiance of snow surface by spectroradiometer have
been used to calibrate and validate these algorithms.
Snow grain size retrieved from Hyperion data using different methods for a part of the Himalayan region.
Spatial distribution of reectance at 1240 nm and its histogrammeasured by Hyperion over a part of middle Himalaya.
8/17/2019 Tf April 2012 Web
7/20
7
The AWiFS and MODIS snow broadband albedo derived
using the developed algorithms have been validated
with in situ observations at dierent eld locations. An
RMSE better than 0.03 and correlation of 0.94 and 0.88
between modelled and observed albedo values have
been obtained for AWiFS and MODIS, respectively.
Snow Grain Size Estimation Snow Grain Size Estimation from Reflectance
Data
A model has been developed to retrieve snow grain
size from reectance data using dierent models
based on ART theory and compared with dierent
snow types in the Himalayan region. This includes
single-channel, two-channel (bi-spectral), two-channel
ratio, and three-channel methods. It was found that
the grain size model using bi-spectral method, one in
visible and another in near infrared (NIR) region, works
well for the seasonal snow-cover in the Himalayas and
is in good agreement with temporal changes of grain
size as the season progresses. It can also take care of
soot eect (if present) in NIR region. This model was
also implemented on Hyperion sensor data for the
Himalayan regions. Only ve spectral bands (440, 500,
Snow Grain Size Estimation from SAR Data
Snow grain size is estimated from SAR data in a
dry snow pack. A dry snow layer is a heterogeneous
medium composed of ice particles with dierent size
and microstructures. In dry snow, volume scattering
from the snow pack is the the principle mechanism
and density and grain size are the most important
variables. Backscattering increases as the size of snow
grain increases. The sensitivity of radar cross-sectionto grain size was assessed by changing the volumetric
correlation length of snow grains.
Net Shortwave Radiation Flux
An algorithm has been developed to estimate net
radiation ux over large snow-covered areas of north-
1050, 1240 and 1650 nm) of Hyperion data were used
for snow grain size retrieval. The grain sizes retrieved
using satellite were compared with grain sizes retrievedfrom eld spectroradiometer and also validated with
snow meteorological data collected from eld. The
model was able to retrieve the spatial variations in
snow grain size parameter in dierent parts of the
western Himalaya, which is natural given the variability
that exists in snow climatic and terrain conditions
of the Himalaya. This methodology is important for
operational snow-cover and glacier monitoring in the
Himalayan region using space- and air-borne sensors.
Broadband albedo of snow-covered area using MODIS data.
Spatial distribution of snow grain size retrieved using 1240 nm byHyperion for a part of lower Himalaya.
8/17/2019 Tf April 2012 Web
8/20
8
west Himalayas for clear sky days using AWS and MODIS
data.
Geospatial maps of air temperature and relative
humidity have been generated using AWS data and
DEM of the study area. These geospatial maps forms
inputs for the parameterisation scheme used for
estimation of incoming shortwave radiation at spatial
level. Geospatial maps of the snow-covered albedo
were generated from MODIS sensor data. Finally,
geospatial maps of net shortwave radiation ux
were generated from the input of incoming shortwave
radiation ux and albedo maps.
Snow Density Estimation using SAR
An empirical model has been developed to estimate
snow density from SAR backscattering. In the absence
of free water and ice layers in the snowpack, the
microwave scattering and emission behaviour are
governed by the snowpack depth and its density.
The complexity in behaviour of snow density to the
total radar backscattering makes it dicult to develop
a model for estimating snow density from singlepolarisation C-band SAR data. Therefore, an empirical
model has been adapted. The total backscattering has
been considered as a third-order polynomial of snow
density.
Net shortwave radiation flux.
Incoming shortwave radiation flux.
Retrieved grain size from SAR data.
8/17/2019 Tf April 2012 Web
9/20
9
Snow Wetness Measurement using SAR
An algorithm has been developed to estimate
snow wetness using Envisat-ASAR C-band data from
the estimated dielectric constant and snow density.
Dielectric constant was derived from back scattering
coecient through inversion equation model.
APPLICATION/ENABLING TECHNOLOGIES
Snow density map of Beas Kund.
the classication of image, thereby decreasing the
classication accuracy of the image contaminated by
mixed pixels. Sub-pixel classication overcomes this
mixed pixel problem by predicting the proportionalmembership of each pixel to each class. The output of
sub-pixel classication is not a single classied image
but a number of images known as fraction images.
Linear mixture model (LMM) sub-pixel classication
technique has been applied for retrieving snow-cover
fraction images in north-west Himalayas and an overall
decreasing snow cover trend has been observed.
Snow-cover Mapping using SAR
Microwave remote sensing is a tool with potential to
estimate snowpack parameters. SASE is working on the
development of algorithms/models/methodologies
using active microwave data of ENVISAT advanced
synthetic aperture radar (ASAR), TerraSAR-X,
RADARSAT-1/2 and ALOS PALSAR sensor data for snow-
Snow wetness (per cent ) estimation using ASAR C-band data.
Snow-cover Mapping, Monitoring and
Snowmelt Run-o Modelling
Snow-cover Monitoring using MODIS and AWiFSData
Snow-cover monitoring in the north-west Himalayas
and its various sub-basins have been carried out using
AWiFS and MODIS sensor data. Binary maps of snow-
covered areas generated using green band and short
wave infrared band (SWIR) of the MODIS and AWiFS
images at sub-pixel accuracy, as per-pixel classication
techniques, have limitations in classifying images
dominated by mixed pixels.
A mixed pixel displays a composite spectral response
that may be dissimilar to the spectral response of
each of its component land cover classes. Therefore,
pixel may not be allocated to any of its component
land cover classes. Hence, error is likely to occur in
8/17/2019 Tf April 2012 Web
10/20
10
North-west Himalayas and its different basins.
covered area mapping and monitoring and estimation
of snowpack parameters. ENVISAT-ASAR C-band data
has been used for estimation of snow-cover area using
Snow and non snow-cover mapping using SAR data.
multi-date image threshold technique for the Manali
(HP) region.
Snow Melt Run-off Modelling
A well-established ‘Degree Day’ approach has
been used for the estimation of snow melt and river
stream ow of a basin. However, the important inputs
used in snow melt run-o modelling, which includes
seasonal snow-covered areas, glacier extent, etc.,
were derived using remote sensing techniques and
hydro-meteorological data obtained from ground
observatories. For estimation of snow-covered areas,
MODIS sensor images were used. The model output is
in good agreement with the in situ discharge data.
Glaciers and Polar Region Study
Sea-Ice Study using Passive Microwave Satellite
Data
Sea-ice cover acts as a barrier between ocean and
atmosphere and thus aects the energy, gases and
momentum transferred between them. Conversion
8/17/2019 Tf April 2012 Web
11/20
11
Comparison of actual and calculated stream flow in Jhelum basin (2004-09).
of open sea water into the frozen ice reduces the
amount of absorbed solar radiation and is one of
the main component responsible for global climatic
perturbations.
The sea-ice around Antarctica is generally seasonal,
i.e., it melts in summer and again freezes in winters.
Due to cold and hostile climate and uctuating
sea-ice conditions, in situ collection of data is very
dicult. Some automatic data collection methods
have been used worldwide but still monitoring of
dierent cryosphere features (shelf ice, sea ice and
seasonal snow-cover), which cover millions of sq km,
is not possible in the true manner. Passive microwave
remote sensing satellites are promising tools for global
monitoring of the cryosphere with high temporal
repetition rate and also due to their working capabilityduring day and night.
Sea-Ice Concentration and Sea-Ice Areal Extent
Special sensor microwave imager (SSM/I) data has
been used for the estimation of sea-ice areal extent
(SIAE). SSM/I scans the earth at the frequencies of 19.3,
37 and 85.5 GHz in vertical and horizontal polarisations
and at 22.2 GHz in the vertical polarisation only. The
Sea-ice concentration (October 2011).
8/17/2019 Tf April 2012 Web
12/20
12
ascending mode data at 25 km resolution has been
used. In processing, rst brightness temperature values
are estimated for dierent frequency and polarisation.Daily, weekly and monthly brightness and temperature
maps have been generated and temporal variation has
been studied. Polarisation ratio (PR) and gradient ratio
(GR) have been calculated using three SSM/I channels,
i.e., 19 GHz (V), 19 GHz (H) and 37 GHz (V). These PR and
GR values were further used as inputs for estimation of
rst year sea-ice concentration (CF) and multiyear sea-
ice concentration (CM). Estimated CF and CM values
were used for estimation of total sea-ice concentration
(CT). To estimate the SIAE, concentration values weremultiplied with the area of pixel, i.e., 25*25 km2.
The MODIS sensor data onboard Aqua satellites
have also been used for the estimation of SIAE. The
validation of SSM/I retrieved SIAE has been done using
the MODIS imageries (1 km spatial resolution) and
linear regression equation generated for corrected
SIAE values using daily available SSM/I satellite data.Glacier Monitoring using Geomatics Techniques
SASE has recently started to monitor a few Indian
Himalayan glaciers using ground, air- and space-borne
geomatics techniques. The main objective is to monitor
Variation of sea-ice areal extent around Antarctica (1988-2011).
Field photograph collected at the end of ablation season of aglacier near Patsio (HP) showing the terminus and snow lineposition (top) and Landsat-7, ETM+ imagery of the same glacier(bottom).
the response of climate change on the Himalayan
glaciers and its associated hazards. SAR interferometry
technique has been used for producing high resolutionDEM and changes on the glacier surface. The
measurement of interferometer correlation provides
the information of changes during the time scale of the
satellite receptivity and size scale on the order of a SAR
wavelength.
8/17/2019 Tf April 2012 Web
13/20
13
Antarctic Ice Sheet Study
Models and methodologies have been developed
using MODIS sensor data for spatial and temporalvariability study of snow/ice albedo, radiative uxes
and other parameters in the Dronning Maudland, east
Antarctica.
Ground Penetrating Radar Survey
Airborne Ground Penetrating Radar Survey for
Snow and Glacier Applications
A ground-based radio echo sounding equipment
is very dicult to operate for snow and glacier studies
in the Indian Himalayas because of the rugged terrain
and topography. Also, a large number of buried/hidden
crevasses can be hazardous during the eld survey.
However, helicopter-mounted air-borne ground
penetrating radar (GPR) has proved to be very useful
for quick data collection from remote and inaccessible
snowbound areas.
The potential of an airborne GPR was explored for
the estimation of snow depth over rugged Himalayanterrain. The 350 MHz antenna was mounted on a
helicopter for the estimation of snow depth over a
glaciated and non-glaciated area of the north-west
Himalaya. The snow depths estimated from GPR signal
were found in good agreement with the snow depths
MODIS image of Dronning Maudland, Antarctica. Inset: Spatial variation of snow/ice albedo near Maitri region.
measured on ground. A GPR survey was conducted at
dierent locations of the north-west Himalayas. The
calibration and validation of estimated snow depthwere carried out using dierent snowpack properties
measured at an experimental site. A detailed survey
was carried out along Samudratapu glacier for 2009-
2010 and a snow accumulation map was generated
using GPR survey. Changes in the snow accumulation
duirng 2009-2010 were recorded for Samudratapu
glacier. Subsequently, a change detection analysisof snow accumulation
was also conducted
over Samudratapu
glacier during 2009-
2010. For the estimation
of changes in snow
accumulation, the snow
depth information
from common GPS
coordinates of survey
data from both theyears were selected.
Thus a limited area was
obtained using numbers
of common points to
nd the changes in snowAirborne GPR assembly.
8/17/2019 Tf April 2012 Web
14/20
14
accumulation in one year. It was observed that snow
depth, for a part of Samudratapu glacier, was lower in
2010 in a fairly larger area of the glacier than in 2009.This was conrmed from the cumulative snowfall data of
2009 and 2010 recorded at meteorological observatory
at Patseo. This proved that the retrieved data using
GPR can identify change in snow accumulation pattern
in a glacier.
Changes in the snow accumulation between 2009-2010 for a partof Samudratapu glacier.
Snow accumulation over Samudratapu glacier estimated fromairborne GPR (March 2010).
Ground Penetrating Radar Experiments for
Snow Depth and Snow Layer Interface
Estimation
Snow depth and snowpack stratigraphy along
with its temporal and spatial evolution are important
inputs for operational avalanche forecasting and
snow-water equivalent assessment for hydrological
applications. GPR with antenna frequency of
1000 MHz was used for snowpack characterisation in
Pir Panjal and greater Himalayan range. Snow depth
was estimated at certain locations in both the ranges
and further validated with the ground measurements.
The estimated snow depth using GPR at Solang (PirPanjal) was having a good correlation coecient
(0.86) with the manually observed values. Snow fork
was also used for the calibration and validation of GPR
data, which gives dielectric constant, volumetric water
content and density of the snowpack.
Experiments were also conducted at Patseo for
snowpack layer identication. By analysing the
proles, the prominent snow layers present within
the snowpack at Patseo were detected. Manual
stratigraphy was also performed along with the GPR
proles and it was found that layer positions in the
radargram correspond fairly well with the startigraphic
layer positions. Real and complex dielectric constant of
snow, which are important parameters for acquisition
of GPR proles were also measured using snow fork.
Topography and Avalanche Hazard
Assessment
Generation of Digital Elevation Models from
Cartosat-I Satellite Imagery
The Cartosat-1 satellite has a pair of panchromatic
cameras having an along-track stereoscopic capability
using its near-nadir viewing and forward viewing
telescopes to acquire stereo image data with a base-
to-height ratio of about 0.63. The spatial resolution is
8/17/2019 Tf April 2012 Web
15/20
15
Snow depth estimation using 1000 MHz GPR antenna. Snow pack charectrisation using snow.
Snow layer interface identification and validation.
2.5 m in the horizontal plane. Each camera has a pixel
array of size 12000 giving a swath of about 27 km.
Presently, Cartosat-1 is the only global stereo capable
satellite for carrying out scientic studies.
The methodology adopted to produce the CartosatDEM involves stereo-strip triangulation of stereo pairs
using high precise ground control points, interactive
cloud-masking and automatic dense conjugate pair
generation. The automatically generated DEM was
further evaluated for quality and editing to remove
anomalies. Two images of same area were taken from
dierent angles using Cartosat-1. Stereo correlation
was applied to extract the matching point in two
stereo images. Sensor geometric model was used
to calculate the elevations. Rational polynomial
coecients were supplied with imagery product anddene the relationship between normalised pixel
and normalised ground coordinates. The rational
polynomial coecients and DGPS points were used by
Photogrammetric software to transform the ground-
to-image geometric correction.
8/17/2019 Tf April 2012 Web
16/20
16
their ability to predict the avalanche hazard and avoid
the dangers they pose. There are three main causative
variables that inuence the occurrence of avalanches:the terrain characteristics, snowpack conditions, and
prevailing meteorological conditions. GIS provides
the possibility to address these complex, spatially and
temporally distributed and multi-parameter dependent
problems. As GIS framework provides capabilities of
data integration/management, geospatial/temporal
analysis and presenting the results in the map/report
forms. The framework was used for the dynamic
avalanche hazard zonation to generate more accurate
and dynamic avalanche hazard maps.
The methodology combines expert knowledge,
computational routines and statistical analysis to
identify the areas aected by avalanche threats. A high
resolution DEM was used for extracting the desired
DEM with a resolution of 10 m derived from Cartosat-1.
GIS-based Dynamic Avalanche Hazard
Zonation
The avalanche researchers need to understand the
spatial and temporal patterns of avalanches to improve
Avalanche hazard zones based on (a) terrain parameters, (b) snow-met parameters and (c) nal hazard map obtainedby combining (a) and (b).
8/17/2019 Tf April 2012 Web
17/20
17
terrain parameters. The model has three distinct sub-
modules. The rst sub-module denes the avalanche
hazard zones on the criteria related to slope, curvatureand land cover. Weights and ratings to these causative
factors and their cumulative eects were assigned
on the basis of experience and knowledge of eld
experts. The second sub-module uses inverse distance
weighted method to generate the meteorological
parameters (air temperature, precipitation and wind)
maps from the eld observatories and AWSs located at
Manali, Dhundi, and Patsio. Weights and ratings to the
meteorological parameters were assigned on the basis
of regression analysis of past avalanche occurrenceand meteorological data. In the third sub-module, the
meteorological-parameter maps were superimposed
on the terrain-based avalanche hazard maps to
generate the dynamic avalanche hazard maps.
Maps of the meteorological parameters, generated
from the observatory and AWS data, were used for
generation of the meteorological parameters-based
hazard maps. Finally, weighted meteorological
parameter maps were combined with the avalanchehazard map (terrain parameters based) using overlay
method to generate the nal avalanche hazard maps
of the study area.
GIS-based Identication and Mapping ofAvalanche Hazard Areas
Avalanche Hazard Data Cards
Avalanche hazard data cards have been prepared
using remote sensing and GIS techniques for
identication and mapping of avalanche proneareas. The data cards have information about DEM,
slope, and ground cover maps validated with ground
reconnaissance. These data cards include salient
features of respective areas, past major avalanche
accidents, route charts with the details of avalanche
hazards. The cards are in the folded form for its easy
handling and reference during move. A large number
Avalanche hazard data card.
8/17/2019 Tf April 2012 Web
18/20
18
of hard copies of each avalanche hazard data card have
been issued to the users.
Digital Avalanche Atlas
SASE has successfully developed a GIS-based digital
avalanche atlas, covering the vast avalanche prone
areas of various parts of the Indian Himalayas. Besides
having all the information available in the conventional
atlases, the digital avalanche atlas is having updatedinformation with wider selection criteria for each site.
DEM, slope aspect, track prole, digital terrain model
(DTM), etc., are some of the GIS-based characteristics
Digital avalanche atlas (top) and its different modules (bottom).
8/17/2019 Tf April 2012 Web
19/20
19
for each site that can be accessed by selecting the
desired option on the graphic user interface (GUI). This
facilitates the user to view the potential avalancheprone areas as well as specic avalanche sites
interactively. Some other useful features are: general
information about avalanches, dierent control
measures, rescue operations, past history of avalanches,
and accidents. It also provides wide selection criteria
for accessing information for specic avalanche sites
for eective decision making. The digital avalanche
atlas is a speedy and user-friendly tool for military
Commanders, BRO personnel and other authorised civil
authorities for identifyingavalanche prone regions,
area familiarisation,
educating the local people
and provides necessary
information to mitigate
the avalanche hazard.
High Resolution Digital
Terrain Mapping and
Avalanche Hazard
Zonation
One of the
primary products of
photogrammetric data is
very accurate and precise
DEM. To exploit the
potential of this state-of-
the-art technology, SASE
had conducted a large
format digital camera-
based aerial survey, covering Manali and its nearbyareas. An aircraft, equipped with global positioning
system (GPS) and inertial measurement unit (IMU),
was used for photogrammetric surveys (altitude range
~23,000-26,000 ft). Large number of digital imagery
were collected using large format digital mapping
Photogrammetricsurvey area.
Contours generated fromphotogrammetry technique.
camera system (focal length = 100.5 mm; pixel size = 6
µm) with forward and side overlaps of 80 per cent and
60 per cent, respectively. For camera calibration andaerial triangulation, 23 ground control points (GCPs)
were established using dual frequency DGPS (PDOP
< 3). Overall accuracy of DTM achieved was between
7-20 cm GSD. Ortho-rectication of images was carried
out and the ortho-rectied mosaic was created at 0.20
m pixel size. DTMs were further analysed to locate
the probable avalanche release zones that was not
possible earlier at coarse resolution data. The detained
high resolution cartographic maps can help in the
preparation of detailedcivil engineering designs,
e.g., bridges, tunnels,
roads, buildings, etc., on
mountain slopes.
The detailed maps also
help in understanding
geomorphic details
of mountain terrains
including the past
glaciations. Thus, conti-
nuous and wide-area
coverage with digital
camera is very cost-
eective technique since
individual measure-
ment points can strongly
deviate due to dierent
error sources, such as
steep slope, x, y shift,
image correlation, scan position, etc. The study hasamply shown that aerial photogrammetric data
can help in qualitative and quantitative mapping
and monitoring of snow cover, glacier movement,
snow accumulation pattern, and onset of avalanche
occurrence, etc., with passage of time.
Rockwell CommanderC-690 aircraft
Captured image.
Digital aerialphotograph
of Dhundiavalanche.
8/17/2019 Tf April 2012 Web
20/20
Technology Focus focuses on the technological developments in the Organisation, covering the products, processes and technologies.
Editorial Committee
Coordinator
Dr AL Moorthy, Director, DESIDOC, Metcalfe House, DelhiMembers
Cmde PK Mishra, Director of Naval Research & Development
DRDO Bhavan, New Delhi
Shri Sudhir K Mishra, Director of Missiles, DRDO Bhavan, New Delhi
Dr K Muraleedharan, Director of Materials, DRDO Bhavan, New Delhi
Dr Rajeev Varshney, SO to SA to RM, DRDO Bhavan, New Delhi
Virtual Reality and Virtual GIS Lab
A virtual reality laboratory has been set-up as part ofSASE’s strategy in avalanche safety and rescue training
mission. The set-up is used for familiarisation of
troops about harsh rugged snowbound mountainous
regions of norht-west Himalaya in the virtual format.
Pre-knowledge of the terrain features, avalanche
locations, formation zone width and run-out zone
length increases chances of survival in an avalancheprone area. The hardware set for virtual reality lab
consists of CRT projector for stereoscopic projection,
at white board screen for 3-D stereo display and 3-D
active glasses for immersive feel. Three-dimensional
terrain rendering has been done by SASE using high
resolution satellite data.Virtual reality lab at Chandigarh.
Virtual reality visualisation using high resolution data.