1014
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008
• Data availability : The existence of various data sets from
spacebome to airborne sensors ,which are acquired in
different time intervals ,and ground measurements ,which
help in the data processing ,information fusion ,and
validation of the final results .
3. DATA SOURCES
Remote sensing techniques are commonly used to determine
glacier surface displacements. These techniques utilize microwave
data and optical imagery .The focus of the present study is to use
multi-temporal optical satellite imagery and aerial photographs .In
order to measure surface displacement from multi-temporal
imagery, the following conditions must be met:
• Surface features have to be detectable in at least two of the
multi-temporal optical data sets in order to track them.
• The multi-temporal data sets have to be accurately
coregistered.
• The spatial resolution of the image has to be finer than the
displacements.
In this research ,High resolution satellite imagery including
Quickbird image with 2,5 m spatial resolution , SPOT 5
panchromatic image (2.5m) acquired at the end of September are
utilized along with six aerial photoghraphs of 1:40000 scale.
4. METHODOLOGY
If the two images are correlated, the resulting offsets in the image
lines and columns are the sums of the contributions from
misregistration, topography, orbits, and attitude as well as the
glacier-dynamics signal. To obtain a valid measurement, we must
remove all the contributions except the glacier flow (E. Berthier et
al:’ 2005).
4.1 Optical satellite imagery
In order to measure glacier displacement map ,Spot 5 and multi-
spectral Quickbird images respectively acquired at September
2003 and 2005 from study site are used .In the first step,for
accurate change detection ,images must be radiometricllay
normalized. Afterwards,performing image to image co
registration is a necessary step to provide a solid base for image
matching and change detection analysis.
Figure. 1: Flowchart of the methodology followed to measure the
surface displacements
4.1.1. Radiometric normalization
The first step of using multi-temporal satellite imagery in this
procedure is to perform normalization so that they share the
similar radiometry ,one of the conditions for an accurate
displacement computation.
Ground reflectance determination from satellite imagery requires,
among other things, an atmospheric correction algorithm and the
associated atmospheric properties at the time of image acquisition.
For most historical satellite scenes such data are not available and
even for planned acquisitions or they may be difficult to obtain.In
this case a relative normalization using the radiometric
information must be performed .In this work ,images were first
normalized by ENVI4.4 software using IAR reflectance method.
Thereafter accurate relative normalization procedure was
conducted utilizing time-invariant objects such as shelter for
climbing groups.
4.1.2. Co-registration of multi-temporal satellite imagery
A critical requirement in change detection applications is that
multitemporal images must be co-registered accurately. In this
study, an image to image co-registration was carried out with a
sub-pixel resulting RMSE of about 0.039 pixel.
4.2 Airborne photographs
4.2.1 DEM & ortho-photo generation
As mentioned earlier ,six aerial photos are used in this research. In
order to generate DEM, firstly ground control points for
photogrametric processing were acquired through a Bundle Block
Adjustments Aerial Triangulation. The DEM was produced using
VirtozouNT photogrammetric software. Thereafter ortho image
mosaic of the entire area was generated.