Geometrical changes of glaciers, creeping mountain permafrost
and slope movements are caused by a sum of complex 3-dimensional
processes. Monitoring surface kinematics helps understanding the
dynamics of glacial and periglacial processes and investigating
the reaction of glaciers and mountain permafrost to climatic forcing.
Aerophotogrammetric determination of digital terrain models (DTM)
and subsequent comparison of multitemporal DTMs is an effective
and well-established technique to exactly define terrain surfaces
and their changes in elevation. The method uses monotemporale
stereo models, composed by at least two overlapping photographs
which are taken from different places. The terrain point A (see
the figure) is computed by intersecting two spatial rays, each
fixed by the known projection centers and the projections (A1'(t1)
and A2'(t1)) of the selected terrain point. Repeating the procedure
at other points gives a DTM of time t1, and repeating the procedure
using photographs taken at time t2 gives point B respectively
a DTM of time t2 and thus the area-wide changes in surface elevation.
These vertical displacements are the result of spatial/dynamic
processes. Thus, determination of surface displacements in full
three dimensions helps the understanding of these processes and,
therefore, improves their mapping, monitoring and modelling. The
analytical photogrammetric method presented here uses multitemporal
stereo models composed by aerial photographs taken at different
times and taken from different places (e.g. Photo 1 (t2) and Photo
2 (t1) in the figure). Between time t1 and time t2 the point A
has moved to point C, like "swimming" on the surface. A block
or stone on the surface, a crevass or other features, for instance,
could be a suitable target. The projection (A2'(t1)) of such a
point is chosen from the photograph taken at time t1. Intersecting
the spatial ray fixed by this image point and the known projection
center with the terrain surface represented by the DTM of time
t1 gives the ground coordinates of point A. This procedure is
called monoplotting. The image point C1'(t2) which corresponds
to A2'(t1) can now be found using the stereoscopic overlap. The
operator thereby cancels the terrain movement which has occurred
between t1 and t2 by displacing one image while simultaneously
looking at the multitemporal photographs. This simultaneous and
stereoscopic observation supports the identification of corresponding
points, improves the accuracy of the measurements and, additionally,
indicates whether a local displacement reflects in a significant
way its surrounding terrain.
In the same way as for the spatial position of point A, the position
of point C is computed by monoplotting. Thus, spatial displacements
of surface points can be deduced area-wide. The terrain surface
of which the velocity field shall be determined has to fulfill
two basic requirements: the displacements must be larger than
the accuracy of the method to obtain significant results and,
second, terrain deformations or destructions between the times
of photography, e.g. caused by thawing or terrain slip, should
not prohibit the identification of corresponding points. Both
requirements can be satisfied by choosing a suitable time interval
between the photo missions. The photogrammetric technique described
here works especially well for determining the creep of permafrost
surfaces but is also suitable for observing glacier flow and slope
movements. Maximum accuracy of the method, as deduced by repeating
measurements using independent multitemporal stereo models and
by comparing the results with geodetic stake measurements, is
estimated to be about 30 micrometer in image scale.
(This work has been carried out at the Laboratory of Hydraulics,
Hydrology and Glaciology, ETH Zurich)
back to the Department of Geography homepage.
