Technical developments for the geomorphological reconstruction of palaeo ice sheets from remotely sensed data.

Ice sheet reconstructions are concerned with an understanding of the dynamics of
palaeo ice sheets through time. This involves ascertaining the configuration of ice
domes and flow patterns, as well as the evolution of glaciation and deglaciation. The
use of remotely sensed data (radar and visual/near infra-red imagery) has been a key
development in these studies, allowing the rapid mapping of individual landforms over
large areas. These data sources record electro-magnetic reflectance of an illuminated
landscape which leads to the introduction of random and systematic bias’. This
research explores the propagation of these bias’ and recommends that, in order to
obtain optimum imagery, sensor spatial resolution is <30m and solar elevation is <20º
at acquisition. More problematic is the selective bias introduced by the solar azimuth in
relation to the orientation of linear landforms. This cannot be removed and requires a
good knowledge of the study area or an additional primary data source (e.g. radar or
digital elevation model).
DEMs are rapidly supplementing, and in many cases replacing, satellite imagery in
landform mapping. However as they record surface elevation, rather than surface
reflectance, they should be able to provide a bias-free data source for landform
mapping. Methods by which the landscape can be visualised are explored with the
purpose of providing a bias free method, although no single visualisation was found to
be satisfactory. A suitable mapping strategy was developed through the application of
these methods to the mapping of glacial landforms from a DEM of the Lake District.
Once landform mapping has been completed it is common for mapped landform data
to be generalised in order to reduce their complexity and so aid interpretation. This is
usually a manual technique that involves reducing several thousand individual
lineaments to summary lines. The thesis concludes with the development of a set of
tools to help with this manual process, as well as provide quantitative verification. This
is then applied to the landforms mapped from the Lake District in the previous chapter.
A method for the automation of this procedure is also suggested.