Geology Reference
In-Depth Information
(Fig. 5.18) in order to detect reliable and
sometimes small (<28 mm) amplitude or slow
deformation (Peltzer
et al
., 2001). In a second
approach, the problem of decorrelation is
circumvented by using
permanent scatterers
,
which are robust features, either man-made or
natural, that can be detected from one image to
the next (Ferretti
et al
., 2001). The relative move-
ment of these features, whether they be outcrops,
buildings, or utility poles, defines a deformation
field that may lack the spatial detail of standard
InSAR analysis in arid areas, but that instead can
capture displacement in urbanized, vegetated,
and mountainous areas (Bürgmann
et al
., 2006)
that are not commonly amenable to typical
InSAR procedures.
10
Haiyuan
Fault
InSAR
Stack
5
0
~5 mm/yr
-5
σ
InSAR mean
1
GPS site & 1
σ
on mean
-10
-100
0
Distance from fault (km)
-50
50
100
Fig. 5.18
Stacked InSAR data across an active fault.
Average fault-parallel velocity profile across the Haiyuan
Fault in north-central China derived by combining more
than 20 InSAR images. The mean displacement was
scaled in order to reconcile with the GPS data from the
same area. This synthesis of stacked images suggests a
relatively weak fault and rather block-like behavior for
long distances (>50 km) on either side of it. Modified
after Cavalié
et al
. (2008).
Lidar imaging
Despite the attractive properties and scientific
insights derived from InSAR studies, in many
heavily forested regions such studies are
impracticable, because the density and seasonal
variation in the canopy simply does not provide
a repeatable image of the ground. As mentioned
in the introductory chapter, a new imaging
technique has ushered in an era of very high-
resolution topographic imaging that can define
the ground surface beneath forest canopies
(Fig. 1.5). Rather than relying on radar, lidar
(light detection and ranging) uses concentrated
pulses of light that can typically “see” the
ground through even small openings in a forest
canopy (Carter
et al
., 2007). Both airborne and
ground-based lidar are now being used to cre-
ate “bare-Earth” DEMs. Their spatial resolution
depends on the distance of the lidar instrument
from its target, but commonly DEMs are created
with 1-m spatial resolution from airborne
instruments and 1-cm to 1-mm resolution with
ground-based ones. As lidar has unmasked
previously unknown faults and has imaged
landslides, channels, and hillslopes beneath
forest canopies, its popularity has grown. In the
United States, two centers for lidar acquisition
sponsored by the National Science Foundation
(NSF) have been created: the National Center
for Airborne Laser Mapping (NCALM) and
tectonic deformation (Bawden
et al
., 2001).
Some recent studies suggest that seasonal
changes in water loading may modulate
tectonism in surprising ways (Box 5.2).
InSAR is most successful for regions with
rapid rates or large magnitudes of deformation
and areas with little vegetation and relatively
minor variations in surface moisture, such as
deserts. In areas with abundant vegetation,
widespread agriculture, or urbanization, sea-
sonal or spatially abrupt changes in the surface
can cause the image to decorrelate, such that
coherence is lost between adjacent pixels.
In such a case, a smoothly varying deformation
pattern, such as seen with the Landers rupture
(Fig. 5.17a), may be impossible to reconstruct.
Even in desert sites with little vegetation or
human interference, if rates of deformation are
quite slow, say a few mm/yr, reliable detection
of this deformation may be impracticable
with traditional InSAR methods. Several new
approaches have been developed to address
such problems. In one approach, rather than
using just two or three images, a temporal
succession of images (perhaps several dozen)
of the same region are stacked and smoothed
