Geoscience Reference
In-Depth Information
the elevation of smaller water bodies (*hundreds of metres across) can be obtained and
used for flood model validation (e.g., Wilson et al.
2007
). The GLAS laser aboard the
ICESat satellite, although primarily designed to measure ice sheet topography, produced
data with a footprint of *70 m, which makes it more suitable for observing river water
levels than radar altimeters. However, GLAS only operated between 2003 and 2009 and
the laser instruments on board suffered from a number of technical issues such that only
limited data with track spacing similar to radar altimeters are available. Nevertheless, the
data have proved useful in particular areas of surface water science, such as geodetically
levelling river gauges in remote basins (Hall et al.
2012
) and determining water surface
slopes in large unmonitored rivers (e.g., O'Loughlin et al.
2013
).
As an alternative to profiling instruments, images of relative water height change over
time (qh/qt) can be obtained from interferometric analysis of pairs of coherent SAR scenes
taken from slightly different viewing geometries. Coregistration of the images to sub-pixel
accuracy and subtraction of the complex phase and amplitude for each image allows
surface displacement to be measured to centimetric accuracy. Such techniques were
originally developed for ground deformation and glaciological studies (see, for example,
Massonnet et al.
1993
; Goldstein et al.
1993
), but have subsequently been employed to map
surface waters in particular circumstances (see Alsdorf et al.
2000
,
2001a
,
b
). For open
water, specular reflection of the radar signal usually results in complete loss of temporal
coherence, but for inundated floodplains, where there is emergent vegetation, Alsdorf et al.
(
2000
) show that it is possible to obtain reliable repeat-pass interferometric measurements
because of the so-called double bounce effect whereby the radar path includes both water
and tree trunk surfaces. This allows relative water elevation change between images to be
mapped to *100-m resolution with centimetric accuracy and has been used to map
complex water height change patterns in the Amazon floodplain (Alsdorf et al.
2007b
) and
to undertake rigorous testing of the ability of two-dimensional floodplain models to sim-
ulate the spatial and temporal dynamics of inundation (Jung et al.
2012
).
Finally, from maps of inundation extent determined using the techniques outlined in
Sect.
2.2
, estimates of water elevation can be obtained by intersecting the shoreline vector
with a suitable DEM. Such techniques are reviewed in detail by Schumann et al. (
2009
)
who note their utility for constraining hydraulic models. The accuracy of water elevation
data derived in this way clearly depends on both the quality of the image processing and
the resolution and accuracy of the DEM, but Schumann et al. (
2010
) show that useful
information for flood wave analysis can be obtained even when using low-resolution (75 m
pixel) ASAR wide swath mode images and the SRTM DEM. Moreover, Mason et al.
(
2009
) show that water elevations obtained by intersecting SAR imagery with DEM data
are better at discriminating between competing model formulations than inundation extent
data.
2.4 Remote Measurements of Water Storage
Change in water storage on the land surface can be measured either indirectly by calcu-
lating the implied volume difference between two flood extent measurements when
intersected with a suitable DEM or directly using observations of the Earth's changing
gravity field. Data on the latter are available from the GRACE and GOCE satellite mis-
sions, although with limited spatial (*hundreds of kilometres) and temporal (*monthly)
resolution. In their raw state, such data may therefore not be terribly useful for surface
flood studies; however, Alsdorf et al. (
2010
) show that by carefully combining GRACE
data with information on precipitation, evaporation and inundation extent, it was possible
