230
G. Kite and A. Pietroniro
tures. Analysis of SIR-C/X-SAR radar images obtained during two space shuttle
flights in 1994 was used to reveal the bedrock structures of different ages that have
controlled the course of the River Nile (Stem and Abdelsalam, 1996). Pa1aeochannels of the Nile have been identified and show the relatively recent deflection of
the river to form the great bend in northern Sudan.
10.7 Flood Extent
Flooding in North America causes an average of $1 billion damages and several
dozen deaths each year (Paterson et aI., 1996). Accurate and timely information on
flood extent can help emergency personnel make better decisions about where to
deploy resources and how to plan evacuations, and can allow more accurate damage
assessments.
Many of the techniques described in the previous section on measuring lake areas
may also be used to measure flood extent. Satellite sensors in the visible and infrared
bands such as NOAA-A VHRR and the Landsat TM and MSS have long been used
to provide estimates of flood extent and flood hazard areas (e.g. Rango and Anderson,
1974). However, such sensors are dependent on cloud-free conditions which are rare
during major floods and cannot be reliably used for under-canopy flooding. For
example, Mertes (1994) used Landsat TM data to map flood events on the Amazon
River and found that the water surface on the flood plain was frequently masked by
vegetation.
Over the last 15 years the use of airborne and satellite synthetic aperture radar
(SAR) with its cloud penetrating day and night capability has developed considerable
potential for measuring flood extent (Pultz and Crevier, 1997). Ormsby and Blanchard (1985) carried out some of the earliest experimental work on SAR response to
flooded vegetation using X-band, C-band and L-band imagery. They concluded that
response depended on wavelength, plant volume and the geometry of the inundated
vegetation (Fig. lO.8). The study notes that at X-band wavelengths, the incoming
energy is scattered within the forest canopy, yet provides enhanced return within
saturated sedges and grasslands. At L-band frequencies, the response from short
vegetation was minimal if any, as noted in Fig. 10.9.
Wang et al. (1995) have also studied radar backscatter from flooded forests in Brazil. They concluded that the ratio of C band (such as on the ERS-1 and RADARSAT
satellites) backscatter from flooded forest to non-flooded forest is about 1.8 at an
incidence angle of 20 0 but decreases to about 1.0 at 60 0 • Brakenridge et al. (1994)
used ERS-1 data to monitor the Mississippi flood of 1993. SAR images are often
combined with optical or infrared images and with data from other sources such as
maps to produce images containing both flood extent and geographic locators such
as roads, and railways. Puyou-Lascassies et al. (1997) describe the use of ERS-I
images before, during and after the 1994 flood event in the Camargue (south-eastern
France) as part of a space-based European risk management system.
Based on the use of C-band airborne experiments, Pultz et al. (1991) suggested that
larger incidence angles and shorter revisit times than available on ERS-1 would give
better results for flood events. Pultz and Crevier (1997) describe the use of
G. Kite and A. Pietroniro
tures. Analysis of SIR-C/X-SAR radar images obtained during two space shuttle
flights in 1994 was used to reveal the bedrock structures of different ages that have
controlled the course of the River Nile (Stem and Abdelsalam, 1996). Pa1aeochannels of the Nile have been identified and show the relatively recent deflection of
the river to form the great bend in northern Sudan.
10.7 Flood Extent
Flooding in North America causes an average of $1 billion damages and several
dozen deaths each year (Paterson et aI., 1996). Accurate and timely information on
flood extent can help emergency personnel make better decisions about where to
deploy resources and how to plan evacuations, and can allow more accurate damage
assessments.
Many of the techniques described in the previous section on measuring lake areas
may also be used to measure flood extent. Satellite sensors in the visible and infrared
bands such as NOAA-A VHRR and the Landsat TM and MSS have long been used
to provide estimates of flood extent and flood hazard areas (e.g. Rango and Anderson,
1974). However, such sensors are dependent on cloud-free conditions which are rare
during major floods and cannot be reliably used for under-canopy flooding. For
example, Mertes (1994) used Landsat TM data to map flood events on the Amazon
River and found that the water surface on the flood plain was frequently masked by
vegetation.
Over the last 15 years the use of airborne and satellite synthetic aperture radar
(SAR) with its cloud penetrating day and night capability has developed considerable
potential for measuring flood extent (Pultz and Crevier, 1997). Ormsby and Blanchard (1985) carried out some of the earliest experimental work on SAR response to
flooded vegetation using X-band, C-band and L-band imagery. They concluded that
response depended on wavelength, plant volume and the geometry of the inundated
vegetation (Fig. lO.8). The study notes that at X-band wavelengths, the incoming
energy is scattered within the forest canopy, yet provides enhanced return within
saturated sedges and grasslands. At L-band frequencies, the response from short
vegetation was minimal if any, as noted in Fig. 10.9.
Wang et al. (1995) have also studied radar backscatter from flooded forests in Brazil. They concluded that the ratio of C band (such as on the ERS-1 and RADARSAT
satellites) backscatter from flooded forest to non-flooded forest is about 1.8 at an
incidence angle of 20 0 but decreases to about 1.0 at 60 0 • Brakenridge et al. (1994)
used ERS-1 data to monitor the Mississippi flood of 1993. SAR images are often
combined with optical or infrared images and with data from other sources such as
maps to produce images containing both flood extent and geographic locators such
as roads, and railways. Puyou-Lascassies et al. (1997) describe the use of ERS-I
images before, during and after the 1994 flood event in the Camargue (south-eastern
France) as part of a space-based European risk management system.
Based on the use of C-band airborne experiments, Pultz et al. (1991) suggested that
larger incidence angles and shorter revisit times than available on ERS-1 would give
better results for flood events. Pultz and Crevier (1997) describe the use of