The application of microwave remote sensing is getting attention due to its
penetrating capability through cloud and tree canopy (Frappart et al. 2005; Matgen
et al. 2011; Schumann and Moller 2015). The most distinctive character of water is
its appearance as a specular reflector to microwave radiation. Thus, most of the
incident energy goes away from the radar antenna after the interaction with the water
surface (Fig. 16.3). Based on this principle, Synthetic Aperture Radar (SAR) and
imaging spectrometer create the opportunity for flood mapping with microwave
remote sensing.
SAR data from various satellite missions such as RADARSAT, ENVISAT,
ALOS, TerraSAR, and Sentinal-1 have been used for flood mapping since the
1990s. Various methods and techniques have been developed for SAR-based flood
mapping including visual image interpretation (Macintosh and Profeti 1995), supervised classification (Townsend 2002), image histogram thresholding (Brivio et al.
2002), image texture algorithm (Schumann et al. 2009), rule-based analysis (Henry
et al. 2006), object-oriented classification (Heremans et al. 2003), and multi-temporal change detection algorithm (Bazi et al. 2005). A threshold value of radar
backscatter can be used to separate flooded and non-flooded areas based on the
image. Two images from before and after flood events are considered in change
detection-based flood mapping. The area is marked as flooded if the amount of radar
backscatter declines considerably from pre-flood to post-flood images. Henry et al.
(2006) used a rule-based approach on the different polarization of SAR data and
concluded that flood mapping is easier with polarized data (HH, VV) compared to
cross-polarization (HV-VH). Figure 16.4 illustrates the flood inundation extracted
from Sentinel 1 like polarization (SAR-C band, VV polarization) data for the 2017
Haor flood event in Bangladesh with histogram thresholding technique.
Fig. 16.3 Schematic diagram of the interaction between microwave and land cover in flooded and
non-flooded condition
328
R. M. Shrestha and M. S. Rahman
penetrating capability through cloud and tree canopy (Frappart et al. 2005; Matgen
et al. 2011; Schumann and Moller 2015). The most distinctive character of water is
its appearance as a specular reflector to microwave radiation. Thus, most of the
incident energy goes away from the radar antenna after the interaction with the water
surface (Fig. 16.3). Based on this principle, Synthetic Aperture Radar (SAR) and
imaging spectrometer create the opportunity for flood mapping with microwave
remote sensing.
SAR data from various satellite missions such as RADARSAT, ENVISAT,
ALOS, TerraSAR, and Sentinal-1 have been used for flood mapping since the
1990s. Various methods and techniques have been developed for SAR-based flood
mapping including visual image interpretation (Macintosh and Profeti 1995), supervised classification (Townsend 2002), image histogram thresholding (Brivio et al.
2002), image texture algorithm (Schumann et al. 2009), rule-based analysis (Henry
et al. 2006), object-oriented classification (Heremans et al. 2003), and multi-temporal change detection algorithm (Bazi et al. 2005). A threshold value of radar
backscatter can be used to separate flooded and non-flooded areas based on the
image. Two images from before and after flood events are considered in change
detection-based flood mapping. The area is marked as flooded if the amount of radar
backscatter declines considerably from pre-flood to post-flood images. Henry et al.
(2006) used a rule-based approach on the different polarization of SAR data and
concluded that flood mapping is easier with polarized data (HH, VV) compared to
cross-polarization (HV-VH). Figure 16.4 illustrates the flood inundation extracted
from Sentinel 1 like polarization (SAR-C band, VV polarization) data for the 2017
Haor flood event in Bangladesh with histogram thresholding technique.
Fig. 16.3 Schematic diagram of the interaction between microwave and land cover in flooded and
non-flooded condition
328
R. M. Shrestha and M. S. Rahman