total number of altered (damaged) crop pixels before and after flood events. These
damaged pixels are further translated into specific area units (acreages etc.)
depending on the image spatial resolution. Additionally, this acreage damage can
be further extrapolated into loss of monetary value based on specific crop types and
geographic location with a well-defined harvest index (Meroni et al. 2013).
Classification-based approach on flood crop damage assessment could be
archived by the utilization of airborne images such as ADAR 5500 multispectral
sensor system or by spaceborne images such as Landsat TM/ETM. The airborne
images provide extremely high spatial resolution (up to 0.25 m), resulting in a more
precise classification of the image, which could lead to highly accurate crop damage
assessment, especially within localized areas. That said due to its high spatial
resolution, the spatial coverage for the airborne images will be limited to a smaller
geographic extent. Furthermore, since these images are mostly on-demand, there
is little to no historical data available. Another drawback of using airborne images
for crop damage assessment is the monetary cost. Most of these images are captured
and/or sold by commercial companies, and the cost of these images would be
extremely high especially when multi-temporal scenes are required to monitor
different stages of crop growth.
Utilization of spaceborne images for the classification-based flood crop damage
assessment is, to some extent, able to address limitations using airborne images.
Compared to airborne images, spaceborne images are affordable (LandSat imagery
is free) and have an extensive historical dataset. Furthermore, satellite images also
have a higher spatial resolution (30–60 m for LandSat), which is fine enough
resolution for land-cover classifications. That said, the temporal resolution for the
satellite images is comparatively coarse (16 days for LandSat) which is not ideal for
crop monitoring as daily information is required to carry out the flood damage
assessments. Similarly, these images are also highly vulnerable to noise contaminations. In fact, satellite-derived images almost always contain some level of impurities
due to various atmospheric effects such as cloud, dust, and aerosol (Cihlar and
Howarth 1994) as well as sensor’s geometric effect such as low solar zenith angle
and off-nadir viewing angle (Jackson et al. 1990; Gatebe et al. 2001). Furthermore,
the classification approach is extremely subjective and time-consuming, and the
accuracy of the result will be highly dependent on the quality of the training sets.
Hence, these approaches are not efficient for global classification analysis (Jensen
et al. 1995; Song et al. 2001). Similarly, areas of considerable topography such as
aspects and slopes will also influence variations in surface spectral responses.
16.3.2 Band Ratioing (Vegetation Indices)
Shortcomings of classification-based flood crop damage assessment can be
addressed by band ratioing, especially utilizing spaceborne images. Band ratioing
is a process of calculating the ratio of the spectral responses between various spectral
bands within a single image. As the infrared band of the spectrum is much more
16 Flood Monitoring and Crop Damage Assessment
333
damaged pixels are further translated into specific area units (acreages etc.)
depending on the image spatial resolution. Additionally, this acreage damage can
be further extrapolated into loss of monetary value based on specific crop types and
geographic location with a well-defined harvest index (Meroni et al. 2013).
Classification-based approach on flood crop damage assessment could be
archived by the utilization of airborne images such as ADAR 5500 multispectral
sensor system or by spaceborne images such as Landsat TM/ETM. The airborne
images provide extremely high spatial resolution (up to 0.25 m), resulting in a more
precise classification of the image, which could lead to highly accurate crop damage
assessment, especially within localized areas. That said due to its high spatial
resolution, the spatial coverage for the airborne images will be limited to a smaller
geographic extent. Furthermore, since these images are mostly on-demand, there
is little to no historical data available. Another drawback of using airborne images
for crop damage assessment is the monetary cost. Most of these images are captured
and/or sold by commercial companies, and the cost of these images would be
extremely high especially when multi-temporal scenes are required to monitor
different stages of crop growth.
Utilization of spaceborne images for the classification-based flood crop damage
assessment is, to some extent, able to address limitations using airborne images.
Compared to airborne images, spaceborne images are affordable (LandSat imagery
is free) and have an extensive historical dataset. Furthermore, satellite images also
have a higher spatial resolution (30–60 m for LandSat), which is fine enough
resolution for land-cover classifications. That said, the temporal resolution for the
satellite images is comparatively coarse (16 days for LandSat) which is not ideal for
crop monitoring as daily information is required to carry out the flood damage
assessments. Similarly, these images are also highly vulnerable to noise contaminations. In fact, satellite-derived images almost always contain some level of impurities
due to various atmospheric effects such as cloud, dust, and aerosol (Cihlar and
Howarth 1994) as well as sensor’s geometric effect such as low solar zenith angle
and off-nadir viewing angle (Jackson et al. 1990; Gatebe et al. 2001). Furthermore,
the classification approach is extremely subjective and time-consuming, and the
accuracy of the result will be highly dependent on the quality of the training sets.
Hence, these approaches are not efficient for global classification analysis (Jensen
et al. 1995; Song et al. 2001). Similarly, areas of considerable topography such as
aspects and slopes will also influence variations in surface spectral responses.
16.3.2 Band Ratioing (Vegetation Indices)
Shortcomings of classification-based flood crop damage assessment can be
addressed by band ratioing, especially utilizing spaceborne images. Band ratioing
is a process of calculating the ratio of the spectral responses between various spectral
bands within a single image. As the infrared band of the spectrum is much more
16 Flood Monitoring and Crop Damage Assessment
333