(geomorphology) is one of the more successful applications of multispectral
remote sensing to reef environments during its history from the 1970s to the
present (Smith et al. 1975; Andréfouët et al. 2006).
Wavelength-dependency of light penetration offers the ability to estimate
bathymetry of reefs, as it is possible to estimate water depths using multispectral
data by tuning some parameters using ground truth data (e.g., light attenuation
coefficients, water depth, and bottom albedo) (Philpot 1989). Recent applications
of this method demonstrate that it is possible to discern water depths with rootmean-square-error (RMS) \30 % of true depths in \25 m of water (Stumpf et al.
2003). Although other methods to measure water depths are effective, and in some
cases significantly more accurate (i.e., Section III Acoustic and Section II LiDAR),
estimation of bathymetry using multispectral satellite data is advantageous due to
its availability in otherwise inaccessible locations by boats or aircrafts equipped
with acoustic or LiDAR sensors.
Extracting shorelines (i.e., the low and high tide limits) are important for
examining dynamics of coasts. Shorelines in coral reef environments can be
extracted successfully using NIR sensor bands, where lack of foam and suspended
sediments (which can affect the NIR wavelength region) and the presence of
remnant water on reef flats during low tide (which can affect the SWIR wavelength
region) contribute to improved shoreline detection. Yamano et al. (2006a) found a
strong linear relationship (r
2 = 0.81) between the error of shoreline position and
spatial resolution of the NIR bands. Extracting waterlines under various water
level conditions can also be utilized as an alternative means to measure the
topography of shallow intertidal areas, since the differential shorelines can be
regarded as contours (Yamano 2007).
Indirect remote sensing. The objective of indirect remote sensing is the environment surrounding the reefs (Table 3.2). For example, coastal land use affects
coral reefs through watershed-based pollution. As its name suggests, Landsat has
been designed for mapping land areas, and Landsat-derived global land cover
estimates have been prepared (Robinson et al. 2006), which can be used to provide
estimates of terrestrial inputs to reef ecosystems. Oceanic and atmospheric environments (e.g., photosynthetically active radiation, light attenuation coefficient,
cloud/dust cover, chlorophyll concentration, algal bloom, turbidity/suspended
sediment concentration, ocean circulation and coastal circulation) are generally
large-scale and variable through space and time. As a result, coarse spatial resolution satellite sensors (e.g., MODIS, MERIS, SeaWiFS) with short observation
intervals (*1 day) may be more feasible for observing these large-scale oceanic
and environmental features, which change on a daily basis. In contrast, most highand moderate-spatial resolution multispectral sensors have relatively long
(*10 days) observation intervals, which limit the ability to examine change at a
fine temporal scale.
58
H. Yamano
remote sensing to reef environments during its history from the 1970s to the
present (Smith et al. 1975; Andréfouët et al. 2006).
Wavelength-dependency of light penetration offers the ability to estimate
bathymetry of reefs, as it is possible to estimate water depths using multispectral
data by tuning some parameters using ground truth data (e.g., light attenuation
coefficients, water depth, and bottom albedo) (Philpot 1989). Recent applications
of this method demonstrate that it is possible to discern water depths with rootmean-square-error (RMS) \30 % of true depths in \25 m of water (Stumpf et al.
2003). Although other methods to measure water depths are effective, and in some
cases significantly more accurate (i.e., Section III Acoustic and Section II LiDAR),
estimation of bathymetry using multispectral satellite data is advantageous due to
its availability in otherwise inaccessible locations by boats or aircrafts equipped
with acoustic or LiDAR sensors.
Extracting shorelines (i.e., the low and high tide limits) are important for
examining dynamics of coasts. Shorelines in coral reef environments can be
extracted successfully using NIR sensor bands, where lack of foam and suspended
sediments (which can affect the NIR wavelength region) and the presence of
remnant water on reef flats during low tide (which can affect the SWIR wavelength
region) contribute to improved shoreline detection. Yamano et al. (2006a) found a
strong linear relationship (r
2 = 0.81) between the error of shoreline position and
spatial resolution of the NIR bands. Extracting waterlines under various water
level conditions can also be utilized as an alternative means to measure the
topography of shallow intertidal areas, since the differential shorelines can be
regarded as contours (Yamano 2007).
Indirect remote sensing. The objective of indirect remote sensing is the environment surrounding the reefs (Table 3.2). For example, coastal land use affects
coral reefs through watershed-based pollution. As its name suggests, Landsat has
been designed for mapping land areas, and Landsat-derived global land cover
estimates have been prepared (Robinson et al. 2006), which can be used to provide
estimates of terrestrial inputs to reef ecosystems. Oceanic and atmospheric environments (e.g., photosynthetically active radiation, light attenuation coefficient,
cloud/dust cover, chlorophyll concentration, algal bloom, turbidity/suspended
sediment concentration, ocean circulation and coastal circulation) are generally
large-scale and variable through space and time. As a result, coarse spatial resolution satellite sensors (e.g., MODIS, MERIS, SeaWiFS) with short observation
intervals (*1 day) may be more feasible for observing these large-scale oceanic
and environmental features, which change on a daily basis. In contrast, most highand moderate-spatial resolution multispectral sensors have relatively long
(*10 days) observation intervals, which limit the ability to examine change at a
fine temporal scale.
58
H. Yamano