measure, observe, and map quantitative and qualitative ocean surface information.
A few of the more mature techniques for analyzing and classifying SAR imagery
applicable to coral reef research and management are presented here.
High resolution winds SAR imagery is directly applicable to the measurement
of wind because the backscatter increases in proportion to wind speed. SARderived winds have some unique properties that make them ideally suited for use
in coral reef management: (1) they have the advantage of high resolution (on the
order of a few hundred meters to 1 km); (2) accurate measurements can be made
right up to the coast, and indeed even in lagoons, straits, and bays (unlike scatterometers that are affected by proximity to land); and (3) accuracy is comparable
to that of scatterometers in the open ocean with better accuracy near the coast
(Yang et al. 2011).
SAR winds are most commonly estimated using a Geophysical Model Function
(GMF) derived originally for use with scatterometers (Monaldo et al. 2004a).
Examples include CMOD-4 (Stofflen and Anderson 1997), CMOD-5 (Hersbach
et al. 2007), and CMOD-IFR2 (Quilfen et al. 1998). Among other parameter inputs
(Christiansen et al. 2008), these GMFs require estimation of the angle between the
wind direction and the radar look direction. In order to determine this angle, it is
necessary to know the wind direction, a parameter that is obtainable with scatterometers, which have more than one look angle, or from meteorological models
or buoys, but which cannot be measured directly with SAR instruments. Alternatively, in the absence of other data sources, wind direction can be obtained (with
180° ambiguity) from wind-aligned features (e.g., roll vortices, island wind
shadows) in the SAR image itself.
Since the existing C-band (4–8 GHz) scatterometer GMFs have all been
derived for vertical transmit, vertical receive (VV) polarization, it is necessary to
use a polarization ratio when applying the GMFs to horizontal transmit, horizontal
receive (HH) polarization SAR imagery. Polarization ratios currently employed
are a function of angle of incidence only (Thompson et al. 1998) or both angle of
incidence and wind direction (Mouche et al. 2007). Wind accuracy using these
GMFs and polarization ratios with C-band SAR data is in the range of ±1.5 to
±2.5 m/s for wind speeds less than 15 m/s when compared against buoy winds or
scatterometer winds (Monaldo et al. 2001, 2004b; Xu et al. 2010). Pichel et al.
(2008) similarly compared SAR CMOD5 winds derived from RADARSAT-1
imagery with buoy winds in the Bering Sea during the period November 2006 to
April 2007. For winds between 0 and 15 m/s (136 matches) the mean difference
was –0.02 m/s and the standard deviation was 2.23 m/s. For winds between 15
and 25 m/s (22 matches) the mean difference was 2.33 m/s and the standard
deviation was 2.37 m/s. Above 25 m/s, the currently available single-polarization
GMFs were determined not to have sufficient sensitivity. Cross-polarization
algorithms currently under development show promise of achieving accurate
winds for these higher wind speeds (Vachon and Wolfe 2010).
Oils spills Marine oil spills can damage coral reefs by killing coral outright or
making them more susceptible to bleaching (Johannes et al. 1972; Haapkylä et al.
2007), as well as harming reef fish and marine mammals, birds, and turtles. Wave
352
M. L. Heron et al.
A few of the more mature techniques for analyzing and classifying SAR imagery
applicable to coral reef research and management are presented here.
High resolution winds SAR imagery is directly applicable to the measurement
of wind because the backscatter increases in proportion to wind speed. SARderived winds have some unique properties that make them ideally suited for use
in coral reef management: (1) they have the advantage of high resolution (on the
order of a few hundred meters to 1 km); (2) accurate measurements can be made
right up to the coast, and indeed even in lagoons, straits, and bays (unlike scatterometers that are affected by proximity to land); and (3) accuracy is comparable
to that of scatterometers in the open ocean with better accuracy near the coast
(Yang et al. 2011).
SAR winds are most commonly estimated using a Geophysical Model Function
(GMF) derived originally for use with scatterometers (Monaldo et al. 2004a).
Examples include CMOD-4 (Stofflen and Anderson 1997), CMOD-5 (Hersbach
et al. 2007), and CMOD-IFR2 (Quilfen et al. 1998). Among other parameter inputs
(Christiansen et al. 2008), these GMFs require estimation of the angle between the
wind direction and the radar look direction. In order to determine this angle, it is
necessary to know the wind direction, a parameter that is obtainable with scatterometers, which have more than one look angle, or from meteorological models
or buoys, but which cannot be measured directly with SAR instruments. Alternatively, in the absence of other data sources, wind direction can be obtained (with
180° ambiguity) from wind-aligned features (e.g., roll vortices, island wind
shadows) in the SAR image itself.
Since the existing C-band (4–8 GHz) scatterometer GMFs have all been
derived for vertical transmit, vertical receive (VV) polarization, it is necessary to
use a polarization ratio when applying the GMFs to horizontal transmit, horizontal
receive (HH) polarization SAR imagery. Polarization ratios currently employed
are a function of angle of incidence only (Thompson et al. 1998) or both angle of
incidence and wind direction (Mouche et al. 2007). Wind accuracy using these
GMFs and polarization ratios with C-band SAR data is in the range of ±1.5 to
±2.5 m/s for wind speeds less than 15 m/s when compared against buoy winds or
scatterometer winds (Monaldo et al. 2001, 2004b; Xu et al. 2010). Pichel et al.
(2008) similarly compared SAR CMOD5 winds derived from RADARSAT-1
imagery with buoy winds in the Bering Sea during the period November 2006 to
April 2007. For winds between 0 and 15 m/s (136 matches) the mean difference
was –0.02 m/s and the standard deviation was 2.23 m/s. For winds between 15
and 25 m/s (22 matches) the mean difference was 2.33 m/s and the standard
deviation was 2.37 m/s. Above 25 m/s, the currently available single-polarization
GMFs were determined not to have sufficient sensitivity. Cross-polarization
algorithms currently under development show promise of achieving accurate
winds for these higher wind speeds (Vachon and Wolfe 2010).
Oils spills Marine oil spills can damage coral reefs by killing coral outright or
making them more susceptible to bleaching (Johannes et al. 1972; Haapkylä et al.
2007), as well as harming reef fish and marine mammals, birds, and turtles. Wave
352
M. L. Heron et al.