less than 100 m in size, were partly lost in the speckle pattern; otherwise, the scale
and the resolution of the simulated imagery proved to be very valuable for large
area strategic ice forecasting and operational planning.
The generation and evaluation of realistic image examples from hypothetical
spaceborne SAR systems is of interest to those nations, which plan to implement
such systems. In Canada, a methodology (Singhroy and Saint-Jean 1999) has been
developed under contract by Intera Technologies Limited to simulate SAR imagery
(CSA 2007; D’Iorio et al. 1997). The main purpose of the SARSIM software package
is to simulate spaceborne SAR data of the ERS-1 and RADARSAT type, as they
have been available since 1990s. The package provides a set of filters that resample
fine resolution airborne SAR data, or those from synthetic test sites, and simulate
the parameters of the spaceborne SAR under consideration. This can be accomplished by generating an idealized yet realistic source target reflectively map and
by generating speckle, thermal system noise and other system perturbations.
Reflectivity map, speckle and thermal noise data are then combined to create the
simulation. Speckle is simulated by generating a two-dimensional array of noise
values. The user can specify the position, shape and weight for each sub-aperture
(‘look’). The usefulness of the resulting image can be determined from the means of
standard digital image analysis techniques. Moreover, new image analysis, interpretation and classification techniques can be refined and tested using simulated imagery.
14.7 Visual Interpretation of SAR Imagery
The most common approach to visual SAR image interpretation is that of a
modified air photo analysis procedure. The majority of airborne radar surveys
conducted during the 1970s and early 1980s relied heavily on visual interpretation
methods. Image analysts were familiar with both air photo and radar techniques.
Major natural resource surveys, such as RADAM BRAZIL (Marcelo et al. 2011) or
the Nigerian NIRAD Project, contributed to the development of manual analysis
techniques. Likewise, airborne radar imagery of ice infected coastal waters of
northern and eastern Canada have been interpreted manually by skilled ice
observers for many years.
Visual image analysis procedures examine various important image elements,
including tone, texture, size, shape and association. These image elements are
equally applicable for the interpretation of SAR imagery. But in modifying the
principles of visual air photo analysis for use in radar image interpretation, one
should keep in mind that radar is a range measuring device. The analyst has to
consider the radar parameters and the ground parameters when interpreting a radar
image, since variations in the average backscatter cross-section (σ
o ) result in
different image characteristics, first and foremost. One must be familiar with
interaction mechanism(s) of radar system parameters and ground parameters and
what effect changes in these parameters might have on σ
o . still, the analyst is left
with two basic questions; first, are the parameters that the analysis of ground
294
S. Pirasteh et al.
and the resolution of the simulated imagery proved to be very valuable for large
area strategic ice forecasting and operational planning.
The generation and evaluation of realistic image examples from hypothetical
spaceborne SAR systems is of interest to those nations, which plan to implement
such systems. In Canada, a methodology (Singhroy and Saint-Jean 1999) has been
developed under contract by Intera Technologies Limited to simulate SAR imagery
(CSA 2007; D’Iorio et al. 1997). The main purpose of the SARSIM software package
is to simulate spaceborne SAR data of the ERS-1 and RADARSAT type, as they
have been available since 1990s. The package provides a set of filters that resample
fine resolution airborne SAR data, or those from synthetic test sites, and simulate
the parameters of the spaceborne SAR under consideration. This can be accomplished by generating an idealized yet realistic source target reflectively map and
by generating speckle, thermal system noise and other system perturbations.
Reflectivity map, speckle and thermal noise data are then combined to create the
simulation. Speckle is simulated by generating a two-dimensional array of noise
values. The user can specify the position, shape and weight for each sub-aperture
(‘look’). The usefulness of the resulting image can be determined from the means of
standard digital image analysis techniques. Moreover, new image analysis, interpretation and classification techniques can be refined and tested using simulated imagery.
14.7 Visual Interpretation of SAR Imagery
The most common approach to visual SAR image interpretation is that of a
modified air photo analysis procedure. The majority of airborne radar surveys
conducted during the 1970s and early 1980s relied heavily on visual interpretation
methods. Image analysts were familiar with both air photo and radar techniques.
Major natural resource surveys, such as RADAM BRAZIL (Marcelo et al. 2011) or
the Nigerian NIRAD Project, contributed to the development of manual analysis
techniques. Likewise, airborne radar imagery of ice infected coastal waters of
northern and eastern Canada have been interpreted manually by skilled ice
observers for many years.
Visual image analysis procedures examine various important image elements,
including tone, texture, size, shape and association. These image elements are
equally applicable for the interpretation of SAR imagery. But in modifying the
principles of visual air photo analysis for use in radar image interpretation, one
should keep in mind that radar is a range measuring device. The analyst has to
consider the radar parameters and the ground parameters when interpreting a radar
image, since variations in the average backscatter cross-section (σ
o ) result in
different image characteristics, first and foremost. One must be familiar with
interaction mechanism(s) of radar system parameters and ground parameters and
what effect changes in these parameters might have on σ
o . still, the analyst is left
with two basic questions; first, are the parameters that the analysis of ground
294
S. Pirasteh et al.