10 Polar SAR Data for Operational Sea Ice Mapping
215
With the rapid and continuous advances in the speed of signal processors, future generations of satellites may feature onboard, real-time processing of SAR data with the
potential capacity for direct broadcast to users. This would allow the satellite to simulate aircraft ice reconnaissance broadcasts, providing image displays on ships and offshore platforms in real-time as the satellite passes overhead. The technological
advances required to permit this are substantial and they are not expected to be realized before the year 2010.
10.S
Manual Analysis Techniques
SAR imagery is the only high-resolution data source capable of penetrating the perpetual cloud-cover conditions and illumination limitations unique to polar regions.
Operational demonstrations using SAR imagery from the ERS-1/ -2 satellites at the US,
Canadian, and Baltic Ice Centers (Bertoia and Carrieres 1994; Ramsay et al. 1994; Herland and Berglund 1995) have proven the feasibility of using SAR images operationally for the creation of ice classification, concentration, and motion products. The techniques developed have led to improvements in a wide variety of ice information and
decision-making products.
Operational support requirements place the most emphasis on defining conditions
in the MIZ. The geophysical parameters that have been deemed to be most significant
to the operational sea ice community are ice concentration, ice type, age and thickness,
lead size and orientation, floe size, topography, ice drift, melt state, and convergence/divergence conditions. Interpretation of SAR imagery can provide information
describing most of these parameters. Operational experience, however, has proven that
SAR imagery alone is of much less value than when it is used in conjunction with other data sources and/or ground truth observations.
Noise associated with SAR imaging is of concern to the ice analyst. Every SAR has a
noise floor or noise level associated with the instrument's detection capability. Any
received signal which falls below this noise floor cannot be distinguished from the background signal. This can be significant for low return targets such as water and new ice.
The ERS-1 noise floor is sufficiently high to allow differentiation between new ice and
smooth open water, although early results from RADARSAT have shown this may not
be true in the far range of the ScanSAR wide images.
Speckle noise is an inevitable byproduct of image formation in a coherent radar system (see Livingstone 1994). Speckle causes a grainy appearance in SAR images that can
make image interpretation difficult. Spatial averaging is the simplest method for reducing image speckle and for this reason the operational ice centers generally use SAR
image products at less than the maximum resolution possible for the sensor. For example, the Alaska SAR Facility produces its ERS quick turn around products at 240-m resolution, acquired through simple 8X8 averaging of the 30-m full resolution product.
The ESA Fast Delivery Product produced by Norway's Troms0 Satellite Station has a
1Oo-m resolution, as does the product prepared by Canada's Gatineau Station.
Calibration is the process of removing radar sensor-dependent parameters from the
radar images so that the digital number of the image intensity accurately represents the
imaged surface features. In a properly calibrated system, the same surface feature will
consistently return the same pixel value. For the purposes of qualitative visual inter-
215
With the rapid and continuous advances in the speed of signal processors, future generations of satellites may feature onboard, real-time processing of SAR data with the
potential capacity for direct broadcast to users. This would allow the satellite to simulate aircraft ice reconnaissance broadcasts, providing image displays on ships and offshore platforms in real-time as the satellite passes overhead. The technological
advances required to permit this are substantial and they are not expected to be realized before the year 2010.
10.S
Manual Analysis Techniques
SAR imagery is the only high-resolution data source capable of penetrating the perpetual cloud-cover conditions and illumination limitations unique to polar regions.
Operational demonstrations using SAR imagery from the ERS-1/ -2 satellites at the US,
Canadian, and Baltic Ice Centers (Bertoia and Carrieres 1994; Ramsay et al. 1994; Herland and Berglund 1995) have proven the feasibility of using SAR images operationally for the creation of ice classification, concentration, and motion products. The techniques developed have led to improvements in a wide variety of ice information and
decision-making products.
Operational support requirements place the most emphasis on defining conditions
in the MIZ. The geophysical parameters that have been deemed to be most significant
to the operational sea ice community are ice concentration, ice type, age and thickness,
lead size and orientation, floe size, topography, ice drift, melt state, and convergence/divergence conditions. Interpretation of SAR imagery can provide information
describing most of these parameters. Operational experience, however, has proven that
SAR imagery alone is of much less value than when it is used in conjunction with other data sources and/or ground truth observations.
Noise associated with SAR imaging is of concern to the ice analyst. Every SAR has a
noise floor or noise level associated with the instrument's detection capability. Any
received signal which falls below this noise floor cannot be distinguished from the background signal. This can be significant for low return targets such as water and new ice.
The ERS-1 noise floor is sufficiently high to allow differentiation between new ice and
smooth open water, although early results from RADARSAT have shown this may not
be true in the far range of the ScanSAR wide images.
Speckle noise is an inevitable byproduct of image formation in a coherent radar system (see Livingstone 1994). Speckle causes a grainy appearance in SAR images that can
make image interpretation difficult. Spatial averaging is the simplest method for reducing image speckle and for this reason the operational ice centers generally use SAR
image products at less than the maximum resolution possible for the sensor. For example, the Alaska SAR Facility produces its ERS quick turn around products at 240-m resolution, acquired through simple 8X8 averaging of the 30-m full resolution product.
The ESA Fast Delivery Product produced by Norway's Troms0 Satellite Station has a
1Oo-m resolution, as does the product prepared by Canada's Gatineau Station.
Calibration is the process of removing radar sensor-dependent parameters from the
radar images so that the digital number of the image intensity accurately represents the
imaged surface features. In a properly calibrated system, the same surface feature will
consistently return the same pixel value. For the purposes of qualitative visual inter-